From c405adc4fc1e727de23c67cb7acee4311406db90 Mon Sep 17 00:00:00 2001 From: yuyr Date: Tue, 23 Dec 2025 14:22:15 +0800 Subject: [PATCH] mvp v1.1 save --- .gitignore | 3 +- specs/mvp/milestones.md | 34 + specs/mvp/mvp_roadmap.md | 348 +++ specs/mvp/v1.1/mvp_plan.md | 169 ++ specs/mvp/v1.1/sdk_submit_refactor.md | 148 ++ specs/mvp/v1.1/v1.1_action.md | 333 +++ src/mvp/v1.1/README.md | 61 + src/mvp/v1.1/docker-compose.yaml | 89 + src/mvp/v1.1/job_spec.schema.json | 33 + src/mvp/v1.1/py/configs/dev.yaml | 20 + src/mvp/v1.1/py/jobspecs/grpo.yaml | 20 + src/mvp/v1.1/py/jobspecs/ppo.yaml | 22 + src/mvp/v1.1/py/jobspecs/sft.yaml | 22 + src/mvp/v1.1/py/mvp_v11/__init__.py | 1 + src/mvp/v1.1/py/mvp_v11/builders.py | 96 + src/mvp/v1.1/py/mvp_v11/driver_entrypoint.py | 63 + src/mvp/v1.1/py/mvp_v11/models.py | 121 ++ src/mvp/v1.1/py/mvp_v11/ray_job_tool.py | 171 ++ src/mvp/v1.1/py/mvp_v11/yaml_io.py | 21 + src/mvp/v1.1/py/requirements.txt | 2 + src/mvp/v1.1/py/run.py | 69 + src/mvp/v1.1/py/sitecustomize.py | 57 + src/mvp/v1.1/scripts/00_prereq_check.sh | 42 + src/mvp/v1.1/scripts/01_up.sh | 16 + src/mvp/v1.1/scripts/02_down.sh | 12 + src/mvp/v1.1/scripts/03_cleanup_v1_legacy.sh | 16 + src/mvp/v1.1/scripts/05_ensure_verl_repo.sh | 23 + src/mvp/v1.1/scripts/12_install_py_deps.sh | 10 + src/mvp/v1.1/scripts/20_start_head.sh | 18 + src/mvp/v1.1/scripts/21_start_workers.sh | 26 + .../v1.1/scripts/30_prepare_data_and_model.sh | 86 + src/mvp/v1.1/scripts/31_snapshot_verl_code.sh | 42 + src/mvp/v1.1/scripts/32_clone_verl_tags.sh | 39 + src/mvp/v1.1/scripts/40_submit_ppo_epoch1.sh | 72 + src/mvp/v1.1/scripts/41_submit_grpo_epoch1.sh | 73 + src/mvp/v1.1/scripts/42_submit_sft_minimal.sh | 62 + src/mvp/v1.1/scripts/43_submit_jobspec.sh | 17 + src/mvp/v1.1/scripts/44_submit_sdk.sh | 19 + .../scripts/46_submit_ppo_two_verl_tags.sh | 86 + src/mvp/v1.1/scripts/50_status.sh | 13 + src/mvp/v1.1/scripts/lib.sh | 52 + src/mvp/v1.1/scripts/run_all.sh | 15 + src/mvp/v1.1/submit_job.py | 282 +++ src/mvp/v1.1/templates/grpo.json | 31 + src/mvp/v1.1/templates/ppo.json | 31 + src/mvp/v1.1/templates/sft.json | 32 + src/mvp/v1/arch.excalidraw | 1877 +++++++++++++++++ 47 files changed, 4894 insertions(+), 1 deletion(-) create mode 100644 specs/mvp/milestones.md create mode 100644 specs/mvp/mvp_roadmap.md create mode 100644 specs/mvp/v1.1/mvp_plan.md create mode 100644 specs/mvp/v1.1/sdk_submit_refactor.md create mode 100644 specs/mvp/v1.1/v1.1_action.md create mode 100644 src/mvp/v1.1/README.md create mode 100644 src/mvp/v1.1/docker-compose.yaml create mode 100644 src/mvp/v1.1/job_spec.schema.json create mode 100644 src/mvp/v1.1/py/configs/dev.yaml create mode 100644 src/mvp/v1.1/py/jobspecs/grpo.yaml create mode 100644 src/mvp/v1.1/py/jobspecs/ppo.yaml create mode 100644 src/mvp/v1.1/py/jobspecs/sft.yaml create mode 100644 src/mvp/v1.1/py/mvp_v11/__init__.py create mode 100644 src/mvp/v1.1/py/mvp_v11/builders.py create mode 100644 src/mvp/v1.1/py/mvp_v11/driver_entrypoint.py create mode 100644 src/mvp/v1.1/py/mvp_v11/models.py create mode 100644 src/mvp/v1.1/py/mvp_v11/ray_job_tool.py create mode 100644 src/mvp/v1.1/py/mvp_v11/yaml_io.py create mode 100644 src/mvp/v1.1/py/requirements.txt create mode 100644 src/mvp/v1.1/py/run.py create mode 100644 src/mvp/v1.1/py/sitecustomize.py create mode 100644 src/mvp/v1.1/scripts/00_prereq_check.sh create mode 100644 src/mvp/v1.1/scripts/01_up.sh create mode 100644 src/mvp/v1.1/scripts/02_down.sh create mode 100644 src/mvp/v1.1/scripts/03_cleanup_v1_legacy.sh create mode 100644 src/mvp/v1.1/scripts/05_ensure_verl_repo.sh create mode 100644 src/mvp/v1.1/scripts/12_install_py_deps.sh create mode 100644 src/mvp/v1.1/scripts/20_start_head.sh create mode 100644 src/mvp/v1.1/scripts/21_start_workers.sh create mode 100644 src/mvp/v1.1/scripts/30_prepare_data_and_model.sh create mode 100644 src/mvp/v1.1/scripts/31_snapshot_verl_code.sh create mode 100644 src/mvp/v1.1/scripts/32_clone_verl_tags.sh create mode 100644 src/mvp/v1.1/scripts/40_submit_ppo_epoch1.sh create mode 100644 src/mvp/v1.1/scripts/41_submit_grpo_epoch1.sh create mode 100644 src/mvp/v1.1/scripts/42_submit_sft_minimal.sh create mode 100644 src/mvp/v1.1/scripts/43_submit_jobspec.sh create mode 100644 src/mvp/v1.1/scripts/44_submit_sdk.sh create mode 100644 src/mvp/v1.1/scripts/46_submit_ppo_two_verl_tags.sh create mode 100644 src/mvp/v1.1/scripts/50_status.sh create mode 100644 src/mvp/v1.1/scripts/lib.sh create mode 100644 src/mvp/v1.1/scripts/run_all.sh create mode 100644 src/mvp/v1.1/submit_job.py create mode 100644 src/mvp/v1.1/templates/grpo.json create mode 100644 src/mvp/v1.1/templates/ppo.json create mode 100644 src/mvp/v1.1/templates/sft.json create mode 100644 src/mvp/v1/arch.excalidraw diff --git a/.gitignore b/.gitignore index c72c0e1..ce0d616 100644 --- a/.gitignore +++ b/.gitignore @@ -1,3 +1,4 @@ verl/ skypilot-ssh-test/ -ray_in_docker/ \ No newline at end of file +ray_in_docker/ +__pycache__/ diff --git a/specs/mvp/milestones.md b/specs/mvp/milestones.md new file mode 100644 index 0000000..c983f1d --- /dev/null +++ b/specs/mvp/milestones.md @@ -0,0 +1,34 @@ + +# milestones + +通过以下几个里程碑来梳理和分析确认可行性，最终目标是产出一套基于Native Ray集群（无k8s底座）的verl 训练平台，支持多用户，运行各类verl任务，提高整体集群的资源利用效率，并且能够通过监测系统进行观察和资源统计，监控报警。未来形成运维SOP后，接入运维智能体，执行自动化运维。 +- Workload + - ppo on ray + - grpo on ray + - sft on ray 可行性 + - model serving on ray + - customize code 自定义代码，任意verl example 提交代码 + - 自定义reward function + - 同时多verl版本支持，同时跑不同的ray任务，但是使用不同版本的verl，甚至是用户魔改版本 +- Ray Job管理 + - 通过python api提交，而不是通过ray cli提交 + - 任务排队机制。无优先级，多个pending job谁先满足资源就谁先执行。 + - 【确认支持】gang scheduling (all or nothing), 指定好trainer.nnodes和trainer.n_gpus_per_node参数，不满足就pending。 + - 无配额管理、公平调度等特性。 + - Ray本身不支持任务超时参数，需要单独job监控，发现超时才停止。 + - Pipeline管理【高级, 暂不实现】 + - 提供对Ray Job进一步封装，串联多个Ray Job，自动完成训练，模型合并等job串联 +- 可观测性 Observability + - 测试本地部署 weight and bias server 可行性，如何集成现有job流程 + - 测试部署 prometheus & grafana，对ray节点进行监测 + - job监控，哪些job使用了多少资源，跑了多长时间，资源利用率是否充分，是否空占着GPU +- 数据、模型存储管理 + - shared dataset管理：所有用户共享的hf数据集 + - hf 模型管理：所有用户共享的hf 基座模型库 + - user dataset 管理: 用户独自的数据集管理 + - user 模型管理：用户独自的模型管理，保存训练好的模型 + - job 作业数据管理，作业产出物，临时目录数据 + - user management：用户可以通过统一界面来管理自己是user dataset/model space和自己运行的job的临时目录，从而灵活组织任务流水线，提供灵活的文件查看方式 +- 网络 + - 确认是否支持IB(H100环境)，以及RoCEv2（H20环境），需要怎样配置 + diff --git a/specs/mvp/mvp_roadmap.md b/specs/mvp/mvp_roadmap.md new file mode 100644 index 0000000..716d002 --- /dev/null +++ b/specs/mvp/mvp_roadmap.md @@ -0,0 +1,348 @@ +# MVP Roadmap（V1 → V2 → … → 训练平台） + +本文档在 `specs/mvp/milestones.md` 的草稿基础上做**扩展与细化**：把目标拆成可迭代的版本（MVP v1/v2/…），保证每个版本都能**独立运行、可验证验收**，并且在上一版本基础上演进。 + +> 总目标（North Star）：产出一套**基于 Native Ray 集群（无 K8s 底座）**的训练平台，面向多用户，支持 `verl` 各类训练/评测/Serving 工作负载，提升集群利用率，并通过可观测系统实现资源统计、监控告警，最终形成运维 SOP 并可接入运维智能体做自动化运维。 + +--- + +## 0. 关键原则（贯穿所有版本） + +1) **版本可独立运行**：每个版本都能从“空环境”按文档跑起来（不依赖未来能力）。 +2) **验收可客观验证**：每个里程碑必须有明确的 DoD（Definition of Done）与可复现步骤。 +3) **强制产物落盘**：模型/数据/日志/ckpt 必须可追踪、可复用、可审计（基于共享存储/NFS）。 +4) **Head 不参与计算**：Head 只承担控制面（GCS/Dashboard/Job server），避免训练抢占控制面资源。 +5) **按 submission id 组织作业**：作业输出目录与 Ray submission id 绑定，方便检索、回收、归档。 +6) **“先把 RL 跑稳”，再扩 workload**：先 PPO（已验证），再 GRPO/SFT/Serving。 + +--- + +## 0.1 里程碑总览（建议交付顺序） + +| 版本 | 定位 | 关键交付 | 核心验收点 | +|---|---|---|---| +| v1 | 可复现实验闭环 | Ray 集群 + PPO 跑通 + 持久化 | driver 不在 head；产物落盘 | +| v1.1 | 实验工程化 | JobSpec 模板 + 新增 1 个 workload | 可回归、可定位、可扩展 | +| v2.0 | 服务化入口 | API + Ray Jobs SDK | API 提交/查询/停止可用 | +| v2.1 | 节点纳管 | SSH 注入 + 资源池/标签 | 节点上线/下线、gang 约束 | +| v3.0 | 平台雏形 | 队列 + 超时 + 最小多用户 | pending→running 自动调度 | +| v3.1 | 可扩展平台 | 自定义代码/reward + 多版本 | 多版本并存、插件可用 | +| v4.0 | 可运营平台 | Prom/Grafana + W&B | 资源核算/告警/归档 | +| v4.1 | 可交接平台 | SOP + 自动化运维接口 | 非开发可按 SOP 运维 | +| v5.0 | 长期形态 | Serving + Pipeline | 训练→发布推理闭环 | + +## 1. 当前基线：MVP v1（已完成/已验证） + +### 1.1 目标 + +在单机（或同一宿主机）用 3 个容器跑通： + +- Ray head（无 GPU，CPU=0/GPU=0） +- 2 个 Ray worker（每个 4 GPU） +- 通过 **head 上的 `ray job submit`** 提交 `verl` PPO（`total_epochs=1`） +- 通过 **entrypoint 自定义资源**强制 driver 在 worker 上 +- 数据/模型/日志/ckpt 全部持久化 + +### 1.2 交付物（repo 中已存在） + +- 脚本与 compose：`src/mvp/v1/` +- 行动与验收文档：`specs/mvp/v1/v1_action.md` +- 共享目录约定：`shared/datasets`、`shared/hf`、`shared/jobs` 等（与 NFS 对齐） + +### 1.3 验收口径（摘要） + +- `ray job list` 的 `driver_info.node_ip_address` ∈ worker IP，且 ≠ head IP +- 训练输出落在 `/mnt/shared/jobs//...` +- checkpoint 按 `save_freq` 产生（避免爆磁盘） + +--- + +## 2. MVP v1.1（Hardening + 多 workload 可行性验证） + +> 目标：把 v1 从“实验脚本”升级成“可长期回归的最小系统”，并验证更多 workload 的可行性边界。 + +### 2.1 主要能力 + +- Workload 扩展（可选顺序）： + - PPO（回归金标） + - GRPO on Ray（可运行验证） + - SFT on Ray（可运行验证：`llamafactory` 或 `verl` 相关 SFT 路径） +- 作业模板化（最小实现）： + - 统一 JobSpec（YAML/JSON）描述：workload 类型、资源（nnodes/n_gpus_per_node）、数据、模型、输出目录、超时 + - 仍然用 `ray job submit`，但把 entrypoint 组装逻辑标准化 +- checkpoint 策略与磁盘保护： + - 默认 `save_freq` ≥ 10（或按训练总 steps 的比例） + - 明确保留策略（至少提供“保留最后 N 个 ckpt”的配置建议/脚本） +- “失败可定位”： + - 统一收敛日志入口（Ray job logs + hydra 日志目录 + 关键参数快照） + - 失败时能定位：是资源不足 / NCCL / 数据 / 模型 / 配置错误 + +### 2.2 验收（DoD） + +- 同一套脚本在同一台机器能连续跑 3 次 PPO 回归，产物目录不互相覆盖 +- 至少新增 1 个 workload（GRPO 或 SFT）可以跑通 “启动→训练→落盘” 闭环 +- 作业目录内包含： + - `config/submit_cmd.txt`（或 job spec 快照） + - `logs/`（可追踪） + - `checkpoints/`（按策略生成） + +--- + +## 3. MVP v2.0（Control Plane 服务化：API + Ray Jobs SDK） + +> 目标：从“人跑脚本”升级为“服务提交任务”。依然是 Native Ray 集群，但引入一个最小控制平面服务。 + +### 3.1 系统形态 + +- Control Plane（建议部署在 head/CPU 机器）： + - FastAPI 服务（REST） + - Job 管理：用 Ray Jobs **Python SDK** 提交/查询/停止（不再依赖 CLI 文本解析） + - 节点视图：读取 Ray state（nodes, actors, placement groups） +- Data Plane： + - 仍然是预先启动的 worker 节点加入集群（先不做 SSH 动态纳管也可） + +### 3.2 API（MVP 级别） + +- `POST /v1/jobs`：提交 JobSpec（ppo/grpo/sft） +- `GET /v1/jobs`：列表（含状态、资源、开始/结束时间） +- `GET /v1/jobs/{id}`：详情（含输出目录、driver node） +- `POST /v1/jobs/{id}:stop`：停止作业 + +### 3.3 验收（DoD） + +- API 提交 PPO，返回 submission id；输出目录为 `/mnt/shared/jobs//...` +- API 查询 job 状态与 driver node（必须是 worker） +- 停止 job 后，资源释放、状态可见 + +--- + +## 4. MVP v2.1（SSH 纳管 + 资源池 + Gang 约束） + +> 目标：对齐你草稿里“SSH 纳管”的约束与需求：控制面能纳管 GPU 节点，形成可运营的资源池。 + +### 4.1 节点纳管（SSH Provisioner） + +- 控制面保存 NodeSpec（ip/user/port/labels/gpu_count） +- 通过 SSH 执行： + - `ray start --address=:6379 --resources=...` + - `ray stop`（drain/下线） +- 维护节点状态机：`pending → online → draining → offline` + +### 4.2 资源池与 gang（All-or-nothing） + +- 资源池最小模型： + - pool 标签（如 `pool_a`、`h20`、`ib_domain_1`） + - 提交 job 时指定 pool 约束 +- Gang 约束（MVP 实现方式）： + - job spec 明确 `trainer.nnodes` + `trainer.n_gpus_per_node` + - 提交前检查 Ray 可用资源是否满足，不满足则进入 pending 队列（见 v3.0） + +### 4.3 验收（DoD） + +- 通过 API 注册 2 个 worker（SSH 注入 ray start）后，`ray status` 可见节点上线 +- 通过 API 下线节点，节点被标记不可调度且不再分配新 job +- gang 不满足时 job 不提交（或提交后一直 pending），满足后可运行 + +--- + +## 5. MVP v3.0（调度与多用户：队列 + 超时 + 最小权限） + +> 目标：平台开始“像个平台”：多用户、队列、超时、审计。仍然不做复杂配额/公平调度。 + +### 5.1 作业队列（简单但可用） + +- FIFO 队列：无优先级 +- “资源满足就调度”：谁先满足谁先跑（可接受非严格 FIFO） +- job 超时：Ray 原生不支持统一 timeout（草稿已指出），因此控制面需： + - 记录 start_time + - 定期扫描超时 job → `stop` + +### 5.2 多用户最小闭环 + +- 认证（MVP）：token 或 basic auth（先不做复杂 RBAC） +- 归属与隔离（文件层）： + - `/mnt/shared/users//datasets/` + - `/mnt/shared/users//models/` + - `/mnt/shared/jobs//` 记录 user/metadata + +### 5.3 验收（DoD） + +- 2 个用户可各自提交 job，能看到自己的 job 列表与输出目录 +- 超时策略可触发（模拟短 timeout），job 被停止且状态标记为 timeout +- 队列在资源不足时保持 pending，资源释放后自动运行 + +--- + +## 6. MVP v3.1（可扩展性：自定义代码/Reward、多版本 VERL） + +> 目标：把“平台内置 workload”升级成“用户可提交自定义代码与 reward”，并支持多版本并存。 + +### 6.1 自定义代码提交（最小实现） + +两种方式二选一（建议先做 A）： + +- A：`working_dir` 指向 NFS 上的代码快照目录（用户自己准备/上传） +- B：上传 zip（控制面落到 NFS 并解压为 code snapshot） + +### 6.2 多版本 VERL 并存 + +约束前提：**基础镜像保持同一个**（生产环境容器由算力平台创建时已固定镜像标签）。 + +目标：在同一 Ray 集群内，不同 job 可以使用不同版本的 `verl`（例如不同分支/commit 或用户魔改版）。 + +已确认优先方案（A）：**必须通过 Ray Job 的 `runtime_env.env_vars` 透传 `PYTHONPATH`**，让 job 粒度优先 import 指定代码快照。 + +建议方案（以 NFS 为中心，最小可行实现）： + +- 在共享存储上以“不可变快照”的方式存放代码版本（推荐 commit hash 命名）： + - `${SHARED_ROOT}/common/code/verl//...` + - `${SHARED_ROOT}/users//code/verl//...`（用户魔改版） +- JobSpec 增加 `code_path`（指向上述目录），控制面在提交 job 时注入（必须走 runtime_env）： + - `runtime_env.env_vars.PYTHONPATH = ":$PYTHONPATH"`（把 code_path 放最前面，确保 import 优先级） + +示例（概念性，实际以 `${SHARED_ROOT}` 为准）： + +```bash +CODE_PATH="${SHARED_ROOT}/common/code/verl/" + +ray job submit \ + --address="http://127.0.0.1:8265" \ + --submission-id="" \ + --runtime-env-json='{"env_vars": {"PYTHONPATH": "'"${CODE_PATH}"':$PYTHONPATH"}}' \ + -- \ + python3 -m verl.trainer.main_ppo ... +``` + +需要验证的关键点（作为 v3.1 的 DoD 之一）： + +- 同时运行两个 job： + - jobA 使用 ``，jobB 使用 `` + - 互不影响，且各自训练/日志/ckpt 正常 +- job 粒度是否能做到“依赖隔离”（至少做到 `verl` 版本隔离；第三方依赖冲突可先假设镜像内一致） + +> 备注：当前 v1 的做法是容器内全局 `pip install -e /workspace/verl`，这会让所有 job 默认使用同一份 `verl`。要实现多版本并存，必须让 job 的 import 优先使用 `code_path`（或为每个 job 单独创建 venv/安装 wheel；后者更重，建议后置）。 + +### 6.3 自定义 reward function + +- JobSpec 支持 `reward_fn_path`（Python 模块路径） +- `reward_fn_path` 可指向共享存储中用户自定义代码目录（例如 `${SHARED_ROOT}/users//code/...`） + - 约束：代码必须在 job runtime 中可 import（由 `working_dir`/`PYTHONPATH` 或 runtime_env 保障） +- 控制面校验模块可导入（basic lint/安全白名单可后置） + +### 6.4 验收（DoD） + +- 同时运行两个 job：使用不同的 `verl` 代码版本（或用户魔改版本），互不影响 +- 用户可在 JobSpec 中替换 reward function 并跑通一个最小训练闭环 + +--- + +## 7. MVP v4.0（可观测性：Prometheus/Grafana + W&B 集成） + +> 目标：平台可运营：能回答“谁在用多少资源、跑了多久、利用率如何、是否空占 GPU”。 + +### 7.1 指标与监控 + +- Ray 指标接入 Prometheus（节点/任务/actor） +- GPU 指标：nvidia exporter 或 DCGM exporter +- Dashboard：Grafana（至少 3 张核心面板） + - 集群总 GPU/CPU 使用率、空闲率 + - 每 job 的 GPU 时间、峰值显存、运行时长 + - 节点健康（心跳/掉线）与告警 + +### 7.2 W&B（或等价）集成验证 + +- 最小可行：单机 self-host W&B server 可用性验证 +- JobSpec 支持启用/关闭 W&B，并传入 project/run name + +### 7.3 验收（DoD） + +- Grafana 上能看到集群与 job 资源视图 +- 某个 job GPU 利用率异常（模拟）能触发告警规则（邮件/IM/日志即可） +- W&B 指标能按 job 维度归档（至少 PPO 能上报） + +--- + +## 8. MVP v4.1（运维化：SOP + 自动化运维接口） + +> 目标：把平台变成“可交接”的系统：运维动作标准化，并为智能体留出接口。 + +### 8.1 SOP 与自动化入口 + +- SOP 文档： + - 节点上线/下线 + - 故障定位（Ray session、Ray job、NCCL、OOM） + - 资源回收（停止 job、清理 ckpt） +- 自动化接口（最小）： + - `/v1/ops/drain_node` + - `/v1/ops/restart_ray_head`（谨慎：需要保护与权限） + - `/v1/ops/cleanup_job_artifacts` + +### 8.2 验收（DoD） + +- 按 SOP，非开发人员可完成一次“节点上线→跑任务→下线→清理” +- 自动化接口至少能完成 1 个高频动作（如清理/停止/下线） + +--- + +## 9. MVP v5.0（Serving 与 Pipeline，偏长期） + +> 目标：训练-部署一体化：支持 model serving，并在平台内串联训练→评测→发布。 + +### 9.1 Serving + +- Ray Serve（或等价）部署模型推理服务 +- Serving 与训练共用模型库与权限（按 user/project） + +### 9.2 Pipeline（草稿里标为高级） + +- Pipeline 是对多个 job 的封装（训练→merge→eval→publish） +- 可先实现最小 DAG（两步串联）作为验证 + +### 9.3 验收（DoD） + +- 训练产物一键发布为一个可访问的推理 endpoint +- Pipeline 能自动串联并产出最终 artifact（可回滚/可追踪） + +--- + +## 10. 并行技术验证（建议尽早做） + +这些属于“跨版本”风险项，建议在 v1.1 ~ v2.0 期间尽早做： + +### 10.1 网络（IB / RoCEv2） + +- 确认环境是否支持 IB（H100）或 RoCEv2（H20） +- 跑最小 NCCL 通信验证（all-reduce / bandwidth） +- 将必要的 NCCL 环境变量注入到 job runtime_env + +### 10.2 Ray + 多节点容器约束 + +- 多容器同宿主机时的 Ray node_ip/临时目录冲突规律（已踩坑，需固化规范） +- 端口范围与防火墙策略（Ray worker 端口、dashboard、metrics） + +--- + +## 11. 已确认的约束与假设（来自讨论结论） + +这些会直接影响 v2.1（SSH 纳管）与后续多用户/存储设计： + +1) **最终形态仍以“每节点容器”运行**（不是裸机 systemd）。 + - H20 开发环境：我们可在宿主机用 `docker compose` 自建容器，并通过 SSH 进入容器调试/纳管。 + - H100 生产环境：容器由算力平台创建/回收；平台侧控制面只能 **SSH 进入这些容器** 做纳管（执行 `ray start/stop`、注入 env 等）。 +2) **认证**：内部 token 即可（MVP 阶段不对接 SSO）。 +3) **存储**：只考虑 NFS。 + - 开发环境：NFS/共享目录可通过宿主机 bind mount 提供给容器。 + - 生产环境：所有容器挂载相同 NFS，容器内共享根路径为 `/private/`（需要在实现时把“共享根路径”做成可配置项，而不是写死 `/mnt/shared`）。 +4) **网络拓扑约束**：暂不做按 IB 域/机架/拓扑的强约束调度（第 10.1 仍需验证 IB/RoCE 是否可用与配置方式，但调度不引入拓扑维度）。 +5) **共享目录分层**：在 `users//...` 之外增加一个可读写的 `common/` 目录用于共享数据/模型/代码： + - `${SHARED_ROOT}/common/datasets/` + - `${SHARED_ROOT}/common/models/` + - `${SHARED_ROOT}/common/code/` + - 权限（MVP）：先默认“所有内部 token 用户可读写”，后续再细化只读/受控写。 + +--- + +## 12. 仍需你确认/讨论的问题（剩余不确定项） + +1) `runtime_env.env_vars` 注入对“子进程/训练框架内部启动进程”的覆盖范围是否足够？ + - 需要确认 `verl`/`sglang` 等子进程是否继承 driver 的环境变量（通常会继承，但建议在 v3.1 验收时明确验证）。 diff --git a/specs/mvp/v1.1/mvp_plan.md b/specs/mvp/v1.1/mvp_plan.md new file mode 100644 index 0000000..99dbbc9 --- /dev/null +++ b/specs/mvp/v1.1/mvp_plan.md @@ -0,0 +1,169 @@ +# MVP v1.1 计划（Hardening + 多 Workload 可行性验证） + +本目录是 `specs/mvp/v1/` 的下一步迭代：在 v1 已经跑通（Ray head + 2 worker，PPO on Ray，持久化落盘）的基础上，把它升级为**可长期回归**的最小系统，并扩展至少一个新 workload 的可行性闭环。 + +> v1.1 的目标不是做平台服务化（API/队列/多用户）——那是 v2/v3 的工作；v1.1 聚焦“工程化 + 可行性边界验证 + 可观测/可排障基础”。 + +--- + +## 1. v1 基线回顾（已完成） + +- 拓扑：1 head（无 GPU，CPU/GPU=0）+ 2 worker（各 4 GPU） +- 提交方式：必须用 head 上的 `ray job submit` +- driver 调度：通过 `worker_node` 自定义资源 + `--entrypoint-resources` 强制 driver 在 worker +- 输出：按 `submission_id` 组织到共享目录（NFS） + +相关实现参考： + +- 脚本：`src/mvp/v1/` +- 验收动作：`specs/mvp/v1/v1_action.md` +- Roadmap：`specs/mvp/mvp_roadmap.md` + +--- + +## 2. v1.1 目标（必须达成） + +### 2.1 工程化（Hardening） + +1) **JobSpec 标准化（最小）** + - 把“提交 job 需要的参数”收敛成结构化文件： + - Ray 基础配置（YAML）：cluster 地址、entrypoint 资源约束、runtime_env 等 + - 训练 JobSpec（YAML）：workload 语义与训练参数 + - 至少覆盖：`submission_id`、workload 类型、资源需求、共享根路径、模型/数据路径、输出目录、超时、环境变量注入。 + - v1.1 实现落点（已在 repo 里提供，SDK 方式）： + - RayConfig 示例：`src/mvp/v1.1/py/configs/dev.yaml` + - JobSpec 示例：`src/mvp/v1.1/py/jobspecs/{ppo,grpo,sft}.yaml` + - 提交入口：`src/mvp/v1.1/py/run.py`（在 head 容器内执行，使用 Ray Python SDK 提交） + - 设计文档：`specs/mvp/v1.1/sdk_submit_refactor.md` + +2) **共享根路径抽象（dev/prod 一致）** + - 引入 `SHARED_ROOT` 作为唯一共享根路径： + - dev：建议也用 `/private`（docker compose 把宿主机 shared 挂到容器内 `/private`，模拟生产） + - prod：固定 `/private`（算力平台容器内 NFS） + - 任何代码/脚本不得写死 `/mnt/shared`（允许兼容旧路径但不得作为主路径）。 + +3) **共享目录分层（新增 `common/` 与 `user/`）** + - 在 `datasets/hf/jobs/outputs` 之外，新增一个所有用户可读写的共享区： + - `${SHARED_ROOT}/common/`：共享模型/数据/代码快照（多版本 verl / 公共数据） + - `${SHARED_ROOT}/user/`：用户自定义代码（例如 `reward_fn_path` 指向这里） + - v1.1 默认策略：先假设“所有用户可写”（后续 v3 再做权限与隔离）。 + +4) **可排障基础** + - 每个 job 目录必须有： + - `config/`：提交命令、JobSpec 快照、关键 env_vars + - `logs/`：Ray job logs + hydra logs（如有） + - `checkpoints/`：按 `save_freq` 控制频率（默认每 10 step） + - 提供“失败快照”能力：收集 `ray status` / `ray job list` / `ray list nodes` / `ray list actors`（最少其中 2 项）写入 job 目录。 + - v1.1 submitter 默认落盘： + - `${SHARED_ROOT}/jobs//config/job_spec.json` + - `${SHARED_ROOT}/jobs//config/runtime_env.json` + - `${SHARED_ROOT}/jobs//config/submit_cmd.txt` + - `${SHARED_ROOT}/jobs//logs/ray_job_submit.out` + - `${SHARED_ROOT}/jobs//debug/ray_status_{pre,post}.txt` + - `${SHARED_ROOT}/jobs//debug/ray_job_list_post.txt` + +### 2.2 Workload 扩展（至少新增 1 个） + +v1.1 需要新增并验收通过两个 workload（都要跑通闭环）： + +- **GRPO on Ray**（推荐优先，复用 PPO 入口，通过算法配置切换） + - 基于 `python -m verl.trainer.main_ppo` + - 通过配置覆盖：`algorithm.adv_estimator=grpo`（以及必要的 rollout 参数） + + - **SFT on Ray（Ray-native）** + - 入口：`python -m verl.trainer.sft_trainer_ray` + - 参考实现：`verl/verl/trainer/sft_trainer_ray.py`（内部会 `ray.init()`） + - 需要确保 `ray.init()` 连接已有集群： + - 优先：`runtime_env.env_vars.RAY_ADDRESS=auto`（配合 `ray job submit`） + - 兜底：在 v1.1 的 launcher 脚本里显式 `ray.init(address="auto")` 再调用 trainer（避免依赖 Ray 的 env var 行为差异） + - 重要细节：Ray Job 的 entrypoint（driver）默认不分配 GPU，因此 SFT driver 侧不要强依赖 CUDA： + - 推荐：`trainer.device=cpu`（driver 只做 orchestration；训练由 Ray workers 占 GPU） + +--- + +## 3. v1.1 关键设计点 + +### 3.1 多版本代码与自定义逻辑（为 v3.1 铺路，但 v1.1 先做最小验证） + +已确定优先方案（A）：通过 **Ray Job 的 `runtime_env.env_vars`** 注入 `PYTHONPATH`。 + +- `code_path`（例如 `${SHARED_ROOT}/common/code/verl/`） +- 提交 job 时设置： + - `runtime_env.env_vars.PYTHONPATH = ":$PYTHONPATH"` + +并约定： + +- `reward_fn_path` 可指向 `${SHARED_ROOT}/user/code/...` 下用户自定义代码 +- 与 `code_path` 一样，必须通过 `runtime_env.env_vars` 确保该路径可被 import（例如把 `${SHARED_ROOT}/user/code` 也加入 `PYTHONPATH`） + +v1.1 中至少做一次“代码覆盖验证”： + +- 在 code_path 下放一个可识别的 `verl` 版本标识（例如 `verl.__version__` 打印差异） +- 提交 job 并在日志中确认 import 的是 code_path 的版本（而不是镜像内默认安装） + +v1.1 的最小落地方式（已实现）： + +- 提供代码快照脚本：`src/mvp/v1.1/scripts/31_snapshot_verl_code.sh` + - 会把 `/workspace/verl`（挂载的 repo）复制到 `${SHARED_ROOT}/common/code/verl//` + - 并写入 `${code_path}/mvp_marker.py`，用于在 Ray job logs 中验证“选用的是哪份 code_path” +- submitter 会在 entrypoint 前运行 preflight： + - 打印 `verl.__file__` 与 `mvp_marker.MARKER` + - 由此确认 job 粒度的 PYTHONPATH 生效，且不同 job 可指向不同 `code_path`（多版本共存） + +### 3.2 Checkpoint 策略（磁盘保护） + +- 默认：`save_freq=10`（每 10 step 保存一次） +- 对于 step 数已知的短任务（例如 29 steps），可以通过配置把 `save_freq` 调整为 10/15/29（按需求权衡） +- 作业目录按 `submission_id` 隔离，方便清理与归档 + +--- + +## 4. v1.1 交付物清单（代码 + 文档） + +### 4.1 代码（建议落点） + +在 `src/mvp/` 下新增 v1.1 级别的提交器与模板（或在 `src/mvp/v1` 原地演进但要保持 v1 可回归）： + +- `src/mvp/v1.1/` + - `docker-compose.yaml`（与 v1 互不干扰的容器名/网络名） + - `scripts/`（Ray 启动/prepare 保留 bash；submit 通过 SDK 工具执行） + - `py/`（工程化提交层：YAML + Ray Python SDK） + - `py/configs/`（Ray 基础配置） + - `py/jobspecs/`（训练 JobSpec） + - `py/run.py`（入口） + +此外，为了对齐 dev 环境约束（远程机固定目录）： + +- 远程机目录必须新增：`argus@h1:/home2/argus/infra/mvp/v1.1/` +- 该目录内需包含 v1.1 的全部内容（compose + scripts + README），可由本 repo 的 `src/mvp/v1.1/` 同步过去 + +### 4.2 文档 + +- `specs/mvp/v1.1/v1.1_action.md`：开发、部署、测试、验收流程（可复现） +- 更新 `specs/mvp/mvp_roadmap.md`：保持路线图与落地一致（按需） + +--- + +## 5. v1.1 验收标准（DoD） + +### 5.1 Hardening DoD + +- [ ] 所有提交均由 head 执行 `ray job submit`，且显式 `--submission-id=` +- [ ] 共享根路径由 `SHARED_ROOT` 控制（dev/prod 可切换），脚本无硬编码 +- [ ] 每个 job 的输出目录为：`${SHARED_ROOT}/jobs//` +- [ ] checkpoint 不会“每 step 保存”导致爆盘：默认 `save_freq=10` +- [ ] job 失败时，`${SHARED_ROOT}/jobs//config/` 中有足够信息定位（命令、env、ray 状态快照） + - [ ] v1.1 测试前会清理 v1 的遗留容器/进程（避免端口、容器名、Ray session 干扰） + +### 5.2 Workload DoD（GRPO + SFT 都必须） + +GRPO（必须）： + +- [ ] `algorithm.adv_estimator=grpo` 的 job 可提交并进入 RUNNING +- [ ] job 能跑完最小训练步数（可设 `total_epochs=1` 或 `total_training_steps`） +- [ ] 输出目录内有日志与至少 1 次 checkpoint（或明确不保存并说明原因） + +SFT（必须）： + +- [ ] `sft_trainer_ray` 可连接集群并跑到至少 1 个 step（推荐最小训练步数/epoch） +- [ ] 输出目录与 checkpoint 策略同 v1.1 规范（落盘到 `${SHARED_ROOT}/jobs//...`） diff --git a/specs/mvp/v1.1/sdk_submit_refactor.md b/specs/mvp/v1.1/sdk_submit_refactor.md new file mode 100644 index 0000000..9c6412b --- /dev/null +++ b/specs/mvp/v1.1/sdk_submit_refactor.md @@ -0,0 +1,148 @@ +# MVP v1.1 工程化重构方案：Ray Python SDK 提交层（YAML Config + YAML JobSpec） + +本文档把 v1.1 的“代码工程化”目标落到一个明确的设计：**保留现有 scripts**（Ray 集群构建、数据准备、模型准备、代码快照），将“任务提交机制”重构为 **Ray Python SDK**（`ray.job_submission.JobSubmissionClient`）驱动的 Python 工具层。 + +> 约束（已确认） +> 1) 基础配置用 YAML，JobSpec 也用 YAML。 +> 2) 工具必须在 **head 容器**执行（从 head 发起提交，满足“在 head 提交”的要求）。 +> 3) 训练参数组织保持与现在一致：仍然使用 **Hydra overrides** 方式构造 entrypoint。 +> 4) 不使用 `requests` 直连 HTTP API（只用 Ray SDK）。 + +--- + +## 1. 当前 Ray SDK 能力验证（关键前提） + +在 head 容器（`mvp11-ray-head`）中验证： + +- Ray 版本：`2.51.1` +- `JobSubmissionClient.submit_job` 支持以下关键字段： + - `submission_id` + - `runtime_env` + - `entrypoint_num_cpus` + - `entrypoint_num_gpus` + - `entrypoint_resources`（用于强制 driver 落 worker） + +因此 v1.1 可以“纯 SDK”完成提交，不需要 `requests` fallback。 + +--- + +## 2. 系统分层（不动 scripts，只重构提交层） + +### 2.1 scripts（保留） + +`src/mvp/v1.1/scripts/` 继续负责： + +- 容器生命周期：`01_up.sh` / `02_down.sh` +- Ray 启动：`20_start_head.sh` / `21_start_workers.sh` +- 数据/模型准备：`30_prepare_data_and_model.sh` +- 代码快照：`31_snapshot_verl_code.sh`（生成 `${SHARED_ROOT}/common/code/verl//`） + +scripts 可以新增一个“薄封装”脚本，负责 `docker exec` 进 head 容器并运行 Python 提交器，但 scripts 不再拼 `ray job submit ...` CLI 字符串。 + +### 2.2 Python 工具层（新增） + +在 `src/mvp/v1.1/py/` 新增提交工具层： + +- 读取 Ray 基础配置（YAML） +- 读取训练 JobSpec（YAML） +- 用 Ray Python SDK 提交/查询/停止/拉日志 +- 将 job 级别产物落盘到：`${SHARED_ROOT}/jobs//...` + +--- + +## 3. 输入定义：两份 YAML + +### 3.1 Ray 基础配置（RayConfig YAML） + +这份配置是“稳定可复用”的，描述 cluster 与 driver placement 等通用信息。 + +字段建议： + +- `address`: `http://127.0.0.1:8265`（从 head 容器内部视角） +- `shared_root`: `/private` +- `entrypoint_num_cpus`: `1` +- `entrypoint_resources`: `{"worker_node": 1}`（强制 driver 使用 worker 才有的资源） +- `runtime_env.env_vars`: HF cache / endpoint 等通用环境变量 +- `user_code_path`: `${shared_root}/user/code`（可选，默认值也可） + +### 3.2 训练 JobSpec（JobSpec YAML） + +这份配置是“一次训练”语义，描述 workload + 训练参数 + code_path 多版本等。 + +字段建议： + +- `workload`: `ppo|grpo|sft` +- `submission_id`: 可选（不填则生成；但最终必须显式传给 SDK） +- `code_path`: `${shared_root}/common/code/verl/`（多版本关键字段） +- `model_id` +- 数据路径：`train_file` / `val_file`（按 workload） +- 训练参数：`nnodes` / `n_gpus_per_node` / `total_training_steps` / `save_freq` / `test_freq` + +注意（SFT 的 driver 设备选择）： + +- Ray job 的 entrypoint（driver）默认不分配 GPU（我们通常不设置 `entrypoint_num_gpus`）。 +- `sft_trainer_ray.py` 的 driver 会用 `trainer.device` 做张量统计；若设置为 `cuda` 且 driver 无 GPU，会报： + - `RuntimeError: No CUDA GPUs are available` +- 因此 v1.1 的 SFT JobSpec 默认应设置：`trainer.device=cpu`（训练 workers 仍会占用 GPU）。 + +--- + +## 4. Python 提交器的职责（tool class） + +建议实现 `RayJobTool`（或类似命名），能力： + +### 4.1 submit（核心） + +输入：`RayConfig + JobSpec` +输出：`submission_id` + +实现要点： + +- `client = JobSubmissionClient(address)` +- 生成/确定 `submission_id` +- `runtime_env` 合并逻辑： + - 合并 config 与 jobspec 的 `env_vars` + - 强制注入多版本： + - `PYTHONPATH = "::$PYTHONPATH"` +- 构造 entrypoint（保持 hydra overrides 风格）： + - PPO/GRPO：`python3 -m verl.trainer.main_ppo ...` + - SFT：`python3 -m verl.trainer.sft_trainer_ray ...` +- 强制 driver 落 worker： + - `entrypoint_resources=config.entrypoint_resources` + - `entrypoint_num_cpus=config.entrypoint_num_cpus` +- 落盘产物： + - `${shared_root}/jobs//config/{ray_config.yaml,jobspec.yaml,submit_payload.json}` + - `${shared_root}/jobs//logs/submit.out` + - `${shared_root}/jobs//debug/{ray_status_pre,ray_job_list_post}.txt`（可用 SDK 或 `ray status` 采集） + +### 4.2 status / stop / logs / list + +- `status(submission_id)` +- `stop(submission_id)` +- `logs(submission_id)`（可支持 tail） +- `list()` + +--- + +## 5. `run.py` 入口（必须在 head 容器执行） + +建议入口： + +- `python3 /workspace/mvp/v1.1/py/run.py --config --jobspec --action submit` +- `--action` 支持：`submit|status|stop|logs|list` + +host 侧执行方式（由 scripts 薄封装）： + +- `docker exec mvp11-ray-head python3 /workspace/mvp/v1.1/py/run.py ...` + +--- + +## 6. 验收口径（工程化部分） + +1) **SDK 提交**：不使用 `ray job submit` CLI，改用 `JobSubmissionClient.submit_job`。 +2) **driver 仍强制在 worker**：SDK 提交时 `entrypoint_resources={"worker_node":1}` 生效。 +3) **多版本共存验证**： + - 通过 `31_snapshot_verl_code.sh` 生成 `codeA/codeB` 两份 code_path + - 通过两份 JobSpec 分别指向不同 `code_path` + - 在 job logs 中看到不同的 marker（例如 `mvp_marker.MARKER`） + diff --git a/specs/mvp/v1.1/v1.1_action.md b/specs/mvp/v1.1/v1.1_action.md new file mode 100644 index 0000000..cd72a7f --- /dev/null +++ b/specs/mvp/v1.1/v1.1_action.md @@ -0,0 +1,333 @@ +# MVP v1.1 行动文档（实施方案 / 部署测试 / 验收口径） + +本文档面向“把 v1 跑通的实验脚本，升级为可长期回归的 v1.1 最小系统”，并给出**开发改造 → 部署测试 → 验收**的可复现流程。 + +> v1.1 的核心约束（来自讨论结论） +> - 仍然必须通过 **head 节点执行 `ray job submit`** 提交任务。 +> - 训练/driver **必须落在 worker**（head 不跑训练）。 +> - 多版本 `verl` 共存：同一镜像不变，必须通过 **Ray Job `runtime_env.env_vars` 注入 `PYTHONPATH`** 让 job 粒度选择代码版本。 +> - 存储只考虑 NFS：dev 环境我们自己 mount；生产环境容器内统一看到 `/private/`。 + +--- + +## 1. 目标与非目标 + +### 1.1 目标（v1.1 必须做到） + +1) **可回归**：同一环境连续跑多次 PPO 回归，不互相覆盖，输出按 submission id 归档。 +2) **可扩展**：新增并验收通过 2 个 workload（**GRPO + SFT**）并跑通闭环。 +3) **可排障**：每个 job 目录包含完整的提交快照、关键 env、Ray 状态快照与日志入口。 +4) **可多版本共存**：同一 Ray 集群内，不同 job 通过 `PYTHONPATH` 选择不同 `verl` 代码快照。 + +### 1.2 非目标（v1.1 不做） + +- 不做平台 API/队列/多租户/RBAC（这是 v2/v3）。 +- 不做复杂调度（拓扑、IB 域、NUMA、Gang 等自动化策略）。 + +--- + +## 2. 运行环境约定（dev / prod 一致抽象） + +### 2.1 拓扑（单机 3 容器） + +- `mvp-ray-head`：无 GPU，`ray start --head --num-cpus=0 --num-gpus=0`（控制面 only） +- `mvp-ray-worker-0`：4 GPU +- `mvp-ray-worker-1`：4 GPU + +### 2.2 “head 不跑训练”的硬约束实现（必须） + +1) **head CPU=0**：从资源层面阻断默认 task/driver 落到 head。 +2) **worker 自定义资源标签**：worker 启动时带 `--resources='{"worker_node": 100}'`。 +3) **ray job submit 强制 entrypoint 落 worker**：提交时必须带： + - `--entrypoint-resources='{"worker_node": 1}'` + - `--entrypoint-num-cpus=1`（显式声明 driver 需要的 CPU） + +> 验证口径：`ray job list` 的 `driver_info.node_ip_address` 必须是 worker 的 IP，而不是 head IP。 + +### 2.3 共享存储（NFS）与路径（关键） + +- 生产环境：容器内共享根路径固定为 `/private/`（算力平台统一挂载 NFS）。 +- 开发环境：docker compose 也应把宿主机共享目录挂载到容器内的 `/private/`，从而做到 dev/prod 一致。 + +统一约定（容器内视角）： + +- `SHARED_ROOT=/private` +- Job 输出：`${SHARED_ROOT}/jobs//` + +建议的共享目录结构（v1.1 新增 `common/` 与 `user/`）： + +- `${SHARED_ROOT}/datasets/`：通用数据（例如 gsm8k parquet） +- `${SHARED_ROOT}/hf/`：HuggingFace cache（模型/分词器/权重） +- `${SHARED_ROOT}/jobs/`：按 submission id 归档的作业目录（强制） +- `${SHARED_ROOT}/outputs/`：临时/非强约束输出（不建议长期依赖） +- `${SHARED_ROOT}/ray/`：Ray 调试痕迹（可选，通常 Ray 默认写 `/tmp/ray`） +- `${SHARED_ROOT}/common/`：所有用户可读写共享区（模型/数据/代码快照） + - `${SHARED_ROOT}/common/models/`：可复用基础模型（可用软链指向 hf cache 或 snapshot） + - `${SHARED_ROOT}/common/datasets/`：共享数据（或与 `datasets/` 统一规划） + - `${SHARED_ROOT}/common/code/`：代码快照（多版本 `verl` / 自定义 reward） +- `${SHARED_ROOT}/user/`：用户自定义内容（默认所有用户可写） + - `${SHARED_ROOT}/user/code/`：reward_fn 等自定义 Python 代码 + +--- + +## 3. 开发实施方案（代码改造清单） + +> v1.1 建议新增 `src/mvp/v1.1/`（保持 v1 可回归不被破坏）。 + +### 3.1 JobSpec（最小标准化） + +v1.1 的工程化目标是把“提交机制”迁移到 Ray Python SDK，因此输入拆为两份 YAML： + +1) Ray 基础配置（YAML）：address / entrypoint resources / runtime_env 等 +2) 训练 JobSpec（YAML）：workload 语义与训练参数（仍由 Hydra overrides 组织） + +训练 JobSpec（YAML）至少包含： + +- `submission_id`：可空；为空时由 submitter 生成（但最终必须显式传给 `ray job submit --submission-id`） +- `workload`：`ppo` / `grpo` / `sft`（v1.1 必须 `ppo` + `grpo` + `sft`） +- `shared_root`：默认 `/private`（容器内路径） +- `code_path`：`verl` 代码快照目录（用于多版本共存） +- `reward_fn_path`（可选）：指向 `${shared_root}/user/code/...` 下的 Python 文件或模块入口 +- `model` / `dataset`：必须指向共享存储的持久化路径（避免每次下载/生成） +- `ray`：`address=http://127.0.0.1:8265`（从 head 容器内部视角） +- `resources`： + - `entrypoint_resources={"worker_node":1}` + - `entrypoint_num_cpus=1` +- `trainer_overrides`：训练参数覆盖（v1.1 默认 `total_epochs=1`、`save_freq=10`） +- `env_vars`：会被透传到 `runtime_env.env_vars`（必须包含 `PYTHONPATH` 注入） + +交付物（v1.1 SDK 方式）： + +- `src/mvp/v1.1/py/configs/dev.yaml`（Ray 基础配置示例） +- `src/mvp/v1.1/py/jobspecs/{ppo,grpo,sft}.yaml`（训练 JobSpec 示例） +- `src/mvp/v1.1/py/run.py`（入口：使用 Ray Python SDK 提交/查询/停止/拉日志） +- 设计文档：`specs/mvp/v1.1/sdk_submit_refactor.md` + +### 3.2 多版本 `verl` 共存（必须） + +原则：**镜像固定不变**；job 粒度通过 `PYTHONPATH` 选择 `verl` 代码快照。 + +提交时必须注入（runtime_env）： + +- `PYTHONPATH=":$PYTHONPATH"`（`CODE_PATH` 放最前面） + +并要求 job 在日志中打印一行确认 import 来源，例如： + +- `python -c "import verl,inspect; print(verl.__file__)"`（或训练入口启动时打印） + +v1.1 具体实现（可复现）： + +- 先用 `src/mvp/v1.1/scripts/31_snapshot_verl_code.sh` 生成代码快照目录 `${SHARED_ROOT}/common/code/verl//` + - 该目录里会包含一个 `mvp_marker.py`（`MARKER=`） +- 提交 job 时让 `code_path` 指向该快照目录；submitter 会在 entrypoint 前打印： + - `MVP_PRECHECK_VERL_FILE`（验证 import 来源） + - `MVP_PRECHECK_MARKER`（验证选择的 code_path） + +### 3.3 `submit_job` 工具（组装 ray job submit） + +新增一个提交器（建议 Python，避免复杂 bash quoting）： + +- 输入：JobSpec JSON +- 产物： + - 生成/确定 `submission_id` + - 创建 `${SHARED_ROOT}/jobs//config/`、`logs/`、`checkpoints/` + - 写入 `config/job_spec.json`（原样快照） + - 写入 `config/runtime_env.json`（最终用于 submit 的 JSON） + - 写入 `config/submit_cmd.txt`（最终命令行） +- 执行：在 **head 容器内**运行 `ray job submit ...` + +### 3.4 可排障：debug bundle（强制落盘） + +在 job 生命周期的关键节点收集并落盘（至少 2 类）： + +- `ray status` +- `ray job list` +- `ray list nodes` +- `ray list actors` + +建议落盘到： + +- `${SHARED_ROOT}/jobs//debug/`（每次收集带时间戳文件名） + +### 3.5 Workload 扩展：GRPO（v1.1 新增闭环） + +优先用与 PPO 相同入口 `python -m verl.trainer.main_ppo`，仅通过配置切换算法： + +- `algorithm.adv_estimator=grpo` +- 其余保持最小可跑：`total_epochs=1`、`save_freq=10` + +### 3.6 Workload 扩展：SFT on Ray（v1.1 必须新增闭环） + +#### 3.6.1 入口与参考实现 + +- 入口：`python -m verl.trainer.sft_trainer_ray` +- 参考代码：`verl/verl/trainer/sft_trainer.py`（非 Ray 版本）与 `verl/verl/trainer/sft_trainer_ray.py`（Ray 版本） + +> v1.1 要验收的是 “SFT on Ray”，因此默认使用 `sft_trainer_ray.py`。 + +#### 3.6.2 连接已有 Ray 集群（必须） + +`sft_trainer_ray.py` 内部直接调用 `ray.init()`，为了确保它连接到**已有集群**（head+workers），v1.1 约定： + +- 提交 job 时通过 `runtime_env.env_vars` 注入：`RAY_ADDRESS=auto` + +如果发现 `ray.init()` 未按预期读取 `RAY_ADDRESS`（Ray 版本差异风险），v1.1 需要提供一个 launcher 兜底： + +- 由 launcher 先显式 `ray.init(address="auto")`，再调用 SFT trainer 逻辑 + +#### 3.6.3 SFT 数据格式（parquet schema） + +`sft_trainer_ray` 默认使用 `MultiTurnSFTDataset`，parquet 中至少需要： + +- `messages` 列：list[dict]，dict 至少含 `role`/`content` + +v1.1 的 `prepare` 阶段需要生成并持久化 SFT 数据，例如： + +- `${SHARED_ROOT}/datasets/gsm8k_sft/train.parquet` +- `${SHARED_ROOT}/datasets/gsm8k_sft/val.parquet`（可选） + +单条样本的 `messages` 形态示例： + +- `[{ "role": "user", "content": "" }, { "role": "assistant", "content": "" }]` + +> 注意：SFT parquet 不能直接复用 PPO/RL 的 parquet（schema 不同）。 + +#### 3.6.4 重要细节：SFT Ray Driver 不应依赖 GPU + +在 `ray job submit` 模式下，我们的 entrypoint（driver）默认 **不会分配 GPU**（我们只指定了 `--entrypoint-num-cpus=1`，没有指定 `--entrypoint-num-gpus`）。 + +而 `verl/verl/trainer/sft_trainer_ray.py` 的 driver 逻辑里会用 `trainer.device` 来创建 `torch.tensor(..., device=...)` 做统计，如果设置为 `cuda` 且 driver 没有 GPU，会触发： + +- `RuntimeError: No CUDA GPUs are available` + +因此 v1.1 的 SFT on Ray 验收默认要求： + +- `trainer.device=cpu`（driver 只做 orchestration；真正训练仍由 Ray 的 TrainingWorker/资源池占用 GPU） + +### 3.7 v1.1 脚本化交付（必须独立完整） + +`src/mvp/v1.1/` 需要像 v1 一样提供一套完整脚本，确保 v1.1 可独立运行、可回归： + +- `src/mvp/v1.1/docker-compose.yaml`（容器名建议与 v1 区分，避免冲突） +- `src/mvp/v1.1/scripts/00_prereq_check.sh`（含 GPU/目录/NFS/verl 代码检查） +- `src/mvp/v1.1/scripts/01_up.sh` / `02_down.sh`（起停） +- `src/mvp/v1.1/scripts/20_start_head.sh` / `21_start_workers.sh` +- `src/mvp/v1.1/scripts/30_prepare_data_and_model.sh`（包含 PPO 数据 + SFT 数据） +- `src/mvp/v1.1/scripts/40_submit_ppo_epoch1.sh` +- `src/mvp/v1.1/scripts/41_submit_grpo_epoch1.sh` +- `src/mvp/v1.1/scripts/42_submit_sft_minimal.sh` +- `src/mvp/v1.1/scripts/50_status.sh` +- `src/mvp/v1.1/scripts/31_snapshot_verl_code.sh`（多版本 code snapshot） +- `src/mvp/v1.1/scripts/43_submit_jobspec.sh`（通过 JobSpec 提交） +- `src/mvp/v1.1/scripts/12_install_py_deps.sh`（安装 PyYAML 等依赖） +- `src/mvp/v1.1/scripts/44_submit_sdk.sh`（通过 Ray Python SDK + YAML 提交） + +--- + +## 4. 部署与测试流程（dev 环境） + +> dev 环境以远程机目录为例：`argus@h1:/home2/argus/infra/mvp`。v1.1 的所有内容要求放在： +> +> - `argus@h1:/home2/argus/infra/mvp/v1.1/` +> +> 并在该目录中通过脚本使用 `docker exec` 协调容器。 + +### 4.0 清理 v1 环境（必须先做） + +v1 已在 `argus@h1` 部署过容器与 Ray。为保证 v1.1 的可重复测试，开始 v1.1 前必须清理 v1： + +1) 停止并删除 v1 容器（推荐用 v1 的 down 脚本） +2) 确认 `docker ps` 中不再有 v1 的 `mvp-ray-head/mvp-ray-worker-*` + +v1.1 的脚本里也提供了一个 best-effort 清理脚本：`src/mvp/v1.1/scripts/03_cleanup_v1_legacy.sh`（远程目录中同名脚本）。 + +### 4.1 环境准备（一次性 / 幂等） + +1) 目录检查（远程机）： + - `${WORKDIR}/shared/` 存在并具备上述子目录（含 `common/`、`user/`） +2) `verl` 代码目录检查： + - `${WORKDIR}/verl` 不存在则执行 `git clone https://github.com/volcengine/verl.git` +3) GPU 可用性检查： + - 设备存在（例如 0-7 可见），并按 worker 容器分配（每个 worker 4 GPU） +4) 模型与数据持久化路径： + - 模型与数据必须落在 `${SHARED_ROOT}` 下；若已存在则跳过下载/生成 + - SFT parquet 同样必须落在 `${SHARED_ROOT}` 下；若已存在则跳过生成 + +### 4.2 启动 Ray 集群（每次测试） + +1) `docker compose up -d` +2) head：`ray start --head --num-cpus=0 --num-gpus=0 ...` +3) workers：`ray start --address=:6379 --resources='{"worker_node":100}' ...` +4) 验证：`ray status` 显示 1 head + 2 worker，且 head `CPU:0 GPU:0` + +### 4.3 提交 PPO 回归（必须跑 2 次） + +1) 生成 JobSpec（可用模板 + 覆盖项） +2) 在 head 容器内执行 submitter（或直接 `ray job submit`） +3) 验证要点： + - `ray job list`：driver node 是 worker + - `${SHARED_ROOT}/jobs//` 下存在 `config/`、`logs/`、`checkpoints/` + - checkpoint 每 10 step 产生（例如 `global_step_10`） + +### 4.4 提交 GRPO（新增 workload 验收） + +同 PPO，但覆盖 `algorithm.adv_estimator=grpo`，确保能进入 RUNNING 并完成最小步数。 + +### 4.5 提交 SFT on Ray（新增 workload 验收，必须） + +1) 确认 `${SHARED_ROOT}/datasets/gsm8k_sft/train.parquet` 已存在（由 v1.1 prepare 生成）。 +2) 通过 head 容器执行 `ray job submit` 提交 `python -m verl.trainer.sft_trainer_ray`。 +3) 关键约束： + - `runtime_env.env_vars.RAY_ADDRESS=auto`（连接已有集群） + - `--entrypoint-resources='{"worker_node": 1}'`（driver 落 worker） + - `PYTHONPATH=:$PYTHONPATH`（多版本 verl） +4) 最小化训练配置建议（避免 OOM/耗时过长）： + - `trainer.total_epochs=1` + - `trainer.total_training_steps=10~30` + - `trainer.save_freq=10` + - `trainer.nnodes=2`、`trainer.n_gpus_per_node=4`（用满 8 卡做一次最小分布式验证） + - `data.train_files=${SHARED_ROOT}/datasets/gsm8k_sft/train.parquet` + - `trainer.default_local_dir=${SHARED_ROOT}/jobs//checkpoints` + +### 4.6 工程化验证：JobSpec + 多版本共存（v1.1 必须） + +1) 生成两个 code snapshot（不同 `CODE_ID`）： + - `CODE_ID=codeA ./scripts/31_snapshot_verl_code.sh` + - `CODE_ID=codeB ./scripts/31_snapshot_verl_code.sh` +2) 分别修改/复制 JobSpec 模板，使 `code_path` 指向不同 snapshot： + - `${SHARED_ROOT}/common/code/verl/codeA` + - `${SHARED_ROOT}/common/code/verl/codeB` +3) 用 JobSpec 提交（必须从 head）： + - `./scripts/43_submit_jobspec.sh /workspace/mvp/v1.1/templates/ppo.json`（示例） +4) 在 Ray job logs 中验证： + - `MVP_PRECHECK_MARKER` 打印为对应的 `codeA`/`codeB` + - `MVP_PRECHECK_VERL_FILE` 指向 `${SHARED_ROOT}/common/code/verl/...` 而不是镜像内 site-packages + +--- + +## 5. 验收标准（Definition of Done） + +### 5.1 Hardening DoD（全部必选） + +- [ ] 提交必须来自 head：能在 head 容器内看到 `ray job submit ...` 的提交记录 +- [ ] driver 不在 head：`ray job list` 的 `driver_info.node_ip_address` ∈ worker IP，且 ≠ head IP +- [ ] 输出目录按 submission id 隔离：`${SHARED_ROOT}/jobs//` 不复用、不覆盖 +- [ ] 数据/模型持久化：再次提交时不重复下载/生成（有 “skip if exists” 的日志） +- [ ] checkpoint 策略有效：默认 `save_freq=10`，不会每 step 保存爆盘 +- [ ] debug bundle 落盘：`${SHARED_ROOT}/jobs//debug/` 至少包含 2 类 Ray 状态快照 +- [ ] 多版本共存验证通过：日志中能确认 `verl` import 来源来自 JobSpec 指定的 `code_path` + +### 5.2 Workload DoD（GRPO + SFT 都必须） + +- [ ] GRPO job 能提交、RUNNING、完成最小训练步数 +- [ ] GRPO job 产物目录满足与 PPO 相同的目录规范与 debug 规范 +- [ ] SFT job 能提交、连接已有集群并跑到至少 1 个 step（建议最小步数/epoch） +- [ ] SFT job 产物目录满足与 PPO 相同的目录规范与 debug 规范 + +--- + +## 6. 生产环境部署注意事项（v1.1 需要考虑但不强制在 dev 全量模拟） + +- 容器由算力平台创建：我们只负责 SSH 进去纳管（启动 ray / 提交 job / 收集产物）。 +- 容器内共享路径为 `/private`：所有脚本必须以 `SHARED_ROOT=/private` 工作，不得写死 `/mnt/shared`。 +- 认证仅内部 token：在 submitter 中把 token 作为 env var 透传（不写入日志明文）。 diff --git a/src/mvp/v1.1/README.md b/src/mvp/v1.1/README.md new file mode 100644 index 0000000..ffbbe67 --- /dev/null +++ b/src/mvp/v1.1/README.md @@ -0,0 +1,61 @@ +# MVP v1.1（GRPO + SFT on Ray）运行说明 + +本目录是一套**独立可运行**的 v1.1 交付：使用 1 个 Ray head（不跑训练）+ 2 个 Ray worker（各 4 GPU）在同一宿主机通过 `docker exec` 协调容器，并通过 **head 上的 `ray job submit`** 提交作业，同时强制 driver 落到 worker。 + +> 远程 dev 环境推荐目录布局： +> +> - `/home2/argus/infra/mvp/` +> - `shared/`（持久化：datasets/hf/jobs/...） +> - `verl/`（代码仓库，用于 prepare / snapshot） +> - `v1.1/`（本目录内容：compose + scripts） + +--- + +## 快速开始（远程机 argus@h1） + +在 `/home2/argus/infra/mvp/v1.1/` 下执行： + +```bash +./scripts/00_prereq_check.sh +./scripts/01_up.sh +./scripts/20_start_head.sh +./scripts/21_start_workers.sh +./scripts/30_prepare_data_and_model.sh +./scripts/12_install_py_deps.sh +./scripts/44_submit_sdk.sh /workspace/mvp/v1.1/py/configs/dev.yaml /workspace/mvp/v1.1/py/jobspecs/ppo.yaml +./scripts/44_submit_sdk.sh /workspace/mvp/v1.1/py/configs/dev.yaml /workspace/mvp/v1.1/py/jobspecs/grpo.yaml +./scripts/44_submit_sdk.sh /workspace/mvp/v1.1/py/configs/dev.yaml /workspace/mvp/v1.1/py/jobspecs/sft.yaml +./scripts/40_submit_ppo_epoch1.sh +./scripts/41_submit_grpo_epoch1.sh +./scripts/42_submit_sft_minimal.sh +./scripts/50_status.sh +``` + +说明： + +- `scripts/40/41/42` 是历史的 “CLI 提交脚本”（仍可用），但 v1.1 的工程化目标是把提交机制迁移到 `scripts/44_submit_sdk.sh`（Ray Python SDK + YAML 配置）。 + +停止并清理： + +```bash +./scripts/02_down.sh +``` + +--- + +## 关键约束（必须满足） + +- **必须通过 head 执行 `ray job submit`** 提交任务（满足“从 head 提交”要求）。 +- **head 不跑训练**：head 以 `--num-cpus=0 --num-gpus=0` 启动；worker 具备自定义资源 `worker_node`；提交时 `--entrypoint-resources='{"worker_node": 1}'` 强制 driver 落 worker。 +- **共享路径统一为 `/private`（容器内）**：compose 将宿主机 `../shared` 挂载到容器内 `/private`，对齐生产环境。 +- **多版本 verl**：通过 Ray Job `runtime_env.env_vars.PYTHONPATH` 注入 `${SHARED_ROOT}/common/code/verl/...`，job 粒度选择代码快照。 + +--- + +## 共享目录（容器内 /private） + +- `/private/datasets/`：数据（PPO 的 gsm8k RL parquet、SFT parquet） +- `/private/hf/`：HF 缓存（模型持久化，避免重复下载） +- `/private/jobs//`：每个 Ray Job 的输出目录（logs/config/debug/checkpoints） +- `/private/common/`：共享区（模型/数据/代码快照） +- `/private/user/`：用户自定义代码（例如 reward_fn） diff --git a/src/mvp/v1.1/docker-compose.yaml b/src/mvp/v1.1/docker-compose.yaml new file mode 100644 index 0000000..a885956 --- /dev/null +++ b/src/mvp/v1.1/docker-compose.yaml @@ -0,0 +1,89 @@ +version: "3.8" + +services: + ray_head: + image: verlai/verl:sgl055.latest + container_name: mvp11-ray-head + command: sleep infinity + ports: + - "8265:8265" + volumes: + - ../verl:/workspace/verl + - ../shared:/private + - .:/workspace/mvp/v1.1 + shm_size: "10g" + ulimits: + nofile: + soft: 65536 + hard: 65536 + cap_add: + - SYS_ADMIN + - SYS_PTRACE + networks: + - mvp11-ray-net + environment: + HF_HOME: "/private/hf" + HUGGINGFACE_HUB_CACHE: "/private/hf/hub" + TRANSFORMERS_CACHE: "/private/hf/transformers" + HF_ENDPOINT: "https://hf-mirror.com" + PYTHONUNBUFFERED: "1" + + ray_worker_0: + image: verlai/verl:sgl055.latest + container_name: mvp11-ray-worker-0 + command: sleep infinity + volumes: + - ../verl:/workspace/verl + - ../shared:/private + - .:/workspace/mvp/v1.1 + shm_size: "10g" + ulimits: + nofile: + soft: 65536 + hard: 65536 + cap_add: + - SYS_ADMIN + - SYS_PTRACE + networks: + - mvp11-ray-net + runtime: nvidia + environment: + NVIDIA_VISIBLE_DEVICES: "0,1,2,3" + NVIDIA_DRIVER_CAPABILITIES: "all" + HF_HOME: "/private/hf" + HUGGINGFACE_HUB_CACHE: "/private/hf/hub" + TRANSFORMERS_CACHE: "/private/hf/transformers" + HF_ENDPOINT: "https://hf-mirror.com" + PYTHONUNBUFFERED: "1" + + ray_worker_1: + image: verlai/verl:sgl055.latest + container_name: mvp11-ray-worker-1 + command: sleep infinity + volumes: + - ../verl:/workspace/verl + - ../shared:/private + - .:/workspace/mvp/v1.1 + shm_size: "10g" + ulimits: + nofile: + soft: 65536 + hard: 65536 + cap_add: + - SYS_ADMIN + - SYS_PTRACE + networks: + - mvp11-ray-net + runtime: nvidia + environment: + NVIDIA_VISIBLE_DEVICES: "4,5,6,7" + NVIDIA_DRIVER_CAPABILITIES: "all" + HF_HOME: "/private/hf" + HUGGINGFACE_HUB_CACHE: "/private/hf/hub" + TRANSFORMERS_CACHE: "/private/hf/transformers" + HF_ENDPOINT: "https://hf-mirror.com" + PYTHONUNBUFFERED: "1" + +networks: + mvp11-ray-net: + driver: bridge diff --git a/src/mvp/v1.1/job_spec.schema.json b/src/mvp/v1.1/job_spec.schema.json new file mode 100644 index 0000000..cb5524f --- /dev/null +++ b/src/mvp/v1.1/job_spec.schema.json @@ -0,0 +1,33 @@ +{ + "$schema": "https://json-schema.org/draft/2020-12/schema", + "title": "mvp-v1.1-job-spec", + "type": "object", + "required": ["workload", "shared_root", "code_path", "model_id", "ray", "runtime_env"], + "properties": { + "submission_id": { "type": "string" }, + "workload": { "type": "string", "enum": ["ppo", "grpo", "sft"] }, + "shared_root": { "type": "string" }, + "code_path": { "type": "string" }, + "model_id": { "type": "string" }, + "ppo": { "type": "object" }, + "grpo": { "type": "object" }, + "sft": { "type": "object" }, + "ray": { + "type": "object", + "required": ["address", "entrypoint_num_cpus", "entrypoint_resources"], + "properties": { + "address": { "type": "string" }, + "entrypoint_num_cpus": { "type": "number" }, + "entrypoint_resources": { "type": "object" } + } + }, + "runtime_env": { + "type": "object", + "required": ["env_vars"], + "properties": { + "env_vars": { "type": "object" } + } + } + } +} + diff --git a/src/mvp/v1.1/py/configs/dev.yaml b/src/mvp/v1.1/py/configs/dev.yaml new file mode 100644 index 0000000..2b7c830 --- /dev/null +++ b/src/mvp/v1.1/py/configs/dev.yaml @@ -0,0 +1,20 @@ +# Ray 基础配置（dev 环境 / head 容器内视角） +address: "http://127.0.0.1:8265" + +# 容器内共享根路径（对齐生产 /private） +shared_root: "/private" + +# 强制 driver 落 worker（head 不跑训练） +entrypoint_num_cpus: 1 +entrypoint_resources: + worker_node: 1 + +# 运行时环境变量（所有 job 通用） +runtime_env: + env_vars: + HF_ENDPOINT: "https://hf-mirror.com" + PYTHONUNBUFFERED: "1" + +# 用户自定义代码目录（可被 PYTHONPATH 注入） +user_code_path: "/private/user/code" + diff --git a/src/mvp/v1.1/py/jobspecs/grpo.yaml b/src/mvp/v1.1/py/jobspecs/grpo.yaml new file mode 100644 index 0000000..83eba66 --- /dev/null +++ b/src/mvp/v1.1/py/jobspecs/grpo.yaml @@ -0,0 +1,20 @@ +workload: "grpo" + +submission_id: "" + +code_path: "/private/common/code/verl/verl_repo" + +model_id: "Qwen/Qwen2.5-0.5B-Instruct" + +train_file: "/private/datasets/gsm8k/train.parquet" +val_file: "/private/datasets/gsm8k/test.parquet" + +nnodes: 2 +n_gpus_per_node: 4 + +total_epochs: 1 +total_training_steps: 10 + +save_freq: 10 +test_freq: -1 + diff --git a/src/mvp/v1.1/py/jobspecs/ppo.yaml b/src/mvp/v1.1/py/jobspecs/ppo.yaml new file mode 100644 index 0000000..05bc7f9 --- /dev/null +++ b/src/mvp/v1.1/py/jobspecs/ppo.yaml @@ -0,0 +1,22 @@ +workload: "ppo" + +# 可选：不填则 submitter 自动生成 +submission_id: "" + +# 多版本：指向 code snapshot（由 scripts/31_snapshot_verl_code.sh 生成） +code_path: "/private/common/code/verl/verl_repo" + +model_id: "Qwen/Qwen2.5-0.5B-Instruct" + +train_file: "/private/datasets/gsm8k/train.parquet" +val_file: "/private/datasets/gsm8k/test.parquet" + +nnodes: 2 +n_gpus_per_node: 4 + +total_epochs: 1 +total_training_steps: 10 + +save_freq: 10 +test_freq: -1 + diff --git a/src/mvp/v1.1/py/jobspecs/sft.yaml b/src/mvp/v1.1/py/jobspecs/sft.yaml new file mode 100644 index 0000000..67637b6 --- /dev/null +++ b/src/mvp/v1.1/py/jobspecs/sft.yaml @@ -0,0 +1,22 @@ +workload: "sft" + +submission_id: "" + +code_path: "/private/common/code/verl/verl_repo" + +model_id: "Qwen/Qwen2.5-0.5B-Instruct" + +train_file: "/private/datasets/gsm8k_sft/train.parquet" +val_file: null + +nnodes: 2 +n_gpus_per_node: 4 + +total_epochs: 1 +total_training_steps: 10 + +save_freq: 10 + +# SFT driver 默认不分配 GPU（ray job entrypoint 不指定 entrypoint_num_gpus），因此 driver 侧不要依赖 CUDA +trainer_device: "cpu" + diff --git a/src/mvp/v1.1/py/mvp_v11/__init__.py b/src/mvp/v1.1/py/mvp_v11/__init__.py new file mode 100644 index 0000000..8b13789 --- /dev/null +++ b/src/mvp/v1.1/py/mvp_v11/__init__.py @@ -0,0 +1 @@ + diff --git a/src/mvp/v1.1/py/mvp_v11/builders.py b/src/mvp/v1.1/py/mvp_v11/builders.py new file mode 100644 index 0000000..d4b9786 --- /dev/null +++ b/src/mvp/v1.1/py/mvp_v11/builders.py @@ -0,0 +1,96 @@ +from __future__ import annotations + +from dataclasses import dataclass + +from .models import JobSpec + + +@dataclass(frozen=True) +class BuiltCommand: + argv: list[str] + + +def build_training_argv(spec: JobSpec, submission_id: str, job_dir: str) -> BuiltCommand: + """ + Returns argv for the actual training process (Hydra overrides preserved). + This argv is executed by a lightweight Python driver entrypoint. + """ + if spec.workload in ("ppo", "grpo"): + algo_overrides: list[str] = [] + if spec.workload == "grpo": + algo_overrides.append("algorithm.adv_estimator=grpo") + + test_freq = spec.test_freq if spec.test_freq is not None else -1 + val_file = spec.val_file if spec.val_file is not None else "null" + + argv = [ + "python3", + "-m", + "verl.trainer.main_ppo", + f"data.train_files={spec.train_file}", + f"data.val_files={val_file}", + "data.train_batch_size=256", + "data.max_prompt_length=512", + "data.max_response_length=512", + f"actor_rollout_ref.model.path={spec.model_id}", + "actor_rollout_ref.actor.optim.lr=1e-6", + "actor_rollout_ref.actor.ppo_mini_batch_size=64", + "actor_rollout_ref.actor.ppo_micro_batch_size_per_gpu=4", + "actor_rollout_ref.rollout.name=sglang", + "actor_rollout_ref.rollout.log_prob_micro_batch_size_per_gpu=8", + "actor_rollout_ref.rollout.tensor_model_parallel_size=1", + "actor_rollout_ref.rollout.gpu_memory_utilization=0.4", + "actor_rollout_ref.ref.log_prob_micro_batch_size_per_gpu=4", + "critic.optim.lr=1e-5", + f"critic.model.path={spec.model_id}", + "critic.ppo_micro_batch_size_per_gpu=4", + "algorithm.kl_ctrl.kl_coef=0.001", + *algo_overrides, + "trainer.logger=console", + "trainer.val_before_train=False", + f"trainer.n_gpus_per_node={spec.n_gpus_per_node}", + f"trainer.nnodes={spec.nnodes}", + f"trainer.save_freq={spec.save_freq}", + f"trainer.test_freq={test_freq}", + f"trainer.total_epochs={spec.total_epochs}", + f"trainer.total_training_steps={spec.total_training_steps}", + "trainer.resume_mode=disable", + f"trainer.default_local_dir={job_dir}/checkpoints", + "+ray_kwargs.ray_init.address=auto", + f"hydra.run.dir={job_dir}/logs/hydra", + ] + return BuiltCommand(argv=argv) + + if spec.workload == "sft": + val_override = "null" if spec.val_file is None else spec.val_file + trainer_device = spec.trainer_device or "cpu" + + argv = [ + "python3", + "-m", + "verl.trainer.sft_trainer_ray", + f"model.path={spec.model_id}", + f"data.train_files={spec.train_file}", + f"data.val_files={val_override}", + "data.train_batch_size=64", + "data.micro_batch_size_per_gpu=1", + "data.max_token_len_per_gpu=2048", + "data.max_length=1024", + "trainer.logger=console", + "trainer.project_name=mvp11-sft", + f"trainer.experiment_name={submission_id}", + f"trainer.total_epochs={spec.total_epochs}", + f"trainer.total_training_steps={spec.total_training_steps}", + f"trainer.save_freq={spec.save_freq}", + "trainer.test_freq=-1", + "trainer.resume_mode=disable", + f"trainer.device={trainer_device}", + f"trainer.default_local_dir={job_dir}/checkpoints", + f"trainer.nnodes={spec.nnodes}", + f"trainer.n_gpus_per_node={spec.n_gpus_per_node}", + f"hydra.run.dir={job_dir}/logs/hydra", + ] + return BuiltCommand(argv=argv) + + raise ValueError(f"unsupported workload: {spec.workload}") + diff --git a/src/mvp/v1.1/py/mvp_v11/driver_entrypoint.py b/src/mvp/v1.1/py/mvp_v11/driver_entrypoint.py new file mode 100644 index 0000000..9f99c58 --- /dev/null +++ b/src/mvp/v1.1/py/mvp_v11/driver_entrypoint.py @@ -0,0 +1,63 @@ +from __future__ import annotations + +import argparse +import os +import shlex +import subprocess +import sys +from pathlib import Path + + +def _preflight() -> None: + print("MVP_PRECHECK_PYTHON:", sys.executable, flush=True) + print("MVP_PRECHECK_PYTHONPATH:", os.environ.get("PYTHONPATH"), flush=True) + print("MVP_PRECHECK_MVP_CODE_PATH:", os.environ.get("MVP_CODE_PATH"), flush=True) + try: + import verl # type: ignore + + print("MVP_PRECHECK_VERL_FILE:", getattr(verl, "__file__", None), flush=True) + except Exception as e: + print("MVP_PRECHECK_VERL_IMPORT_ERROR:", repr(e), flush=True) + + try: + import mvp_marker # type: ignore + + print("MVP_PRECHECK_MARKER:", getattr(mvp_marker, "MARKER", None), flush=True) + except Exception as e: + print("MVP_PRECHECK_MARKER_MISSING:", repr(e), flush=True) + + +def main() -> int: + parser = argparse.ArgumentParser() + parser.add_argument("--job-dir", required=True) + parser.add_argument("cmd", nargs=argparse.REMAINDER) + args = parser.parse_args() + + job_dir = Path(args.job_dir) + job_dir.mkdir(parents=True, exist_ok=True) + + _preflight() + + if not args.cmd: + print("no command provided", file=sys.stderr) + return 2 + + # argparse includes the leading "--" if the caller uses it; strip it. + cmd = list(args.cmd) + if cmd and cmd[0] == "--": + cmd = cmd[1:] + if not cmd: + print("no command provided", file=sys.stderr) + return 2 + + # Execute training command as a subprocess so that logs are captured by Ray job logs. + cmd_str = " ".join(shlex.quote(x) for x in cmd) + print("MVP_DRIVER_EXEC:", cmd_str, flush=True) + + proc = subprocess.run(cmd, check=False) + print("MVP_DRIVER_EXIT_CODE:", proc.returncode, flush=True) + return proc.returncode + + +if __name__ == "__main__": + raise SystemExit(main()) diff --git a/src/mvp/v1.1/py/mvp_v11/models.py b/src/mvp/v1.1/py/mvp_v11/models.py new file mode 100644 index 0000000..41f8804 --- /dev/null +++ b/src/mvp/v1.1/py/mvp_v11/models.py @@ -0,0 +1,121 @@ +from __future__ import annotations + +from dataclasses import dataclass +from typing import Any + + +def _require(d: dict[str, Any], key: str) -> Any: + if key not in d or d[key] in (None, ""): + raise ValueError(f"missing required field: {key}") + return d[key] + + +@dataclass(frozen=True) +class RayConfig: + address: str + shared_root: str + entrypoint_num_cpus: float + entrypoint_resources: dict[str, float] + runtime_env_env_vars: dict[str, str] + user_code_path: str + + @staticmethod + def from_dict(d: dict[str, Any]) -> "RayConfig": + runtime_env = d.get("runtime_env") or {} + env_vars = (runtime_env.get("env_vars") or {}) if isinstance(runtime_env, dict) else {} + if not isinstance(env_vars, dict): + raise ValueError("runtime_env.env_vars must be a mapping") + + entrypoint_resources = d.get("entrypoint_resources") or {} + if not isinstance(entrypoint_resources, dict): + raise ValueError("entrypoint_resources must be a mapping") + + return RayConfig( + address=str(_require(d, "address")), + shared_root=str(_require(d, "shared_root")), + entrypoint_num_cpus=float(d.get("entrypoint_num_cpus", 1)), + entrypoint_resources={str(k): float(v) for k, v in entrypoint_resources.items()}, + runtime_env_env_vars={str(k): str(v) for k, v in env_vars.items()}, + user_code_path=str(d.get("user_code_path", f"{_require(d, 'shared_root')}/user/code")), + ) + + def to_public_dict(self) -> dict[str, Any]: + return { + "address": self.address, + "shared_root": self.shared_root, + "entrypoint_num_cpus": self.entrypoint_num_cpus, + "entrypoint_resources": self.entrypoint_resources, + "runtime_env": {"env_vars": self.runtime_env_env_vars}, + "user_code_path": self.user_code_path, + } + + +@dataclass(frozen=True) +class JobSpec: + workload: str # ppo|grpo|sft + submission_id: str | None + code_path: str + model_id: str + + train_file: str + val_file: str | None + + nnodes: int + n_gpus_per_node: int + + total_epochs: int + total_training_steps: int + + save_freq: int + test_freq: int | None + + trainer_device: str | None # only for sft (driver-side device) + + @staticmethod + def from_dict(d: dict[str, Any]) -> "JobSpec": + workload = str(_require(d, "workload")) + if workload not in ("ppo", "grpo", "sft"): + raise ValueError(f"unsupported workload: {workload}") + + val_file = d.get("val_file", None) + if val_file in ("", "null"): + val_file = None + + test_freq = d.get("test_freq", None) + if test_freq in ("", "null"): + test_freq = None + + return JobSpec( + workload=workload, + submission_id=(str(d["submission_id"]) if d.get("submission_id") else None), + code_path=str(_require(d, "code_path")), + model_id=str(_require(d, "model_id")), + train_file=str(_require(d, "train_file")), + val_file=(str(val_file) if val_file is not None else None), + nnodes=int(d.get("nnodes", 2)), + n_gpus_per_node=int(d.get("n_gpus_per_node", 4)), + total_epochs=int(d.get("total_epochs", 1)), + total_training_steps=int(d.get("total_training_steps", 10)), + save_freq=int(d.get("save_freq", 10)), + test_freq=(int(test_freq) if test_freq is not None else None), + trainer_device=(str(d.get("trainer_device")) if d.get("trainer_device") else None), + ) + + def to_public_dict(self) -> dict[str, Any]: + out: dict[str, Any] = { + "workload": self.workload, + "submission_id": self.submission_id or "", + "code_path": self.code_path, + "model_id": self.model_id, + "train_file": self.train_file, + "val_file": self.val_file, + "nnodes": self.nnodes, + "n_gpus_per_node": self.n_gpus_per_node, + "total_epochs": self.total_epochs, + "total_training_steps": self.total_training_steps, + "save_freq": self.save_freq, + "test_freq": self.test_freq, + } + if self.workload == "sft": + out["trainer_device"] = self.trainer_device or "cpu" + return out diff --git a/src/mvp/v1.1/py/mvp_v11/ray_job_tool.py b/src/mvp/v1.1/py/mvp_v11/ray_job_tool.py new file mode 100644 index 0000000..01b899a --- /dev/null +++ b/src/mvp/v1.1/py/mvp_v11/ray_job_tool.py @@ -0,0 +1,171 @@ +from __future__ import annotations + +import json +import os +import shlex +from datetime import datetime +from pathlib import Path +from typing import Any + +import ray +from ray.job_submission import JobSubmissionClient + +from .builders import build_training_argv +from .models import JobSpec, RayConfig +from .yaml_io import dump_yaml + + +def _ts() -> str: + return datetime.now().strftime("%Y%m%d_%H%M%S") + + +def _mkdir(p: Path) -> None: + p.mkdir(parents=True, exist_ok=True) + + +def _write_text(p: Path, content: str) -> None: + _mkdir(p.parent) + p.write_text(content, encoding="utf-8") + + +def _write_json(p: Path, obj: Any) -> None: + _write_text(p, json.dumps(obj, indent=2, ensure_ascii=False) + "\n") + + +def _safe_basename(path: str) -> str: + return path.rstrip("/").split("/")[-1] + + +class RayJobTool: + def __init__(self, cfg: RayConfig): + self.cfg = cfg + self.client = JobSubmissionClient(cfg.address) + + def _job_dir(self, submission_id: str) -> str: + return f"{self.cfg.shared_root}/jobs/{submission_id}" + + def _runtime_env(self, spec: JobSpec) -> dict[str, Any]: + env_vars = dict(self.cfg.runtime_env_env_vars) + + # Default HF cache + env_vars.setdefault("HF_HOME", f"{self.cfg.shared_root}/hf") + env_vars.setdefault("HUGGINGFACE_HUB_CACHE", f"{self.cfg.shared_root}/hf/hub") + env_vars.setdefault("TRANSFORMERS_CACHE", f"{self.cfg.shared_root}/hf/transformers") + env_vars.setdefault("PYTHONUNBUFFERED", "1") + + # Tool code path must be importable on workers (compose mounts v1.1 into all containers). + # Place it before verl code to avoid interfering with verl import priority. + tool_code_path = os.environ.get("MVP_TOOL_CODE_PATH", "/workspace/mvp/v1.1/py") + + user_code_path = self.cfg.user_code_path + code_path = spec.code_path + + existing = env_vars.get("PYTHONPATH", "") + prefix = f"{tool_code_path}:{code_path}:{user_code_path}" + env_vars["PYTHONPATH"] = f"{prefix}:{existing}" if existing else prefix + + # For debugging / log visibility + env_vars["MVP_CODE_PATH"] = code_path + + # SFT: ensure ray.init() connects to the cluster + if spec.workload == "sft": + env_vars.setdefault("RAY_ADDRESS", "auto") + + return {"env_vars": env_vars} + + def submit(self, spec: JobSpec, no_wait: bool) -> str: + submission_id = spec.submission_id or f"mvp11_{spec.workload}_{_ts()}_{os.getpid()}" + job_dir = self._job_dir(submission_id) + + built = build_training_argv(spec, submission_id=submission_id, job_dir=job_dir) + entrypoint_argv = [ + "python3", + "-m", + "mvp_v11.driver_entrypoint", + "--job-dir", + job_dir, + *built.argv, + ] + entrypoint = " ".join(shlex.quote(x) for x in entrypoint_argv) + + runtime_env = self._runtime_env(spec) + + # Prepare job artifacts directory + job_root = Path(job_dir) + _mkdir(job_root / "config") + _mkdir(job_root / "logs") + _mkdir(job_root / "debug") + _mkdir(job_root / "checkpoints") + + _write_text(job_root / "config" / "ray_config.yaml", dump_yaml(self.cfg.to_public_dict())) + _write_text(job_root / "config" / "jobspec.yaml", dump_yaml(spec.to_public_dict())) + _write_json(job_root / "config" / "submit_payload.json", { + "submission_id": submission_id, + "address": self.cfg.address, + "entrypoint": entrypoint, + "entrypoint_num_cpus": self.cfg.entrypoint_num_cpus, + "entrypoint_resources": self.cfg.entrypoint_resources, + "runtime_env": runtime_env, + }) + + # Pre-submit debug snapshot (ray cluster resources via ray.init) + try: + ray.init(address="auto", ignore_reinit_error=True, log_to_driver=False) + _write_json(job_root / "debug" / "ray_cluster_resources_pre.json", ray.cluster_resources()) + _write_json(job_root / "debug" / "ray_available_resources_pre.json", ray.available_resources()) + except Exception as e: + _write_text(job_root / "debug" / "ray_resources_pre.error.txt", repr(e) + "\n") + + try: + submitted = self.client.submit_job( + entrypoint=entrypoint, + submission_id=submission_id, + runtime_env=runtime_env, + entrypoint_num_cpus=self.cfg.entrypoint_num_cpus, + entrypoint_resources=self.cfg.entrypoint_resources, + ) + except Exception as e: + _write_text(job_root / "logs" / "submit.error.txt", repr(e) + "\n") + raise + + _write_text(job_root / "config" / "ray_submission_id.txt", submitted + "\n") + + # Post-submit debug snapshot via SDK + try: + jobs = self.client.list_jobs() + _write_text( + job_root / "debug" / "ray_job_list_post.json", + json.dumps([_job_details_to_dict(j) for j in jobs], indent=2) + "\n", + ) + except Exception as e: + _write_text(job_root / "debug" / "ray_job_list_post.error.txt", repr(e) + "\n") + + if not no_wait: + # caller can separately wait; keep submit non-blocking by default in scripts + pass + + return submitted + + def status(self, submission_id: str) -> str: + return str(self.client.get_job_status(submission_id)) + + def stop(self, submission_id: str) -> bool: + return bool(self.client.stop_job(submission_id)) + + def logs(self, submission_id: str) -> str: + return self.client.get_job_logs(submission_id) + + def list(self) -> list[dict[str, Any]]: + return [_job_details_to_dict(j) for j in self.client.list_jobs()] + + +def _job_details_to_dict(obj: Any) -> dict[str, Any]: + # Ray uses pydantic models internally, but depending on bundled pydantic version + # we might get `.model_dump()` (v2) or `.dict()` (v1). + if hasattr(obj, "model_dump"): + return obj.model_dump() # type: ignore[no-any-return] + if hasattr(obj, "dict"): + return obj.dict() # type: ignore[no-any-return] + if hasattr(obj, "__dict__"): + return dict(obj.__dict__) + return {"repr": repr(obj)} diff --git a/src/mvp/v1.1/py/mvp_v11/yaml_io.py b/src/mvp/v1.1/py/mvp_v11/yaml_io.py new file mode 100644 index 0000000..c321688 --- /dev/null +++ b/src/mvp/v1.1/py/mvp_v11/yaml_io.py @@ -0,0 +1,21 @@ +from __future__ import annotations + +from pathlib import Path +from typing import Any + +import yaml + + +def load_yaml(path: str) -> dict[str, Any]: + p = Path(path) + data = yaml.safe_load(p.read_text(encoding="utf-8")) + if data is None: + return {} + if not isinstance(data, dict): + raise ValueError(f"yaml root must be a mapping: {path}") + return data + + +def dump_yaml(data: dict[str, Any]) -> str: + return yaml.safe_dump(data, sort_keys=False, allow_unicode=True) + diff --git a/src/mvp/v1.1/py/requirements.txt b/src/mvp/v1.1/py/requirements.txt new file mode 100644 index 0000000..e3af026 --- /dev/null +++ b/src/mvp/v1.1/py/requirements.txt @@ -0,0 +1,2 @@ +PyYAML>=6.0.1 + diff --git a/src/mvp/v1.1/py/run.py b/src/mvp/v1.1/py/run.py new file mode 100644 index 0000000..67fe975 --- /dev/null +++ b/src/mvp/v1.1/py/run.py @@ -0,0 +1,69 @@ +#!/usr/bin/env python3 +from __future__ import annotations + +import argparse +import json +import os +import sys + + +def _ensure_import_path() -> None: + # Allow `python3 /workspace/.../py/run.py` to import `mvp_v11.*` + here = os.path.dirname(os.path.abspath(__file__)) + if here not in sys.path: + sys.path.insert(0, here) + + +def main() -> int: + _ensure_import_path() + + from mvp_v11.models import JobSpec, RayConfig + from mvp_v11.ray_job_tool import RayJobTool + from mvp_v11.yaml_io import load_yaml + + parser = argparse.ArgumentParser() + parser.add_argument("--config", required=True, help="Ray base config yaml") + parser.add_argument("--jobspec", help="Training jobspec yaml (required for submit)") + parser.add_argument("--action", required=True, choices=["submit", "status", "stop", "logs", "list"]) + parser.add_argument("--submission-id", help="For status/stop/logs") + parser.add_argument("--no-wait", action="store_true", help="Submit and return immediately") + parser.add_argument("--tail", type=int, default=0, help="Tail N lines for logs") + args = parser.parse_args() + + cfg = RayConfig.from_dict(load_yaml(args.config)) + tool = RayJobTool(cfg) + + if args.action == "submit": + if not args.jobspec: + raise SystemExit("--jobspec is required for submit") + spec = JobSpec.from_dict(load_yaml(args.jobspec)) + submitted = tool.submit(spec, no_wait=args.no_wait) + print(submitted) + return 0 + + if args.action in ("status", "stop", "logs"): + sid = args.submission_id or "" + if not sid: + raise SystemExit("--submission-id is required for status/stop/logs") + if args.action == "status": + print(tool.status(sid)) + return 0 + if args.action == "stop": + print(tool.stop(sid)) + return 0 + logs = tool.logs(sid) + if args.tail and args.tail > 0: + lines = logs.splitlines() + logs = "\n".join(lines[-args.tail :]) + ("\n" if lines else "") + print(logs, end="") + return 0 + + if args.action == "list": + print(json.dumps(tool.list(), indent=2)) + return 0 + + raise SystemExit(f"unknown action: {args.action}") + + +if __name__ == "__main__": + raise SystemExit(main()) diff --git a/src/mvp/v1.1/py/sitecustomize.py b/src/mvp/v1.1/py/sitecustomize.py new file mode 100644 index 0000000..478b284 --- /dev/null +++ b/src/mvp/v1.1/py/sitecustomize.py @@ -0,0 +1,57 @@ +""" +Job-scoped compatibility shims loaded automatically by Python at startup. + +This is intentionally lightweight and safe-by-default: +- Only patches missing symbols. +- Never raises (best-effort). + +Primary use case in MVP v1.1: +- Allow multiple `verl` versions (e.g. v0.6.0 vs v0.6.1) to run on the same + base image where `sglang` APIs may differ slightly. +""" + +from __future__ import annotations + + +def _patch_sglang_get_ip() -> None: + try: + import sglang.srt.utils as srt_utils # type: ignore + except Exception: + return + + if hasattr(srt_utils, "get_ip"): + return + + def get_ip() -> str: + # Best-effort local IP without external dependency. + try: + import socket + + s = socket.socket(socket.AF_INET, socket.SOCK_DGRAM) + try: + # Doesn't send packets; used to pick the default route/interface. + s.connect(("8.8.8.8", 80)) + return str(s.getsockname()[0]) + finally: + s.close() + except Exception: + # Fallback: hostname resolution + try: + import socket + + return str(socket.gethostbyname(socket.gethostname())) + except Exception: + return "127.0.0.1" + + try: + setattr(srt_utils, "get_ip", get_ip) + except Exception: + return + + +try: + _patch_sglang_get_ip() +except Exception: + # Never block interpreter startup. + pass + diff --git a/src/mvp/v1.1/scripts/00_prereq_check.sh b/src/mvp/v1.1/scripts/00_prereq_check.sh new file mode 100644 index 0000000..e6479f8 --- /dev/null +++ b/src/mvp/v1.1/scripts/00_prereq_check.sh @@ -0,0 +1,42 @@ +#!/usr/bin/env bash +set -euo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +# shellcheck source=lib.sh +source "${SCRIPT_DIR}/lib.sh" + +echo "[host] prereq check" + +require_cmd docker +require_cmd bash + +if ! docker info >/dev/null 2>&1; then + echo "docker is not available (docker info failed)" >&2 + exit 1 +fi + +if ! docker compose version >/dev/null 2>&1; then + echo "docker compose is not available" >&2 + exit 1 +fi + +if ! command -v nvidia-smi >/dev/null 2>&1; then + echo "WARN: nvidia-smi not found on host; GPU validation skipped" +else + echo "[host] GPU summary" + nvidia-smi -L || true +fi + +echo "[host] ensure shared dirs exist under ../shared" +mkdir -p "${ROOT_DIR}/../shared"/{datasets,hf,jobs,outputs,ray,common,user} +mkdir -p "${ROOT_DIR}/../shared/common"/{code,datasets,models} +mkdir -p "${ROOT_DIR}/../shared/user"/{code} + +echo "[host] ensure verl repo exists under ../verl (required by prepare scripts)" +if [[ ! -d "${ROOT_DIR}/../verl" ]]; then + echo "missing ../verl. On remote, ensure /home2/argus/infra/mvp/verl exists (git clone volcengine/verl)." >&2 + exit 1 +fi + +echo "ok" + diff --git a/src/mvp/v1.1/scripts/01_up.sh b/src/mvp/v1.1/scripts/01_up.sh new file mode 100644 index 0000000..5dffc6c --- /dev/null +++ b/src/mvp/v1.1/scripts/01_up.sh @@ -0,0 +1,16 @@ +#!/usr/bin/env bash +set -euo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +# shellcheck source=lib.sh +source "${SCRIPT_DIR}/lib.sh" + +if [[ "${SKIP_CLEANUP_V1:-0}" != "1" ]]; then + "${SCRIPT_DIR}/03_cleanup_v1_legacy.sh" || true +fi + +echo "[host] docker compose up -d (v1.1)" +dc up -d + +echo "[host] containers:" +docker ps --format 'table {{.Names}}\t{{.Status}}\t{{.Ports}}' | (head -n 1 && grep -E "mvp11-ray-") || true diff --git a/src/mvp/v1.1/scripts/02_down.sh b/src/mvp/v1.1/scripts/02_down.sh new file mode 100644 index 0000000..a0d6eb0 --- /dev/null +++ b/src/mvp/v1.1/scripts/02_down.sh @@ -0,0 +1,12 @@ +#!/usr/bin/env bash +set -euo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +# shellcheck source=lib.sh +source "${SCRIPT_DIR}/lib.sh" + +echo "[host] docker compose down (v1.1)" +dc down -v || true + +echo "[host] done" + diff --git a/src/mvp/v1.1/scripts/03_cleanup_v1_legacy.sh b/src/mvp/v1.1/scripts/03_cleanup_v1_legacy.sh new file mode 100644 index 0000000..0f28585 --- /dev/null +++ b/src/mvp/v1.1/scripts/03_cleanup_v1_legacy.sh @@ -0,0 +1,16 @@ +#!/usr/bin/env bash +set -euo pipefail + +echo "[host] cleanup v1 legacy containers (best-effort)" + +LEGACY=(mvp-ray-head mvp-ray-worker-0 mvp-ray-worker-1) + +for c in "${LEGACY[@]}"; do + if docker ps -a --format '{{.Names}}' | grep -qx "${c}"; then + echo "[host] removing legacy container: ${c}" + docker rm -f "${c}" >/dev/null 2>&1 || true + fi +done + +echo "[host] legacy cleanup done" + diff --git a/src/mvp/v1.1/scripts/05_ensure_verl_repo.sh b/src/mvp/v1.1/scripts/05_ensure_verl_repo.sh new file mode 100644 index 0000000..3a6da0d --- /dev/null +++ b/src/mvp/v1.1/scripts/05_ensure_verl_repo.sh @@ -0,0 +1,23 @@ +#!/usr/bin/env bash +set -euo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +# shellcheck source=lib.sh +source "${SCRIPT_DIR}/lib.sh" + +VERL_DIR="${ROOT_DIR}/../verl" + +echo "[host] ensure verl repo exists at: ${VERL_DIR}" +if [[ -d "${VERL_DIR}/.git" ]]; then + echo "verl_repo_exists: skip" + exit 0 +fi + +if [[ -d "${VERL_DIR}" && ! -d "${VERL_DIR}/.git" ]]; then + echo "ERROR: ${VERL_DIR} exists but is not a git repo; please fix manually." >&2 + exit 1 +fi + +echo "cloning volcengine/verl -> ${VERL_DIR}" +git clone https://github.com/volcengine/verl.git "${VERL_DIR}" + diff --git a/src/mvp/v1.1/scripts/12_install_py_deps.sh b/src/mvp/v1.1/scripts/12_install_py_deps.sh new file mode 100644 index 0000000..da1649c --- /dev/null +++ b/src/mvp/v1.1/scripts/12_install_py_deps.sh @@ -0,0 +1,10 @@ +#!/usr/bin/env bash +set -euo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +# shellcheck source=lib.sh +source "${SCRIPT_DIR}/lib.sh" + +echo "[head] install python deps for v1.1 SDK submitter (PyYAML)" +dexec "${HEAD_CONTAINER}" bash -lc "pip install -r /workspace/mvp/v1.1/py/requirements.txt" + diff --git a/src/mvp/v1.1/scripts/20_start_head.sh b/src/mvp/v1.1/scripts/20_start_head.sh new file mode 100644 index 0000000..fb3b7b7 --- /dev/null +++ b/src/mvp/v1.1/scripts/20_start_head.sh @@ -0,0 +1,18 @@ +#!/usr/bin/env bash +set -euo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +# shellcheck source=lib.sh +source "${SCRIPT_DIR}/lib.sh" + +HEAD_IP="$(container_ip "${HEAD_CONTAINER}")" + +echo "[head] ray stop (best-effort)" +dexec "${HEAD_CONTAINER}" bash -lc "ray stop --force || true" + +echo "[head] start ray head (CPU=0 GPU=0): ${HEAD_IP}" +dexec "${HEAD_CONTAINER}" bash -lc "ray start --head --node-ip-address='${HEAD_IP}' --port=6379 --dashboard-host=0.0.0.0 --dashboard-port=8265 --num-cpus=0 --num-gpus=0 --disable-usage-stats" + +echo "[head] ray status" +dexec "${HEAD_CONTAINER}" bash -lc "ray status || true" + diff --git a/src/mvp/v1.1/scripts/21_start_workers.sh b/src/mvp/v1.1/scripts/21_start_workers.sh new file mode 100644 index 0000000..2f55a95 --- /dev/null +++ b/src/mvp/v1.1/scripts/21_start_workers.sh @@ -0,0 +1,26 @@ +#!/usr/bin/env bash +set -euo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +# shellcheck source=lib.sh +source "${SCRIPT_DIR}/lib.sh" + +HEAD_IP="$(container_ip "${HEAD_CONTAINER}")" +HEAD_ADDR="${HEAD_IP}:6379" + +start_one() { + local worker="$1" + local ip + ip="$(container_ip "${worker}")" + echo "[${worker}] ray stop (best-effort)" + dexec "${worker}" bash -lc "ray stop --force || true" + echo "[${worker}] start ray worker -> head ${HEAD_ADDR}" + dexec "${worker}" bash -lc "ray start --address='${HEAD_ADDR}' --node-ip-address='${ip}' --resources='{\"worker_node\": 100}' --disable-usage-stats" +} + +start_one "${WORKER0_CONTAINER}" +start_one "${WORKER1_CONTAINER}" + +echo "[head] ray status" +dexec "${HEAD_CONTAINER}" bash -lc "ray status || true" + diff --git a/src/mvp/v1.1/scripts/30_prepare_data_and_model.sh b/src/mvp/v1.1/scripts/30_prepare_data_and_model.sh new file mode 100644 index 0000000..157e743 --- /dev/null +++ b/src/mvp/v1.1/scripts/30_prepare_data_and_model.sh @@ -0,0 +1,86 @@ +#!/usr/bin/env bash +set -euo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +# shellcheck source=lib.sh +source "${SCRIPT_DIR}/lib.sh" + +MODEL_ID="${MODEL_ID:-Qwen/Qwen2.5-0.5B-Instruct}" + +PPO_DATA_DIR="${SHARED_ROOT}/datasets/gsm8k" +SFT_DATA_DIR="${SHARED_ROOT}/datasets/gsm8k_sft" + +CODE_SNAPSHOT_DIR="${SHARED_ROOT}/common/code/verl/verl_repo" + +echo "[head] ensure dataset dirs exist" +dexec "${HEAD_CONTAINER}" bash -lc "mkdir -p '${PPO_DATA_DIR}' '${SFT_DATA_DIR}'" + +echo "[head] prepare PPO dataset (gsm8k RL parquet) -> ${PPO_DATA_DIR}" +dexec "${HEAD_CONTAINER}" bash -lc "if [[ -f '${PPO_DATA_DIR}/train.parquet' && -f '${PPO_DATA_DIR}/test.parquet' ]]; then echo 'ppo_dataset_exists: skip'; else python3 /workspace/verl/examples/data_preprocess/gsm8k.py --local_save_dir '${PPO_DATA_DIR}'; fi" + +echo "[head] prepare SFT dataset (gsm8k messages parquet) -> ${SFT_DATA_DIR}" +if dexec "${HEAD_CONTAINER}" bash -lc "test -f '${SFT_DATA_DIR}/train.parquet'"; then + echo "[head] sft_dataset_exists: skip" +else + SFT_PY_CODE="$(cat <<'PY' +import os + +import pandas as pd +from datasets import load_dataset + +out_dir = os.environ["SFT_DATA_DIR"] +os.makedirs(out_dir, exist_ok=True) + +ds = load_dataset("openai/gsm8k", "main") + +instruction = "Let's think step by step and output the final answer after \"####\"." + +def to_messages(example): + q = example["question"].strip() + " " + instruction + a = example["answer"] + return { + "messages": [ + {"role": "user", "content": q}, + {"role": "assistant", "content": a}, + ] + } + +train = ds["train"].map(to_messages, remove_columns=ds["train"].column_names) +test = ds["test"].map(to_messages, remove_columns=ds["test"].column_names) + +pd.DataFrame(train).to_parquet(os.path.join(out_dir, "train.parquet"), index=False) +pd.DataFrame(test).to_parquet(os.path.join(out_dir, "test.parquet"), index=False) + +print("sft_dataset_written_ok:", out_dir) +PY +)" + printf "%s\n" "${SFT_PY_CODE}" | dexec "${HEAD_CONTAINER}" bash -lc "SFT_DATA_DIR='${SFT_DATA_DIR}' python3 -" +fi + +echo "[head] ensure model cached to persistent HF_HOME (idempotent) -> ${MODEL_ID}" +PY_CODE="$(cat <<'PY' +import os + +model_id = os.environ["MODEL_ID"] + +hf_home = os.environ.get("HF_HOME", "/private/hf") +os.environ.setdefault("HF_HOME", hf_home) +os.environ.setdefault("HUGGINGFACE_HUB_CACHE", os.path.join(hf_home, "hub")) +os.environ.setdefault("TRANSFORMERS_CACHE", os.path.join(hf_home, "transformers")) + +from huggingface_hub import snapshot_download + +try: + snapshot_download(repo_id=model_id, local_files_only=True) + print("model_cache_exists: skip", model_id) +except Exception: + print("model_cache_missing: downloading", model_id) + snapshot_download(repo_id=model_id) + print("model_cached_ok:", model_id) +PY +)" + +printf "%s\n" "${PY_CODE}" | dexec "${HEAD_CONTAINER}" bash -lc "MODEL_ID='${MODEL_ID}' python3 -" + +echo "[head] snapshot verl repo into shared common code path (idempotent best-effort) -> ${CODE_SNAPSHOT_DIR}" +dexec "${HEAD_CONTAINER}" bash -lc "mkdir -p '${CODE_SNAPSHOT_DIR}' && if command -v rsync >/dev/null 2>&1; then rsync -a --delete /workspace/verl/ '${CODE_SNAPSHOT_DIR}/'; else rm -rf '${CODE_SNAPSHOT_DIR:?}/'* && cp -a /workspace/verl/. '${CODE_SNAPSHOT_DIR}/'; fi && echo 'code_snapshot_ok'" diff --git a/src/mvp/v1.1/scripts/31_snapshot_verl_code.sh b/src/mvp/v1.1/scripts/31_snapshot_verl_code.sh new file mode 100644 index 0000000..0399e3e --- /dev/null +++ b/src/mvp/v1.1/scripts/31_snapshot_verl_code.sh @@ -0,0 +1,42 @@ +#!/usr/bin/env bash +set -euo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +# shellcheck source=lib.sh +source "${SCRIPT_DIR}/lib.sh" + +# Create an immutable-ish code snapshot under: +# ${SHARED_ROOT}/common/code/verl/ +# +# By default, code_id is the git commit hash of /workspace/verl (mounted from ../verl). +# +# This enables job-level multi-version coexistence via runtime_env PYTHONPATH injection. + +CODE_ID="${CODE_ID:-}" + +if [[ -z "${CODE_ID}" ]]; then + CODE_ID="$(dexec "${HEAD_CONTAINER}" bash -lc "git config --global --add safe.directory /workspace/verl >/dev/null 2>&1 || true; git -C /workspace/verl rev-parse HEAD")" +fi + +DEST_DIR="${SHARED_ROOT}/common/code/verl/${CODE_ID}" + +echo "[head] snapshot verl repo -> ${DEST_DIR}" + +if dexec "${HEAD_CONTAINER}" bash -lc "test -d '${DEST_DIR}'"; then + echo "[head] code_snapshot_exists: skip" + exit 0 +fi + +dexec "${HEAD_CONTAINER}" bash -lc "mkdir -p '${DEST_DIR}'" + +# Copy code (no .git needed for runtime) +if dexec "${HEAD_CONTAINER}" bash -lc "command -v rsync >/dev/null 2>&1"; then + dexec "${HEAD_CONTAINER}" bash -lc "rsync -a --delete --exclude='.git' /workspace/verl/ '${DEST_DIR}/'" +else + dexec "${HEAD_CONTAINER}" bash -lc "tar -C /workspace/verl -cf - --exclude='.git' . | tar -C '${DEST_DIR}' -xf -" +fi + +# Add a tiny marker module for multi-version validation in Ray job logs. +dexec "${HEAD_CONTAINER}" bash -lc "printf \"%s\\n\" \"MARKER = '${CODE_ID}'\" > '${DEST_DIR}/mvp_marker.py'" + +echo "[head] code_snapshot_ok: ${CODE_ID}" diff --git a/src/mvp/v1.1/scripts/32_clone_verl_tags.sh b/src/mvp/v1.1/scripts/32_clone_verl_tags.sh new file mode 100644 index 0000000..dfcc470 --- /dev/null +++ b/src/mvp/v1.1/scripts/32_clone_verl_tags.sh @@ -0,0 +1,39 @@ +#!/usr/bin/env bash +set -euo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +# shellcheck source=lib.sh +source "${SCRIPT_DIR}/lib.sh" + +VERL_REPO_URL="${VERL_REPO_URL:-https://github.com/volcengine/verl.git}" +DEST_BASE="${SHARED_ROOT}/common/code/verl" + +TAGS=("v0.6.0" "v0.6.1") + +echo "[head] ensure base dir: ${DEST_BASE}" +dexec "${HEAD_CONTAINER}" bash -lc "mkdir -p '${DEST_BASE}'" + +for tag in "${TAGS[@]}"; do + dest="${DEST_BASE}/verl_${tag}" + echo "[head] prepare verl tag ${tag} -> ${dest}" + + verify_repo_cmd="test -d '${dest}/.git' && git -C '${dest}' rev-parse --is-inside-work-tree >/dev/null 2>&1" + if dexec "${HEAD_CONTAINER}" bash -lc "${verify_repo_cmd}"; then + echo "[head] exists: verified git repo: ${dest}" + else + echo "[head] cloning ${tag} (retry with HTTP/1.1 if needed)" + dexec "${HEAD_CONTAINER}" bash -lc "rm -rf '${dest}'" + # Retry a few times because GitHub over HTTP/2 can occasionally fail with curl framing errors/timeouts. + dexec "${HEAD_CONTAINER}" bash -lc "set -euo pipefail; for i in 1 2 3; do echo \"clone_attempt=\\$i\"; if git -c http.version=HTTP/1.1 clone --filter=blob:none --single-branch --branch '${tag}' --depth 1 '${VERL_REPO_URL}' '${dest}'; then exit 0; fi; rm -rf '${dest}'; sleep 3; done; exit 1" + dexec "${HEAD_CONTAINER}" bash -lc "${verify_repo_cmd}" || { echo \"[head] clone failed or repo invalid: ${dest}\" >&2; exit 1; } + fi + + # Avoid git safe.directory issues when reading repo state + dexec "${HEAD_CONTAINER}" bash -lc "git config --global --add safe.directory '${dest}' >/dev/null 2>&1 || true" + dexec "${HEAD_CONTAINER}" bash -lc "printf 'tag='; git -C '${dest}' describe --tags --exact-match 2>/dev/null || true; printf '\\nhead='; git -C '${dest}' rev-parse HEAD; printf '\\n'" + + # Add marker for multi-version verification in Ray job logs + dexec "${HEAD_CONTAINER}" bash -lc "printf \"%s\\n\" \"MARKER = '${tag}'\" > '${dest}/mvp_marker.py'" +done + +echo "[head] done" diff --git a/src/mvp/v1.1/scripts/40_submit_ppo_epoch1.sh b/src/mvp/v1.1/scripts/40_submit_ppo_epoch1.sh new file mode 100644 index 0000000..ac11ab9 --- /dev/null +++ b/src/mvp/v1.1/scripts/40_submit_ppo_epoch1.sh @@ -0,0 +1,72 @@ +#!/usr/bin/env bash +set -euo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +# shellcheck source=lib.sh +source "${SCRIPT_DIR}/lib.sh" + +SUBMISSION_ID="${SUBMISSION_ID:-mvp11_ppo_$(timestamp)_$RANDOM}" +JOB_DIR="${SHARED_ROOT}/jobs/${SUBMISSION_ID}" + +MODEL_ID="${MODEL_ID:-Qwen/Qwen2.5-0.5B-Instruct}" +TRAIN_FILE="${SHARED_ROOT}/datasets/gsm8k/train.parquet" +VAL_FILE="${SHARED_ROOT}/datasets/gsm8k/test.parquet" + +CODE_PATH="${CODE_PATH:-${SHARED_ROOT}/common/code/verl/verl_repo}" +TOTAL_TRAINING_STEPS="${TOTAL_TRAINING_STEPS:-10}" +SAVE_FREQ="${SAVE_FREQ:-10}" +TEST_FREQ="${TEST_FREQ:--1}" + +echo "[head] create job dir: ${JOB_DIR}" +dexec "${HEAD_CONTAINER}" bash -lc "mkdir -p '${JOB_DIR}'/{logs,checkpoints,config,debug}" + +SUBMIT_CMD="python3 -m verl.trainer.main_ppo \ +data.train_files=${TRAIN_FILE} \ +data.val_files=${VAL_FILE} \ +data.train_batch_size=256 \ +data.max_prompt_length=512 \ +data.max_response_length=512 \ +actor_rollout_ref.model.path=${MODEL_ID} \ +actor_rollout_ref.actor.optim.lr=1e-6 \ +actor_rollout_ref.actor.ppo_mini_batch_size=64 \ +actor_rollout_ref.actor.ppo_micro_batch_size_per_gpu=4 \ +actor_rollout_ref.rollout.name=sglang \ +actor_rollout_ref.rollout.log_prob_micro_batch_size_per_gpu=8 \ +actor_rollout_ref.rollout.tensor_model_parallel_size=1 \ +actor_rollout_ref.rollout.gpu_memory_utilization=0.4 \ +actor_rollout_ref.ref.log_prob_micro_batch_size_per_gpu=4 \ +critic.optim.lr=1e-5 \ +critic.model.path=${MODEL_ID} \ +critic.ppo_micro_batch_size_per_gpu=4 \ +algorithm.kl_ctrl.kl_coef=0.001 \ +trainer.logger=console \ +trainer.val_before_train=False \ +trainer.n_gpus_per_node=4 \ +trainer.nnodes=2 \ +trainer.save_freq=${SAVE_FREQ} \ +trainer.test_freq=${TEST_FREQ} \ +trainer.total_epochs=1 \ +trainer.total_training_steps=${TOTAL_TRAINING_STEPS} \ +trainer.resume_mode=disable \ +trainer.default_local_dir=${JOB_DIR}/checkpoints \ ++ray_kwargs.ray_init.address=auto \ +hydra.run.dir=${JOB_DIR}/logs/hydra" + +printf "%s\n" "${SUBMIT_CMD}" | dexec "${HEAD_CONTAINER}" bash -lc "cat > '${JOB_DIR}/config/submit_cmd.txt'" + +echo "[head] debug snapshot (pre-submit)" +dexec "${HEAD_CONTAINER}" bash -lc "ray status >'${JOB_DIR}/debug/ray_status_pre.txt' 2>&1 || true" + +echo "[head] submit PPO via ray job submit (driver forced on worker)" +SUBMIT_OUT="$(dexec "${HEAD_CONTAINER}" bash -lc "ray job submit --address='${RAY_DASHBOARD_ADDR}' --submission-id='${SUBMISSION_ID}' --entrypoint-num-cpus=1 --entrypoint-resources='{\"worker_node\": 1}' --runtime-env-json='{\"env_vars\":{\"HF_HOME\":\"${SHARED_ROOT}/hf\",\"HUGGINGFACE_HUB_CACHE\":\"${SHARED_ROOT}/hf/hub\",\"TRANSFORMERS_CACHE\":\"${SHARED_ROOT}/hf/transformers\",\"HF_ENDPOINT\":\"https://hf-mirror.com\",\"PYTHONUNBUFFERED\":\"1\",\"PYTHONPATH\":\"${CODE_PATH}:${SHARED_ROOT}/user/code\"}}' --no-wait -- ${SUBMIT_CMD}")" + +printf "%s\n" "${SUBMIT_OUT}" +printf "%s\n" "${SUBMIT_OUT}" | dexec "${HEAD_CONTAINER}" bash -lc "cat > '${JOB_DIR}/logs/ray_job_submit.out'" +echo "${SUBMISSION_ID}" | dexec "${HEAD_CONTAINER}" bash -lc "cat > '${JOB_DIR}/config/ray_submission_id.txt'" + +echo "[head] debug snapshot (post-submit)" +dexec "${HEAD_CONTAINER}" bash -lc "ray job list >'${JOB_DIR}/debug/ray_job_list_post.txt' 2>&1 || true" +dexec "${HEAD_CONTAINER}" bash -lc "ray status >'${JOB_DIR}/debug/ray_status_post.txt' 2>&1 || true" + +echo "submitted: ${SUBMISSION_ID}" +echo "job dir: ${JOB_DIR}" diff --git a/src/mvp/v1.1/scripts/41_submit_grpo_epoch1.sh b/src/mvp/v1.1/scripts/41_submit_grpo_epoch1.sh new file mode 100644 index 0000000..835a6da --- /dev/null +++ b/src/mvp/v1.1/scripts/41_submit_grpo_epoch1.sh @@ -0,0 +1,73 @@ +#!/usr/bin/env bash +set -euo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +# shellcheck source=lib.sh +source "${SCRIPT_DIR}/lib.sh" + +SUBMISSION_ID="${SUBMISSION_ID:-mvp11_grpo_$(timestamp)_$RANDOM}" +JOB_DIR="${SHARED_ROOT}/jobs/${SUBMISSION_ID}" + +MODEL_ID="${MODEL_ID:-Qwen/Qwen2.5-0.5B-Instruct}" +TRAIN_FILE="${SHARED_ROOT}/datasets/gsm8k/train.parquet" +VAL_FILE="${SHARED_ROOT}/datasets/gsm8k/test.parquet" + +CODE_PATH="${CODE_PATH:-${SHARED_ROOT}/common/code/verl/verl_repo}" +TOTAL_TRAINING_STEPS="${TOTAL_TRAINING_STEPS:-10}" +SAVE_FREQ="${SAVE_FREQ:-10}" +TEST_FREQ="${TEST_FREQ:--1}" + +echo "[head] create job dir: ${JOB_DIR}" +dexec "${HEAD_CONTAINER}" bash -lc "mkdir -p '${JOB_DIR}'/{logs,checkpoints,config,debug}" + +SUBMIT_CMD="python3 -m verl.trainer.main_ppo \ +data.train_files=${TRAIN_FILE} \ +data.val_files=${VAL_FILE} \ +data.train_batch_size=256 \ +data.max_prompt_length=512 \ +data.max_response_length=512 \ +actor_rollout_ref.model.path=${MODEL_ID} \ +actor_rollout_ref.actor.optim.lr=1e-6 \ +actor_rollout_ref.actor.ppo_mini_batch_size=64 \ +actor_rollout_ref.actor.ppo_micro_batch_size_per_gpu=4 \ +actor_rollout_ref.rollout.name=sglang \ +actor_rollout_ref.rollout.log_prob_micro_batch_size_per_gpu=8 \ +actor_rollout_ref.rollout.tensor_model_parallel_size=1 \ +actor_rollout_ref.rollout.gpu_memory_utilization=0.4 \ +actor_rollout_ref.ref.log_prob_micro_batch_size_per_gpu=4 \ +critic.optim.lr=1e-5 \ +critic.model.path=${MODEL_ID} \ +critic.ppo_micro_batch_size_per_gpu=4 \ +algorithm.adv_estimator=grpo \ +algorithm.kl_ctrl.kl_coef=0.001 \ +trainer.logger=console \ +trainer.val_before_train=False \ +trainer.n_gpus_per_node=4 \ +trainer.nnodes=2 \ +trainer.save_freq=${SAVE_FREQ} \ +trainer.test_freq=${TEST_FREQ} \ +trainer.total_epochs=1 \ +trainer.total_training_steps=${TOTAL_TRAINING_STEPS} \ +trainer.resume_mode=disable \ +trainer.default_local_dir=${JOB_DIR}/checkpoints \ ++ray_kwargs.ray_init.address=auto \ +hydra.run.dir=${JOB_DIR}/logs/hydra" + +printf "%s\n" "${SUBMIT_CMD}" | dexec "${HEAD_CONTAINER}" bash -lc "cat > '${JOB_DIR}/config/submit_cmd.txt'" + +echo "[head] debug snapshot (pre-submit)" +dexec "${HEAD_CONTAINER}" bash -lc "ray status >'${JOB_DIR}/debug/ray_status_pre.txt' 2>&1 || true" + +echo "[head] submit GRPO via ray job submit (driver forced on worker)" +SUBMIT_OUT="$(dexec "${HEAD_CONTAINER}" bash -lc "ray job submit --address='${RAY_DASHBOARD_ADDR}' --submission-id='${SUBMISSION_ID}' --entrypoint-num-cpus=1 --entrypoint-resources='{\"worker_node\": 1}' --runtime-env-json='{\"env_vars\":{\"HF_HOME\":\"${SHARED_ROOT}/hf\",\"HUGGINGFACE_HUB_CACHE\":\"${SHARED_ROOT}/hf/hub\",\"TRANSFORMERS_CACHE\":\"${SHARED_ROOT}/hf/transformers\",\"HF_ENDPOINT\":\"https://hf-mirror.com\",\"PYTHONUNBUFFERED\":\"1\",\"PYTHONPATH\":\"${CODE_PATH}:${SHARED_ROOT}/user/code\"}}' --no-wait -- ${SUBMIT_CMD}")" + +printf "%s\n" "${SUBMIT_OUT}" +printf "%s\n" "${SUBMIT_OUT}" | dexec "${HEAD_CONTAINER}" bash -lc "cat > '${JOB_DIR}/logs/ray_job_submit.out'" +echo "${SUBMISSION_ID}" | dexec "${HEAD_CONTAINER}" bash -lc "cat > '${JOB_DIR}/config/ray_submission_id.txt'" + +echo "[head] debug snapshot (post-submit)" +dexec "${HEAD_CONTAINER}" bash -lc "ray job list >'${JOB_DIR}/debug/ray_job_list_post.txt' 2>&1 || true" +dexec "${HEAD_CONTAINER}" bash -lc "ray status >'${JOB_DIR}/debug/ray_status_post.txt' 2>&1 || true" + +echo "submitted: ${SUBMISSION_ID}" +echo "job dir: ${JOB_DIR}" diff --git a/src/mvp/v1.1/scripts/42_submit_sft_minimal.sh b/src/mvp/v1.1/scripts/42_submit_sft_minimal.sh new file mode 100644 index 0000000..d30c7f4 --- /dev/null +++ b/src/mvp/v1.1/scripts/42_submit_sft_minimal.sh @@ -0,0 +1,62 @@ +#!/usr/bin/env bash +set -euo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +# shellcheck source=lib.sh +source "${SCRIPT_DIR}/lib.sh" + +SUBMISSION_ID="${SUBMISSION_ID:-mvp11_sft_$(timestamp)_$RANDOM}" +JOB_DIR="${SHARED_ROOT}/jobs/${SUBMISSION_ID}" + +MODEL_ID="${MODEL_ID:-Qwen/Qwen2.5-0.5B-Instruct}" +TRAIN_FILE="${SHARED_ROOT}/datasets/gsm8k_sft/train.parquet" +VAL_FILE="${SHARED_ROOT}/datasets/gsm8k_sft/test.parquet" + +CODE_PATH="${CODE_PATH:-${SHARED_ROOT}/common/code/verl/verl_repo}" +TOTAL_TRAINING_STEPS="${TOTAL_TRAINING_STEPS:-10}" +SAVE_FREQ="${SAVE_FREQ:-10}" +SFT_DRIVER_DEVICE="${SFT_DRIVER_DEVICE:-cpu}" + +echo "[head] create job dir: ${JOB_DIR}" +dexec "${HEAD_CONTAINER}" bash -lc "mkdir -p '${JOB_DIR}'/{logs,checkpoints,config,debug}" + +SUBMIT_CMD="python3 -m verl.trainer.sft_trainer_ray \ +model.path=${MODEL_ID} \ +data.train_files=${TRAIN_FILE} \ +data.val_files=null \ +data.train_batch_size=64 \ +data.micro_batch_size_per_gpu=1 \ +data.max_token_len_per_gpu=2048 \ +data.max_length=1024 \ +trainer.logger=console \ +trainer.project_name=mvp11-sft \ +trainer.experiment_name=${SUBMISSION_ID} \ +trainer.total_epochs=1 \ +trainer.total_training_steps=${TOTAL_TRAINING_STEPS} \ +trainer.save_freq=${SAVE_FREQ} \ +trainer.test_freq=-1 \ +trainer.resume_mode=disable \ +trainer.device=${SFT_DRIVER_DEVICE} \ +trainer.default_local_dir=${JOB_DIR}/checkpoints \ +trainer.nnodes=2 \ +trainer.n_gpus_per_node=4 \ +hydra.run.dir=${JOB_DIR}/logs/hydra" + +printf "%s\n" "${SUBMIT_CMD}" | dexec "${HEAD_CONTAINER}" bash -lc "cat > '${JOB_DIR}/config/submit_cmd.txt'" + +echo "[head] debug snapshot (pre-submit)" +dexec "${HEAD_CONTAINER}" bash -lc "ray status >'${JOB_DIR}/debug/ray_status_pre.txt' 2>&1 || true" + +echo "[head] submit SFT via ray job submit (driver forced on worker)" +SUBMIT_OUT="$(dexec "${HEAD_CONTAINER}" bash -lc "ray job submit --address='${RAY_DASHBOARD_ADDR}' --submission-id='${SUBMISSION_ID}' --entrypoint-num-cpus=1 --entrypoint-resources='{\"worker_node\": 1}' --runtime-env-json='{\"env_vars\":{\"HF_HOME\":\"${SHARED_ROOT}/hf\",\"HUGGINGFACE_HUB_CACHE\":\"${SHARED_ROOT}/hf/hub\",\"TRANSFORMERS_CACHE\":\"${SHARED_ROOT}/hf/transformers\",\"HF_ENDPOINT\":\"https://hf-mirror.com\",\"PYTHONUNBUFFERED\":\"1\",\"RAY_ADDRESS\":\"auto\",\"PYTHONPATH\":\"${CODE_PATH}:${SHARED_ROOT}/user/code\"}}' --no-wait -- ${SUBMIT_CMD}")" + +printf "%s\n" "${SUBMIT_OUT}" +printf "%s\n" "${SUBMIT_OUT}" | dexec "${HEAD_CONTAINER}" bash -lc "cat > '${JOB_DIR}/logs/ray_job_submit.out'" +echo "${SUBMISSION_ID}" | dexec "${HEAD_CONTAINER}" bash -lc "cat > '${JOB_DIR}/config/ray_submission_id.txt'" + +echo "[head] debug snapshot (post-submit)" +dexec "${HEAD_CONTAINER}" bash -lc "ray job list >'${JOB_DIR}/debug/ray_job_list_post.txt' 2>&1 || true" +dexec "${HEAD_CONTAINER}" bash -lc "ray status >'${JOB_DIR}/debug/ray_status_post.txt' 2>&1 || true" + +echo "submitted: ${SUBMISSION_ID}" +echo "job dir: ${JOB_DIR}" diff --git a/src/mvp/v1.1/scripts/43_submit_jobspec.sh b/src/mvp/v1.1/scripts/43_submit_jobspec.sh new file mode 100644 index 0000000..c4a1e19 --- /dev/null +++ b/src/mvp/v1.1/scripts/43_submit_jobspec.sh @@ -0,0 +1,17 @@ +#!/usr/bin/env bash +set -euo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +# shellcheck source=lib.sh +source "${SCRIPT_DIR}/lib.sh" + +SPEC_PATH="${1:-}" +if [[ -z "${SPEC_PATH}" ]]; then + echo "usage: $0 " >&2 + echo "example: $0 /workspace/mvp/v1.1/templates/ppo.json" >&2 + exit 1 +fi + +# Submit from head container (required), but with driver forced onto worker via entrypoint resources in spec. +dexec "${HEAD_CONTAINER}" bash -lc "SHARED_ROOT='${SHARED_ROOT}' python3 /workspace/mvp/v1.1/submit_job.py --spec '${SPEC_PATH}' --no-wait" + diff --git a/src/mvp/v1.1/scripts/44_submit_sdk.sh b/src/mvp/v1.1/scripts/44_submit_sdk.sh new file mode 100644 index 0000000..d7bf010 --- /dev/null +++ b/src/mvp/v1.1/scripts/44_submit_sdk.sh @@ -0,0 +1,19 @@ +#!/usr/bin/env bash +set -euo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +# shellcheck source=lib.sh +source "${SCRIPT_DIR}/lib.sh" + +CONFIG_PATH="${1:-/workspace/mvp/v1.1/py/configs/dev.yaml}" +JOBSPEC_PATH="${2:-}" + +if [[ -z "${JOBSPEC_PATH}" ]]; then + echo "usage: $0 " >&2 + echo "example: $0 /workspace/mvp/v1.1/py/configs/dev.yaml /workspace/mvp/v1.1/py/jobspecs/ppo.yaml" >&2 + exit 1 +fi + +echo "[head] submit via Ray Python SDK" +dexec "${HEAD_CONTAINER}" bash -lc "python3 /workspace/mvp/v1.1/py/run.py --config '${CONFIG_PATH}' --jobspec '${JOBSPEC_PATH}' --action submit --no-wait" + diff --git a/src/mvp/v1.1/scripts/46_submit_ppo_two_verl_tags.sh b/src/mvp/v1.1/scripts/46_submit_ppo_two_verl_tags.sh new file mode 100644 index 0000000..9ccf831 --- /dev/null +++ b/src/mvp/v1.1/scripts/46_submit_ppo_two_verl_tags.sh @@ -0,0 +1,86 @@ +#!/usr/bin/env bash +set -euo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +# shellcheck source=lib.sh +source "${SCRIPT_DIR}/lib.sh" + +CONFIG_PATH="${1:-/workspace/mvp/v1.1/py/configs/dev.yaml}" + +TS="$(timestamp)" +BASE="/workspace/mvp/v1.1/py/jobspecs" + +NNODES="${NNODES:-1}" +N_GPUS_PER_NODE="${N_GPUS_PER_NODE:-1}" + +wait_job() { + local sid="$1" + echo "[head] wait: ${sid}" + while true; do + # Ray returns one of: PENDING/RUNNING/SUCCEEDED/FAILED/STOPPED + st="$(dexec "${HEAD_CONTAINER}" bash -lc "python3 /workspace/mvp/v1.1/py/run.py --config '${CONFIG_PATH}' --action status --submission-id '${sid}'" | tr -d '\r' | tail -n 1)" + echo "[head] status: ${sid} -> ${st}" + case "${st}" in + *SUCCEEDED*) + return 0 + ;; + *FAILED*|*STOPPED*) + echo "[head] job not successful: ${sid} (${st})" >&2 + echo "[head] last logs:" + dexec "${HEAD_CONTAINER}" bash -lc "python3 /workspace/mvp/v1.1/py/run.py --config '${CONFIG_PATH}' --action logs --submission-id '${sid}' --tail 200" || true + return 1 + ;; + *) + sleep 10 + ;; + esac + done +} + +show_precheck() { + local sid="$1" + echo "[head] verify precheck: ${sid}" + dexec "${HEAD_CONTAINER}" bash -lc "python3 /workspace/mvp/v1.1/py/run.py --config '${CONFIG_PATH}' --action logs --submission-id '${sid}' --tail 2000 | egrep 'MVP_PRECHECK_VERL_FILE|MVP_PRECHECK_MARKER' || true" +} + +make_spec() { + local tag="$1" + local code_path="$2" + local out_path="$3" + local sid="mvp11_ppo_${tag//./_}_${TS}" + dexec "${HEAD_CONTAINER}" bash -lc "cat > '${out_path}' <<'YAML' +workload: \"ppo\" +submission_id: \"${sid}\" +code_path: \"${code_path}\" +model_id: \"Qwen/Qwen2.5-0.5B-Instruct\" +train_file: \"/private/datasets/gsm8k/train.parquet\" +val_file: \"/private/datasets/gsm8k/test.parquet\" +nnodes: ${NNODES} +n_gpus_per_node: ${N_GPUS_PER_NODE} +total_epochs: 1 +total_training_steps: 10 +save_freq: 10 +test_freq: -1 +YAML" + echo "${sid}" +} + +echo "[head] submit PPO sequentially with verl v0.6.0 then v0.6.1" +echo "[head] resources: nnodes=${NNODES} n_gpus_per_node=${N_GPUS_PER_NODE}" + +sid0="$(make_spec "v0.6.0" "/private/common/code/verl/verl_v0.6.0" "${BASE}/tmp_ppo_verl_v0.6.0_${TS}.yaml")" +echo "[head] submit: ${sid0}" +dexec "${HEAD_CONTAINER}" bash -lc "python3 /workspace/mvp/v1.1/py/run.py --config '${CONFIG_PATH}' --jobspec '${BASE}/tmp_ppo_verl_v0.6.0_${TS}.yaml' --action submit --no-wait" +wait_job "${sid0}" +show_precheck "${sid0}" + +sid1="$(make_spec "v0.6.1" "/private/common/code/verl/verl_v0.6.1" "${BASE}/tmp_ppo_verl_v0.6.1_${TS}.yaml")" +echo "[head] submit: ${sid1}" +dexec "${HEAD_CONTAINER}" bash -lc "python3 /workspace/mvp/v1.1/py/run.py --config '${CONFIG_PATH}' --jobspec '${BASE}/tmp_ppo_verl_v0.6.1_${TS}.yaml' --action submit --no-wait" +wait_job "${sid1}" +show_precheck "${sid1}" + +echo "[head] done" +echo "submitted:" +echo " ${sid0} (verl v0.6.0)" +echo " ${sid1} (verl v0.6.1)" diff --git a/src/mvp/v1.1/scripts/50_status.sh b/src/mvp/v1.1/scripts/50_status.sh new file mode 100644 index 0000000..03847a0 --- /dev/null +++ b/src/mvp/v1.1/scripts/50_status.sh @@ -0,0 +1,13 @@ +#!/usr/bin/env bash +set -euo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +# shellcheck source=lib.sh +source "${SCRIPT_DIR}/lib.sh" + +echo "[head] ray status" +dexec "${HEAD_CONTAINER}" bash -lc "ray status || true" + +echo "[head] ray job list" +dexec "${HEAD_CONTAINER}" bash -lc "ray job list || true" + diff --git a/src/mvp/v1.1/scripts/lib.sh b/src/mvp/v1.1/scripts/lib.sh new file mode 100644 index 0000000..23e0997 --- /dev/null +++ b/src/mvp/v1.1/scripts/lib.sh @@ -0,0 +1,52 @@ +#!/usr/bin/env bash +set -euo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +ROOT_DIR="$(cd "${SCRIPT_DIR}/.." && pwd)" + +COMPOSE_FILE="${ROOT_DIR}/docker-compose.yaml" + +HEAD_CONTAINER="mvp11-ray-head" +WORKER0_CONTAINER="mvp11-ray-worker-0" +WORKER1_CONTAINER="mvp11-ray-worker-1" + +SHARED_ROOT="${SHARED_ROOT:-/private}" +RAY_DASHBOARD_ADDR="${RAY_DASHBOARD_ADDR:-http://127.0.0.1:8265}" + +dc() { + docker compose --project-directory "${ROOT_DIR}" -f "${COMPOSE_FILE}" "$@" +} + +require_cmd() { + local cmd="$1" + command -v "${cmd}" >/dev/null 2>&1 || { + echo "missing required command: ${cmd}" >&2 + exit 1 + } +} + +ensure_container_running() { + local name="$1" + if ! docker ps --format '{{.Names}}' | grep -qx "${name}"; then + echo "container not running: ${name}" >&2 + exit 1 + fi +} + +dexec() { + local name="$1" + shift + ensure_container_running "${name}" + docker exec -i "${name}" "$@" +} + +container_ip() { + local name="$1" + ensure_container_running "${name}" + docker inspect -f '{{range .NetworkSettings.Networks}}{{.IPAddress}}{{end}}' "${name}" +} + +timestamp() { + date +"%Y%m%d_%H%M%S" +} + diff --git a/src/mvp/v1.1/scripts/run_all.sh b/src/mvp/v1.1/scripts/run_all.sh new file mode 100644 index 0000000..cf84c01 --- /dev/null +++ b/src/mvp/v1.1/scripts/run_all.sh @@ -0,0 +1,15 @@ +#!/usr/bin/env bash +set -euo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" + +"${SCRIPT_DIR}/00_prereq_check.sh" +"${SCRIPT_DIR}/01_up.sh" +"${SCRIPT_DIR}/20_start_head.sh" +"${SCRIPT_DIR}/21_start_workers.sh" +"${SCRIPT_DIR}/30_prepare_data_and_model.sh" +"${SCRIPT_DIR}/40_submit_ppo_epoch1.sh" +"${SCRIPT_DIR}/41_submit_grpo_epoch1.sh" +"${SCRIPT_DIR}/42_submit_sft_minimal.sh" +"${SCRIPT_DIR}/50_status.sh" + diff --git a/src/mvp/v1.1/submit_job.py b/src/mvp/v1.1/submit_job.py new file mode 100644 index 0000000..9e8f54c --- /dev/null +++ b/src/mvp/v1.1/submit_job.py @@ -0,0 +1,282 @@ +#!/usr/bin/env python3 +import argparse +import json +import os +import shlex +import subprocess +from datetime import datetime +from pathlib import Path + + +def _ts(): + return datetime.now().strftime("%Y%m%d_%H%M%S") + + +def _expand(value: str) -> str: + return os.path.expandvars(value) + + +def _mkdir(path: Path) -> None: + path.mkdir(parents=True, exist_ok=True) + + +def _write_text(path: Path, content: str) -> None: + _mkdir(path.parent) + path.write_text(content, encoding="utf-8") + + +def _write_json(path: Path, obj) -> None: + _mkdir(path.parent) + path.write_text(json.dumps(obj, indent=2, ensure_ascii=False) + "\n", encoding="utf-8") + + +def _require(spec: dict, key: str): + if key not in spec: + raise SystemExit(f"missing required key in spec: {key}") + return spec[key] + + +def _default_submission_id(workload: str) -> str: + return f"mvp11_{workload}_{_ts()}_{os.getpid()}" + + +def _runtime_env(spec: dict) -> dict: + shared_root = _expand(_require(spec, "shared_root")) + code_path = _expand(_require(spec, "code_path")) + env_vars = dict(spec.get("runtime_env", {}).get("env_vars", {})) + + env_vars.setdefault("HF_HOME", f"{shared_root}/hf") + env_vars.setdefault("HUGGINGFACE_HUB_CACHE", f"{shared_root}/hf/hub") + env_vars.setdefault("TRANSFORMERS_CACHE", f"{shared_root}/hf/transformers") + env_vars.setdefault("PYTHONUNBUFFERED", "1") + + user_code = f"{shared_root}/user/code" + existing = env_vars.get("PYTHONPATH", "") + prefix = f"{code_path}:{user_code}" + env_vars["PYTHONPATH"] = f"{prefix}:{existing}" if existing else prefix + + # Helpful marker for logs/debugging + env_vars.setdefault("MVP_CODE_PATH", code_path) + + return {"env_vars": env_vars} + + +def _preflight_shell() -> str: + # Make multi-version/debugging observable in Ray job logs. + # `mvp_marker.py` is written by our snapshot script (optional); if missing, ignore. + py = r""" +import os +import sys +print("MVP_PRECHECK_PYTHON:", sys.executable) +print("MVP_PRECHECK_PYTHONPATH:", os.environ.get("PYTHONPATH")) +try: + import verl + print("MVP_PRECHECK_VERL_FILE:", getattr(verl, "__file__", None)) +except Exception as e: + print("MVP_PRECHECK_VERL_IMPORT_ERROR:", repr(e)) +try: + import mvp_marker + print("MVP_PRECHECK_MARKER:", getattr(mvp_marker, "MARKER", None)) +except Exception as e: + print("MVP_PRECHECK_MARKER_MISSING:", repr(e)) +""" + return f"python3 - <<'PY'\n{py.strip()}\nPY" + + +def _build_entrypoint(spec: dict, submission_id: str, job_dir: str) -> str: + workload = _require(spec, "workload") + model_id = _expand(_require(spec, "model_id")) + shared_root = _expand(_require(spec, "shared_root")) + + if workload in ("ppo", "grpo"): + cfg = spec.get(workload, {}) + train_file = _expand(cfg.get("train_file", f"{shared_root}/datasets/gsm8k/train.parquet")) + val_file = _expand(cfg.get("val_file", f"{shared_root}/datasets/gsm8k/test.parquet")) + nnodes = int(cfg.get("nnodes", 2)) + gpus_per_node = int(cfg.get("n_gpus_per_node", 4)) + total_epochs = int(cfg.get("total_epochs", 1)) + total_steps = int(cfg.get("total_training_steps", 10)) + save_freq = int(cfg.get("save_freq", 10)) + test_freq = int(cfg.get("test_freq", -1)) + + algo_overrides = "" + if workload == "grpo": + algo_overrides = "algorithm.adv_estimator=grpo" + + cmd = f"""python3 -m verl.trainer.main_ppo \ +data.train_files={train_file} \ +data.val_files={val_file} \ +data.train_batch_size=256 \ +data.max_prompt_length=512 \ +data.max_response_length=512 \ +actor_rollout_ref.model.path={model_id} \ +actor_rollout_ref.actor.optim.lr=1e-6 \ +actor_rollout_ref.actor.ppo_mini_batch_size=64 \ +actor_rollout_ref.actor.ppo_micro_batch_size_per_gpu=4 \ +actor_rollout_ref.rollout.name=sglang \ +actor_rollout_ref.rollout.log_prob_micro_batch_size_per_gpu=8 \ +actor_rollout_ref.rollout.tensor_model_parallel_size=1 \ +actor_rollout_ref.rollout.gpu_memory_utilization=0.4 \ +actor_rollout_ref.ref.log_prob_micro_batch_size_per_gpu=4 \ +critic.optim.lr=1e-5 \ +critic.model.path={model_id} \ +critic.ppo_micro_batch_size_per_gpu=4 \ +algorithm.kl_ctrl.kl_coef=0.001 \ +{algo_overrides} \ +trainer.logger=console \ +trainer.val_before_train=False \ +trainer.n_gpus_per_node={gpus_per_node} \ +trainer.nnodes={nnodes} \ +trainer.save_freq={save_freq} \ +trainer.test_freq={test_freq} \ +trainer.total_epochs={total_epochs} \ +trainer.total_training_steps={total_steps} \ +trainer.resume_mode=disable \ +trainer.default_local_dir={job_dir}/checkpoints \ ++ray_kwargs.ray_init.address=auto \ +hydra.run.dir={job_dir}/logs/hydra""" + return "\n".join([_preflight_shell(), "exec " + cmd]) + + if workload == "sft": + cfg = spec.get("sft", {}) + train_file = _expand(cfg.get("train_file", f"{shared_root}/datasets/gsm8k_sft/train.parquet")) + val_file = cfg.get("val_file", None) + nnodes = int(cfg.get("nnodes", 2)) + gpus_per_node = int(cfg.get("n_gpus_per_node", 4)) + total_epochs = int(cfg.get("total_epochs", 1)) + total_steps = int(cfg.get("total_training_steps", 10)) + save_freq = int(cfg.get("save_freq", 10)) + device = cfg.get("device", "cpu") + + val_override = "data.val_files=null" if val_file is None else f"data.val_files={_expand(val_file)}" + + # Note: driver should not require CUDA under ray job submit (no entrypoint GPUs by default). + cmd = f"""python3 -m verl.trainer.sft_trainer_ray \ +model.path={model_id} \ +data.train_files={train_file} \ +{val_override} \ +data.train_batch_size=64 \ +data.micro_batch_size_per_gpu=1 \ +data.max_token_len_per_gpu=2048 \ +data.max_length=1024 \ +trainer.logger=console \ +trainer.project_name=mvp11-sft \ +trainer.experiment_name={submission_id} \ +trainer.total_epochs={total_epochs} \ +trainer.total_training_steps={total_steps} \ +trainer.save_freq={save_freq} \ +trainer.test_freq=-1 \ +trainer.resume_mode=disable \ +trainer.device={device} \ +trainer.default_local_dir={job_dir}/checkpoints \ +trainer.nnodes={nnodes} \ +trainer.n_gpus_per_node={gpus_per_node} \ +hydra.run.dir={job_dir}/logs/hydra""" + return "\n".join([_preflight_shell(), "exec " + cmd]) + + raise SystemExit(f"unsupported workload: {workload}") + + +def main() -> int: + parser = argparse.ArgumentParser() + parser.add_argument("--spec", required=True, help="Path to JobSpec json (inside this container)") + parser.add_argument("--no-wait", action="store_true", help="Submit and return immediately") + parser.add_argument("--dry-run", action="store_true", help="Only print the submit command") + args = parser.parse_args() + + spec_path = Path(args.spec) + spec = json.loads(spec_path.read_text(encoding="utf-8")) + + shared_root = _expand(_require(spec, "shared_root")) + workload = _require(spec, "workload") + + submission_id = spec.get("submission_id") or _default_submission_id(workload) + job_dir = f"{shared_root}/jobs/{submission_id}" + + ray_cfg = _require(spec, "ray") + ray_addr = ray_cfg.get("address", "http://127.0.0.1:8265") + entrypoint_num_cpus = ray_cfg.get("entrypoint_num_cpus", 1) + entrypoint_resources = ray_cfg.get("entrypoint_resources", {"worker_node": 1}) + + runtime_env = _runtime_env(spec) + entrypoint = _build_entrypoint(spec, submission_id=submission_id, job_dir=job_dir) + + # Prepare job dir + job_root = Path(job_dir) + _mkdir(job_root / "config") + _mkdir(job_root / "logs") + _mkdir(job_root / "checkpoints") + _mkdir(job_root / "debug") + + # Snapshot config for audit/debug + _write_json(job_root / "config" / "job_spec.json", spec) + _write_json(job_root / "config" / "runtime_env.json", runtime_env) + _write_text(job_root / "config" / "ray_submission_id.txt", submission_id + "\n") + + submit_cmd_txt = "\n".join( + [ + "ray job submit", + f" --address={ray_addr}", + f" --submission-id={submission_id}", + f" --entrypoint-num-cpus={entrypoint_num_cpus}", + f" --entrypoint-resources={json.dumps(entrypoint_resources)}", + f" --runtime-env-json=", + f" {'--no-wait' if args.no_wait else ''}", + " -- bash -lc ''", + ] + ) + _write_text(job_root / "config" / "submit_cmd.txt", submit_cmd_txt + "\n") + + # Debug snapshot (pre-submit) + try: + pre = subprocess.run(["ray", "status"], capture_output=True, text=True, check=False) + _write_text(job_root / "debug" / "ray_status_pre.txt", (pre.stdout or "") + (pre.stderr or "")) + except FileNotFoundError: + _write_text(job_root / "debug" / "ray_status_pre.txt", "ray cli not found\n") + + submit_args = [ + "ray", + "job", + "submit", + "--address", + ray_addr, + "--submission-id", + submission_id, + "--entrypoint-num-cpus", + str(entrypoint_num_cpus), + "--entrypoint-resources", + json.dumps(entrypoint_resources), + "--runtime-env-json", + json.dumps(runtime_env), + ] + if args.no_wait: + submit_args.append("--no-wait") + submit_args += ["--", "bash", "-lc", entrypoint] + + if args.dry_run: + print(" ".join(shlex.quote(x) for x in submit_args)) + return 0 + + proc = subprocess.run(submit_args, capture_output=True, text=True, check=False) + _write_text(job_root / "logs" / "ray_job_submit.out", (proc.stdout or "") + (proc.stderr or "")) + print(proc.stdout, end="") + if proc.returncode != 0: + print(proc.stderr, end="", file=os.sys.stderr) + return proc.returncode + + # Debug snapshot (post-submit) + try: + post = subprocess.run(["ray", "job", "list", "--log-style=record", "-v"], capture_output=True, text=True, check=False) + _write_text(job_root / "debug" / "ray_job_list_post.txt", (post.stdout or "") + (post.stderr or "")) + post2 = subprocess.run(["ray", "status"], capture_output=True, text=True, check=False) + _write_text(job_root / "debug" / "ray_status_post.txt", (post2.stdout or "") + (post2.stderr or "")) + except FileNotFoundError: + pass + + print(f"job_dir: {job_dir}") + return 0 + + +if __name__ == "__main__": + raise SystemExit(main()) + diff --git a/src/mvp/v1.1/templates/grpo.json b/src/mvp/v1.1/templates/grpo.json new file mode 100644 index 0000000..cd98915 --- /dev/null +++ b/src/mvp/v1.1/templates/grpo.json @@ -0,0 +1,31 @@ +{ + "submission_id": "", + "workload": "grpo", + "shared_root": "${SHARED_ROOT}", + "code_path": "${SHARED_ROOT}/common/code/verl/verl_repo", + "model_id": "Qwen/Qwen2.5-0.5B-Instruct", + "grpo": { + "train_file": "${SHARED_ROOT}/datasets/gsm8k/train.parquet", + "val_file": "${SHARED_ROOT}/datasets/gsm8k/test.parquet", + "nnodes": 2, + "n_gpus_per_node": 4, + "total_epochs": 1, + "total_training_steps": 10, + "save_freq": 10, + "test_freq": -1 + }, + "ray": { + "address": "http://127.0.0.1:8265", + "entrypoint_num_cpus": 1, + "entrypoint_resources": { + "worker_node": 1 + } + }, + "runtime_env": { + "env_vars": { + "HF_ENDPOINT": "https://hf-mirror.com", + "PYTHONUNBUFFERED": "1" + } + } +} + diff --git a/src/mvp/v1.1/templates/ppo.json b/src/mvp/v1.1/templates/ppo.json new file mode 100644 index 0000000..89241a4 --- /dev/null +++ b/src/mvp/v1.1/templates/ppo.json @@ -0,0 +1,31 @@ +{ + "submission_id": "", + "workload": "ppo", + "shared_root": "${SHARED_ROOT}", + "code_path": "${SHARED_ROOT}/common/code/verl/verl_repo", + "model_id": "Qwen/Qwen2.5-0.5B-Instruct", + "ppo": { + "train_file": "${SHARED_ROOT}/datasets/gsm8k/train.parquet", + "val_file": "${SHARED_ROOT}/datasets/gsm8k/test.parquet", + "nnodes": 2, + "n_gpus_per_node": 4, + "total_epochs": 1, + "total_training_steps": 10, + "save_freq": 10, + "test_freq": -1 + }, + "ray": { + "address": "http://127.0.0.1:8265", + "entrypoint_num_cpus": 1, + "entrypoint_resources": { + "worker_node": 1 + } + }, + "runtime_env": { + "env_vars": { + "HF_ENDPOINT": "https://hf-mirror.com", + "PYTHONUNBUFFERED": "1" + } + } +} + diff --git a/src/mvp/v1.1/templates/sft.json b/src/mvp/v1.1/templates/sft.json new file mode 100644 index 0000000..9d4f8af --- /dev/null +++ b/src/mvp/v1.1/templates/sft.json @@ -0,0 +1,32 @@ +{ + "submission_id": "", + "workload": "sft", + "shared_root": "${SHARED_ROOT}", + "code_path": "${SHARED_ROOT}/common/code/verl/verl_repo", + "model_id": "Qwen/Qwen2.5-0.5B-Instruct", + "sft": { + "train_file": "${SHARED_ROOT}/datasets/gsm8k_sft/train.parquet", + "val_file": null, + "nnodes": 2, + "n_gpus_per_node": 4, + "total_epochs": 1, + "total_training_steps": 10, + "save_freq": 10, + "device": "cpu" + }, + "ray": { + "address": "http://127.0.0.1:8265", + "entrypoint_num_cpus": 1, + "entrypoint_resources": { + "worker_node": 1 + } + }, + "runtime_env": { + "env_vars": { + "HF_ENDPOINT": "https://hf-mirror.com", + "PYTHONUNBUFFERED": "1", + "RAY_ADDRESS": "auto" + } + } +} + diff --git a/src/mvp/v1/arch.excalidraw b/src/mvp/v1/arch.excalidraw new file mode 100644 index 0000000..b9e50ce --- /dev/null +++ b/src/mvp/v1/arch.excalidraw @@ -0,0 +1,1877 @@ +{ + "type": "excalidraw", + "version": 2, + "source": "https://marketplace.visualstudio.com/items?itemName=pomdtr.excalidraw-editor", + "elements": [ + { + "id": "9Ql3xet2wP4Oy9RJ4s-8H", + "type": "rectangle", + "x": 165.33361053466797, + "y": 124.66671752929688, + "width": 201.66666412353516, + "height": 144.66665649414062, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "a0", + "roundness": { + "type": 3 + }, + "seed": 1490759950, + "version": 404, + "versionNonce": 1106431423, + "isDeleted": false, + "boundElements": [ + { + "type": "text", + "id": "Yx_fL1nYHbxINMBHYImpi" + }, + { + "id": "bAFwE3wo46b9BFt_2Gnm2", + "type": "arrow" + }, + { + "id": "Qzy3gfKzdZyrTwzOpuh86", + "type": "arrow" + }, + { + "id": "zj5n4D3014Kl-BdrNuRZp", + "type": "arrow" + }, + { + "id": "cqQ2Ij98wYdbKqXQcPhxe", + "type": "arrow" + } + ], + "updated": 1766373694946, + "link": null, + "locked": false + }, + { + "id": "Yx_fL1nYHbxINMBHYImpi", + "type": "text", + "x": 232.60698318481445, + "y": 184.5000457763672, + "width": 67.11991882324219, + "height": 25, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "a1", + "roundness": null, + "seed": 122704078, + "version": 259, + "versionNonce": 104768575, + "isDeleted": false, + "boundElements": [], + "updated": 1766372702157, + "link": null, + "locked": false, + "text": "scripts", + "fontSize": 20, + "fontFamily": 5, + "textAlign": "center", + "verticalAlign": "middle", + "containerId": "9Ql3xet2wP4Oy9RJ4s-8H", + "originalText": "scripts", + "autoResize": true, + "lineHeight": 1.25 + }, + { + "id": "1BW7RJyVw0yjg93vAIlWT", + "type": "rectangle", + "x": 813.8334465026855, + "y": 81.00010681152344, + "width": 159.9999771118164, + "height": 88.66665649414062, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "a2", + "roundness": { + "type": 3 + }, + "seed": 929976718, + "version": 789, + "versionNonce": 761000657, + "isDeleted": false, + "boundElements": [ + { + "id": "qcyVAzD12V9EAxE15fwAq", + "type": "arrow" + }, + { + "id": "Qzy3gfKzdZyrTwzOpuh86", + "type": "arrow" + } + ], + "updated": 1766373762851, + "link": null, + "locked": false + }, + { + "id": "nIO4qa9Aj4LF_WlHt2BNO", + "type": "rectangle", + "x": 455.8334159851074, + "y": 125.66667175292969, + "width": 201.66666412353516, + "height": 144.66665649414062, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "a4", + "roundness": { + "type": 3 + }, + "seed": 2137052434, + "version": 311, + "versionNonce": 1823398289, + "isDeleted": false, + "boundElements": [ + { + "id": "bAFwE3wo46b9BFt_2Gnm2", + "type": "arrow" + }, + { + "id": "QgI8Gn67BTTI1MdOhOjjn", + "type": "arrow" + }, + { + "id": "qcyVAzD12V9EAxE15fwAq", + "type": "arrow" + } + ], + "updated": 1766373745383, + "link": null, + "locked": false + }, + { + "id": "bAFwE3wo46b9BFt_2Gnm2", + "type": "arrow", + "x": 372.0002746582031, + "y": 196.9000457763672, + "width": 252.06685104370115, + "height": 90.56665649414063, + "angle": 0, + "strokeColor": "#e03131", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "a6", + "roundness": null, + "seed": 1158070290, + "version": 571, + "versionNonce": 231145393, + "isDeleted": false, + "boundElements": [], + "updated": 1766373712586, + "link": null, + "locked": false, + "points": [ + [ + 0, + 0 + ], + [ + 114.08342552185059, + 0 + ], + [ + 114.08342552185059, + -90.56665649414063 + ], + [ + 252.06685104370115, + -90.56665649414063 + ], + [ + 252.06685104370115, + -55.56665649414063 + ] + ], + "lastCommittedPoint": null, + "startBinding": { + "elementId": "9Ql3xet2wP4Oy9RJ4s-8H", + "fixedPoint": [ + 1.0247933887424108, + 0.4993087557117625 + ], + "focus": 0, + "gap": 0 + }, + "endBinding": { + "elementId": "e58le7CfnSkWVb-LoMmj4", + "fixedPoint": [ + 0.49736842105263096, + -0.13157894736842105 + ], + "focus": 0, + "gap": 0 + }, + "startArrowhead": null, + "endArrowhead": "arrow", + "elbowed": true, + "fixedSegments": null, + "startIsSpecial": null, + "endIsSpecial": null + }, + { + "id": "aBQ9npStDj077n0B7_JZ3", + "type": "rectangle", + "x": 814.000072479248, + "y": 227.6667022705078, + "width": 159.9999771118164, + "height": 88.66665649414062, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "a7", + "roundness": { + "type": 3 + }, + "seed": 1963206674, + "version": 814, + "versionNonce": 371608031, + "isDeleted": false, + "boundElements": [ + { + "id": "QgI8Gn67BTTI1MdOhOjjn", + "type": "arrow" + }, + { + "id": "zj5n4D3014Kl-BdrNuRZp", + "type": "arrow" + } + ], + "updated": 1766373791812, + "link": null, + "locked": false + }, + { + "id": "QgI8Gn67BTTI1MdOhOjjn", + "type": "arrow", + "x": 648.1671257019043, + "y": 165.23338928222657, + "width": 145.8328742980957, + "height": 121.4333282470703, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "aC", + "roundness": null, + "seed": 990553042, + "version": 371, + "versionNonce": 1356695409, + "isDeleted": false, + "boundElements": [], + "updated": 1766373712586, + "link": null, + "locked": false, + "points": [ + [ + 0, + 0 + ], + [ + 80.41647338867188, + 0 + ], + [ + 80.41647338867188, + 121.4333282470703 + ], + [ + 145.8328742980957, + 121.4333282470703 + ] + ], + "lastCommittedPoint": null, + "startBinding": { + "elementId": "e58le7CfnSkWVb-LoMmj4", + "fixedPoint": [ + 1.131578947368421, + 0.49736842105263174 + ], + "focus": 0, + "gap": 0 + }, + "endBinding": { + "elementId": "aBQ9npStDj077n0B7_JZ3", + "fixedPoint": [ + -0.12500047087676108, + 0.66541378226761 + ], + "focus": 0, + "gap": 0 + }, + "startArrowhead": null, + "endArrowhead": "arrow", + "elbowed": true, + "fixedSegments": null, + "startIsSpecial": null, + "endIsSpecial": null + }, + { + "id": "atIqduRWNX97Sp6yKe4hj", + "type": "rectangle", + "x": 431.1368826709601, + "y": 26.199444213977472, + "width": 611.63434968004, + "height": 645.8789403469411, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "dotted", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "aw", + "roundness": { + "type": 3 + }, + "seed": 311282494, + "version": 376, + "versionNonce": 190431967, + "isDeleted": false, + "boundElements": [], + "updated": 1766373009377, + "link": null, + "locked": false + }, + { + "id": "k6SXmpRXbeQSbRrXG_o6a", + "type": "text", + "x": 656.1800675420305, + "y": -12.200757945542989, + "width": 199.19985961914062, + "height": 25, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "dotted", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "ax", + "roundness": null, + "seed": 570087970, + "version": 192, + "versionNonce": 158307807, + "isDeleted": false, + "boundElements": [], + "updated": 1766373821110, + "link": null, + "locked": false, + "text": "a single H20 machine", + "fontSize": 20, + "fontFamily": 5, + "textAlign": "left", + "verticalAlign": "top", + "containerId": null, + "originalText": "a single H20 machine", + "autoResize": true, + "lineHeight": 1.25 + }, + { + "id": "qcyVAzD12V9EAxE15fwAq", + "type": "arrow", + "x": 648.1671257019043, + "y": 165.23338928222657, + "width": 160.66632080078125, + "height": 39.99995422363281, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b03", + "roundness": null, + "seed": 696982833, + "version": 370, + "versionNonce": 1628299665, + "isDeleted": false, + "boundElements": null, + "updated": 1766373712586, + "link": null, + "locked": false, + "points": [ + [ + 0, + 0 + ], + [ + 80.33316040039062, + 0 + ], + [ + 80.33316040039062, + -39.99995422363281 + ], + [ + 160.66632080078125, + -39.99995422363281 + ] + ], + "lastCommittedPoint": null, + "startBinding": { + "elementId": "e58le7CfnSkWVb-LoMmj4", + "fixedPoint": [ + 1.131578947368421, + 0.49736842105263174 + ], + "focus": 0, + "gap": 0 + }, + "endBinding": { + "elementId": "1BW7RJyVw0yjg93vAIlWT", + "fixedPoint": [ + -0.031250004470349, + 0.4988721803217357 + ], + "focus": 0, + "gap": 0 + }, + "startArrowhead": null, + "endArrowhead": "arrow", + "elbowed": true, + "fixedSegments": null, + "startIsSpecial": null, + "endIsSpecial": null + }, + { + "id": "Qzy3gfKzdZyrTwzOpuh86", + "type": "arrow", + "x": 371.8835678755345, + "y": 173.74970208077258, + "width": 520.5167182267604, + "height": 132.74959526924914, + "angle": 0, + "strokeColor": "#e03131", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b04", + "roundness": null, + "seed": 626515153, + "version": 428, + "versionNonce": 451690431, + "isDeleted": false, + "boundElements": null, + "updated": 1766372702159, + "link": null, + "locked": false, + "points": [ + [ + 0, + 0 + ], + [ + 21.28340523847919, + 0 + ], + [ + 21.28340523847919, + -132.74959526924914 + ], + [ + 520.5167182267604, + -132.74959526924914 + ], + [ + 520.5167182267604, + -111.41628228096789 + ] + ], + "lastCommittedPoint": null, + "startBinding": { + "elementId": "9Ql3xet2wP4Oy9RJ4s-8H", + "fixedPoint": [ + 1.024214677416095, + 0.3392833272086004 + ], + "focus": 0, + "gap": 0 + }, + "endBinding": { + "elementId": "eeIv7mr_n3Lg6B3DlXJHR", + "fixedPoint": [ + 0.49736842105263096, + -0.13157894736842105 + ], + "focus": 0, + "gap": 0 + }, + "startArrowhead": null, + "endArrowhead": "arrow", + "elbowed": true, + "fixedSegments": [ + { + "index": 2, + "start": [ + 21.28340523847919, + 0 + ], + "end": [ + 21.28340523847919, + -132.74959526924914 + ] + } + ], + "startIsSpecial": false, + "endIsSpecial": false + }, + { + "id": "zj5n4D3014Kl-BdrNuRZp", + "type": "arrow", + "x": 371.87357276423296, + "y": 221.26066759479318, + "width": 525.8600263263431, + "height": 135.07269116985526, + "angle": 0, + "strokeColor": "#e03131", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b05", + "roundness": null, + "seed": 2137319889, + "version": 378, + "versionNonce": 1170250751, + "isDeleted": false, + "boundElements": null, + "updated": 1766372702159, + "link": null, + "locked": false, + "points": [ + [ + 0, + 0 + ], + [ + 35.29340034978071, + 0 + ], + [ + 35.29340034978071, + 135.07269116985526 + ], + [ + 525.8600263263431, + 135.07269116985526 + ], + [ + 525.8600263263431, + 116.40612622844901 + ] + ], + "lastCommittedPoint": null, + "startBinding": { + "elementId": "9Ql3xet2wP4Oy9RJ4s-8H", + "fixedPoint": [ + 1.0241651148800903, + 0.6677001626107852 + ], + "focus": 0, + "gap": 0 + }, + "endBinding": { + "elementId": "c4sciwjA9ZH_GjmspcBan", + "fixedPoint": [ + 0.49736842105262796, + 1.131578947368421 + ], + "focus": 0, + "gap": 0 + }, + "startArrowhead": null, + "endArrowhead": "arrow", + "elbowed": true, + "fixedSegments": [ + { + "index": 2, + "start": [ + 35.29340034978071, + 0 + ], + "end": [ + 35.29340034978071, + 135.07269116985526 + ] + } + ], + "startIsSpecial": false, + "endIsSpecial": false + }, + { + "id": "bzQAQDSv4fPQAKJkqw9an", + "type": "arrow", + "x": 287.8335380554199, + "y": -76.33323669433594, + "width": 70.00006103515625, + "height": 26, + "angle": 0, + "strokeColor": "#e03131", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b06", + "roundness": null, + "seed": 301869471, + "version": 103, + "versionNonce": 2132861809, + "isDeleted": false, + "boundElements": null, + "updated": 1766372404307, + "link": null, + "locked": false, + "points": [ + [ + 0, + 0 + ], + [ + 35.000030517578125, + 0 + ], + [ + 35.000030517578125, + -26 + ], + [ + 70.00006103515625, + -26 + ] + ], + "lastCommittedPoint": null, + "startBinding": null, + "endBinding": null, + "startArrowhead": null, + "endArrowhead": "arrow", + "elbowed": true, + "fixedSegments": null, + "startIsSpecial": null, + "endIsSpecial": null + }, + { + "id": "JoAr6EMunnvVMp7Uo00g5", + "type": "text", + "x": 379.1668510437012, + "y": -102.99992370605469, + "width": 165.33987426757812, + "height": 25, + "angle": 0, + "strokeColor": "#e03131", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b07", + "roundness": null, + "seed": 1372433471, + "version": 19, + "versionNonce": 1494393073, + "isDeleted": false, + "boundElements": null, + "updated": 1766372417246, + "link": null, + "locked": false, + "text": "start ray cluster", + "fontSize": 20, + "fontFamily": 5, + "textAlign": "left", + "verticalAlign": "top", + "containerId": null, + "originalText": "start ray cluster", + "autoResize": true, + "lineHeight": 1.25 + }, + { + "id": "mWLybQDTFo_Um3eT0cEDa", + "type": "arrow", + "x": 577.8335380554199, + "y": -84.33323669433594, + "width": 59.33331298828125, + "height": 24.66668701171875, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b08", + "roundness": null, + "seed": 2218161, + "version": 62, + "versionNonce": 49511103, + "isDeleted": false, + "boundElements": null, + "updated": 1766372425873, + "link": null, + "locked": false, + "points": [ + [ + 0, + 0 + ], + [ + 29.666656494140625, + 0 + ], + [ + 29.666656494140625, + -24.66668701171875 + ], + [ + 59.33331298828125, + -24.66668701171875 + ] + ], + "lastCommittedPoint": null, + "startBinding": null, + "endBinding": null, + "startArrowhead": null, + "endArrowhead": "arrow", + "elbowed": true, + "fixedSegments": null, + "startIsSpecial": null, + "endIsSpecial": null + }, + { + "id": "sy0AQFPiKMUpC_fNA1u5Y", + "type": "text", + "x": 653.8335380554199, + "y": -102.99992370605469, + "width": 312.499755859375, + "height": 25, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b09", + "roundness": null, + "seed": 516560529, + "version": 57, + "versionNonce": 1162137087, + "isDeleted": false, + "boundElements": null, + "updated": 1766372458567, + "link": null, + "locked": false, + "text": "ray job submit / drivier - worker", + "fontSize": 20, + "fontFamily": 5, + "textAlign": "left", + "verticalAlign": "top", + "containerId": null, + "originalText": "ray job submit / drivier - worker", + "autoResize": true, + "lineHeight": 1.25 + }, + { + "id": "e58le7CfnSkWVb-LoMmj4", + "type": "ellipse", + "x": 605.1671257019043, + "y": 146.33338928222656, + "width": 38, + "height": 38, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "#99e9f2", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0C", + "roundness": { + "type": 2 + }, + "seed": 915719889, + "version": 349, + "versionNonce": 1238458833, + "isDeleted": false, + "boundElements": [ + { + "id": "bAFwE3wo46b9BFt_2Gnm2", + "type": "arrow" + }, + { + "id": "qcyVAzD12V9EAxE15fwAq", + "type": "arrow" + }, + { + "id": "QgI8Gn67BTTI1MdOhOjjn", + "type": "arrow" + }, + { + "id": "cqQ2Ij98wYdbKqXQcPhxe", + "type": "arrow" + } + ], + "updated": 1766373712585, + "link": null, + "locked": false + }, + { + "id": "eeIv7mr_n3Lg6B3DlXJHR", + "type": "ellipse", + "x": 873.5002861022949, + "y": 67.33341979980469, + "width": 38, + "height": 38, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "#ffc9c9", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0D", + "roundness": { + "type": 2 + }, + "seed": 1674226303, + "version": 428, + "versionNonce": 1555106719, + "isDeleted": false, + "boundElements": [ + { + "id": "Qzy3gfKzdZyrTwzOpuh86", + "type": "arrow" + } + ], + "updated": 1766372702159, + "link": null, + "locked": false + }, + { + "id": "c4sciwjA9ZH_GjmspcBan", + "type": "ellipse", + "x": 878.8335990905762, + "y": 294.6667938232422, + "width": 38, + "height": 38, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "#ffc9c9", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0E", + "roundness": { + "type": 2 + }, + "seed": 2058462079, + "version": 476, + "versionNonce": 803991519, + "isDeleted": false, + "boundElements": [ + { + "id": "zj5n4D3014Kl-BdrNuRZp", + "type": "arrow" + } + ], + "updated": 1766372702159, + "link": null, + "locked": false + }, + { + "id": "QQdAB0FVqHTGhVCKstHrm", + "type": "ellipse", + "x": 308.1669120788574, + "y": -185.9999237060547, + "width": 38, + "height": 38, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "#99e9f2", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0F", + "roundness": { + "type": 2 + }, + "seed": 156527057, + "version": 305, + "versionNonce": 1422415839, + "isDeleted": false, + "boundElements": [], + "updated": 1766372581044, + "link": null, + "locked": false + }, + { + "id": "37ojV_5RQu-H5_oLEw_CC", + "type": "text", + "x": 375.1668510437012, + "y": -176.9999237060547, + "width": 136.61990356445312, + "height": 25, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "#ffc9c9", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0G", + "roundness": null, + "seed": 122498079, + "version": 15, + "versionNonce": 121837457, + "isDeleted": false, + "boundElements": null, + "updated": 1766372589741, + "link": null, + "locked": false, + "text": "ray head node", + "fontSize": 20, + "fontFamily": 5, + "textAlign": "left", + "verticalAlign": "top", + "containerId": null, + "originalText": "ray head node", + "autoResize": true, + "lineHeight": 1.25 + }, + { + "id": "ZyQitTq4F7KArc1LsSU0L", + "type": "ellipse", + "x": 578.1668510437012, + "y": -185.33323669433594, + "width": 38, + "height": 38, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "#ffc9c9", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0H", + "roundness": { + "type": 2 + }, + "seed": 666768689, + "version": 440, + "versionNonce": 1110150815, + "isDeleted": false, + "boundElements": [], + "updated": 1766372594958, + "link": null, + "locked": false + }, + { + "id": "B_AKAU5jOYUjwmhK54Ddr", + "type": "text", + "x": 657.5235862731934, + "y": -178.16656494140625, + "width": 154.65988159179688, + "height": 25, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "#ffc9c9", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0I", + "roundness": null, + "seed": 1500425873, + "version": 72, + "versionNonce": 150248991, + "isDeleted": false, + "boundElements": [], + "updated": 1766372606781, + "link": null, + "locked": false, + "text": "ray worker node", + "fontSize": 20, + "fontFamily": 5, + "textAlign": "left", + "verticalAlign": "top", + "containerId": null, + "originalText": "ray worker node", + "autoResize": true, + "lineHeight": 1.25 + }, + { + "id": "WPOvDTFRcRTqysrgfilw_", + "type": "rectangle", + "x": 453.1667900085449, + "y": 395.0000762939453, + "width": 572.6666870117189, + "height": 251.3333740234375, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0K", + "roundness": null, + "seed": 62506143, + "version": 121, + "versionNonce": 1380555377, + "isDeleted": false, + "boundElements": null, + "updated": 1766373299556, + "link": null, + "locked": false + }, + { + "id": "icDpH3wy-c2_99TXtmru5", + "type": "rectangle", + "x": 469.1667900085449, + "y": 410.87457554413584, + "width": 83.33331298828125, + "height": 138.7922487966845, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0L", + "roundness": null, + "seed": 292536607, + "version": 130, + "versionNonce": 398901169, + "isDeleted": false, + "boundElements": [ + { + "type": "text", + "id": "oYnPwXC5ulNTf8ayVoUHX" + } + ], + "updated": 1766376371670, + "link": null, + "locked": false + }, + { + "id": "oYnPwXC5ulNTf8ayVoUHX", + "type": "text", + "x": 476.93347549438477, + "y": 455.2706999424781, + "width": 67.79994201660156, + "height": 50, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0M", + "roundness": null, + "seed": 964724095, + "version": 75, + "versionNonce": 1019940831, + "isDeleted": false, + "boundElements": null, + "updated": 1766376371670, + "link": null, + "locked": false, + "text": "datase\nts", + "fontSize": 20, + "fontFamily": 5, + "textAlign": "center", + "verticalAlign": "middle", + "containerId": "icDpH3wy-c2_99TXtmru5", + "originalText": "datasets", + "autoResize": true, + "lineHeight": 1.25 + }, + { + "id": "s3rGKdwzC87Z1v9gN0N-5", + "type": "rectangle", + "x": 582.166820526123, + "y": 410.4639351318977, + "width": 83.33331298828125, + "height": 138.7922487966845, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0N", + "roundness": null, + "seed": 401691633, + "version": 190, + "versionNonce": 700169215, + "isDeleted": false, + "boundElements": [ + { + "type": "text", + "id": "D2W7CAX99N7IpvqJQP6eX" + } + ], + "updated": 1766376371670, + "link": null, + "locked": false + }, + { + "id": "D2W7CAX99N7IpvqJQP6eX", + "type": "text", + "x": 613.1934852600098, + "y": 467.36005953023994, + "width": 21.279983520507812, + "height": 25, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0O", + "roundness": null, + "seed": 325299665, + "version": 138, + "versionNonce": 770525041, + "isDeleted": false, + "boundElements": [], + "updated": 1766376371670, + "link": null, + "locked": false, + "text": "hf", + "fontSize": 20, + "fontFamily": 5, + "textAlign": "center", + "verticalAlign": "middle", + "containerId": "s3rGKdwzC87Z1v9gN0N-5", + "originalText": "hf", + "autoResize": true, + "lineHeight": 1.25 + }, + { + "id": "EGfEHSzmWYNx5ec9bP-9v", + "type": "rectangle", + "x": 701.5001945495605, + "y": 409.6427294956321, + "width": 83.33331298828125, + "height": 138.7922487966845, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0P", + "roundness": null, + "seed": 974114385, + "version": 214, + "versionNonce": 2090152273, + "isDeleted": false, + "boundElements": [ + { + "type": "text", + "id": "XIfw_ABOr1dlUCUXEBJqA" + } + ], + "updated": 1766376371670, + "link": null, + "locked": false + }, + { + "id": "XIfw_ABOr1dlUCUXEBJqA", + "type": "text", + "x": 722.9068641662598, + "y": 466.53885389397436, + "width": 40.51997375488281, + "height": 25, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0Q", + "roundness": null, + "seed": 881056817, + "version": 164, + "versionNonce": 663619647, + "isDeleted": false, + "boundElements": [], + "updated": 1766376371670, + "link": null, + "locked": false, + "text": "jobs", + "fontSize": 20, + "fontFamily": 5, + "textAlign": "center", + "verticalAlign": "middle", + "containerId": "EGfEHSzmWYNx5ec9bP-9v", + "originalText": "jobs", + "autoResize": true, + "lineHeight": 1.25 + }, + { + "id": "Fe9YUb1NNu32IhvievWnr", + "type": "rectangle", + "x": 812.8334465026855, + "y": 408.8214486711559, + "width": 83.33331298828125, + "height": 138.7922487966845, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0R", + "roundness": null, + "seed": 66661759, + "version": 268, + "versionNonce": 1178642527, + "isDeleted": false, + "boundElements": [ + { + "type": "text", + "id": "UZhPJS_QBKW_LMYwu1enR" + } + ], + "updated": 1766376371670, + "link": null, + "locked": false + }, + { + "id": "UZhPJS_QBKW_LMYwu1enR", + "type": "text", + "x": 821.1101188659668, + "y": 465.71757306949814, + "width": 66.77996826171875, + "height": 25, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0S", + "roundness": null, + "seed": 1284842911, + "version": 227, + "versionNonce": 969134353, + "isDeleted": false, + "boundElements": [], + "updated": 1766376371670, + "link": null, + "locked": false, + "text": "output", + "fontSize": 20, + "fontFamily": 5, + "textAlign": "center", + "verticalAlign": "middle", + "containerId": "Fe9YUb1NNu32IhvievWnr", + "originalText": "output", + "autoResize": true, + "lineHeight": 1.25 + }, + { + "id": "r8maqUMI8CEoTqgUHwtnZ", + "type": "rectangle", + "x": 924.8335075378418, + "y": 408.0001678466797, + "width": 83.33331298828125, + "height": 138.7922487966845, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0T", + "roundness": null, + "seed": 1203268831, + "version": 320, + "versionNonce": 204048113, + "isDeleted": false, + "boundElements": [ + { + "type": "text", + "id": "nL0ZOKRltZyanG1aa7hOW" + } + ], + "updated": 1766376371670, + "link": null, + "locked": false + }, + { + "id": "nL0ZOKRltZyanG1aa7hOW", + "type": "text", + "x": 951.7201805114746, + "y": 464.8962922450219, + "width": 29.559967041015625, + "height": 25, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0U", + "roundness": null, + "seed": 150305023, + "version": 273, + "versionNonce": 1246452895, + "isDeleted": false, + "boundElements": [], + "updated": 1766376371670, + "link": null, + "locked": false, + "text": "ray", + "fontSize": 20, + "fontFamily": 5, + "textAlign": "center", + "verticalAlign": "middle", + "containerId": "r8maqUMI8CEoTqgUHwtnZ", + "originalText": "ray", + "autoResize": true, + "lineHeight": 1.25 + }, + { + "id": "OmfvpsuW0dhubnI5NH1Ni", + "type": "text", + "x": 675.8334770202637, + "y": 593.0000762939453, + "width": 121.67990112304688, + "height": 25, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0V", + "roundness": null, + "seed": 1166095967, + "version": 60, + "versionNonce": 643125617, + "isDeleted": false, + "boundElements": null, + "updated": 1766373325167, + "link": null, + "locked": false, + "text": "/mnt/shared", + "fontSize": 20, + "fontFamily": 5, + "textAlign": "left", + "verticalAlign": "top", + "containerId": null, + "originalText": "/mnt/shared", + "autoResize": true, + "lineHeight": 1.25 + }, + { + "id": "cqQ2Ij98wYdbKqXQcPhxe", + "type": "arrow", + "x": 371.9714008624129, + "y": 208.42023305299114, + "width": 252.09572483949137, + "height": 50.3333740234375, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0W", + "roundness": null, + "seed": 1948514207, + "version": 127, + "versionNonce": 977608497, + "isDeleted": false, + "boundElements": null, + "updated": 1766373713002, + "link": null, + "locked": false, + "points": [ + [ + 0, + 0 + ], + [ + 96.77889477662518, + 0 + ], + [ + 96.77889477662518, + 31.246530252672926 + ], + [ + 252.09572483949137, + 31.246530252672926 + ], + [ + 252.09572483949137, + -19.086843770764574 + ] + ], + "lastCommittedPoint": null, + "startBinding": { + "elementId": "9Ql3xet2wP4Oy9RJ4s-8H", + "fixedPoint": [ + 1.0246502128937116, + 0.5789413922556957 + ], + "focus": 0, + "gap": 0 + }, + "endBinding": { + "elementId": "e58le7CfnSkWVb-LoMmj4", + "fixedPoint": [ + 0.49736842105263096, + 1.131578947368421 + ], + "focus": 0, + "gap": 0 + }, + "startArrowhead": null, + "endArrowhead": "arrow", + "elbowed": true, + "fixedSegments": [ + { + "index": 3, + "start": [ + 96.77889477662518, + 31.246530252672926 + ], + "end": [ + 252.09572483949137, + 31.246530252672926 + ] + } + ], + "startIsSpecial": false, + "endIsSpecial": false + }, + { + "id": "xn_M6D4MoCOvg2I2eSA3o", + "type": "text", + "x": 491.16682052612305, + "y": 237.66676330566406, + "width": 137.33990478515625, + "height": 25, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0X", + "roundness": null, + "seed": 485981873, + "version": 51, + "versionNonce": 224706367, + "isDeleted": false, + "boundElements": null, + "updated": 1766373736215, + "link": null, + "locked": false, + "text": "ray job submit", + "fontSize": 20, + "fontFamily": 5, + "textAlign": "left", + "verticalAlign": "top", + "containerId": null, + "originalText": "ray job submit", + "autoResize": true, + "lineHeight": 1.25 + }, + { + "id": "n0XX1HzCpzHfeo4hTRson", + "type": "text", + "x": 456.5001640319824, + "y": 281.00013732910156, + "width": 196.61985778808594, + "height": 25, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0Y", + "roundness": null, + "seed": 960522577, + "version": 46, + "versionNonce": 1758456319, + "isDeleted": false, + "boundElements": null, + "updated": 1766373758962, + "link": null, + "locked": false, + "text": "head node container", + "fontSize": 20, + "fontFamily": 5, + "textAlign": "left", + "verticalAlign": "top", + "containerId": null, + "originalText": "head node container", + "autoResize": true, + "lineHeight": 1.25 + }, + { + "id": "-rnnyMPU-tQs8UVNBCb7C", + "type": "text", + "x": 792.5001945495605, + "y": 176.33338928222656, + "width": 225.5198211669922, + "height": 50, + "angle": 0, + "strokeColor": "#1e1e1e", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0Z", + "roundness": null, + "seed": 501756561, + "version": 78, + "versionNonce": 1893991729, + "isDeleted": false, + "boundElements": null, + "updated": 1766373794871, + "link": null, + "locked": false, + "text": "worker node containers\n 4 x H20", + "fontSize": 20, + "fontFamily": 5, + "textAlign": "left", + "verticalAlign": "top", + "containerId": null, + "originalText": "worker node containers\n 4 x H20", + "autoResize": true, + "lineHeight": 1.25 + }, + { + "id": "qWTdVy6o3A4fkZsGRMCKB", + "type": "text", + "x": 483.4334098815918, + "y": 557.5000762939453, + "width": 40.1335388183593, + "height": 16.722307840983046, + "angle": 0, + "strokeColor": "#e03131", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0a", + "roundness": null, + "seed": 438165041, + "version": 42, + "versionNonce": 1297545873, + "isDeleted": false, + "boundElements": null, + "updated": 1766376434059, + "link": null, + "locked": false, + "text": "数据集", + "fontSize": 13.377846272786437, + "fontFamily": 5, + "textAlign": "left", + "verticalAlign": "top", + "containerId": null, + "originalText": "数据集", + "autoResize": true, + "lineHeight": 1.25 + }, + { + "id": "Zr04JN1SwCMGrPB7Otn5n", + "type": "text", + "x": 599.7667686462403, + "y": 556.6389223734539, + "width": 53.47998046875, + "height": 16.722307840983046, + "angle": 0, + "strokeColor": "#e03131", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0b", + "roundness": null, + "seed": 1427393105, + "version": 108, + "versionNonce": 20002801, + "isDeleted": false, + "boundElements": [], + "updated": 1766376447293, + "link": null, + "locked": false, + "text": "基座模型", + "fontSize": 13.377846272786437, + "fontFamily": 5, + "textAlign": "left", + "verticalAlign": "top", + "containerId": null, + "originalText": "基座模型", + "autoResize": true, + "lineHeight": 1.25 + }, + { + "id": "U5mN9zuDxufwaTsXAez1v", + "type": "text", + "x": 683.1001121520997, + "y": 553.9721743265789, + "width": 116.947265625, + "height": 33.44461568196609, + "angle": 0, + "strokeColor": "#e03131", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0c", + "roundness": null, + "seed": 1016566161, + "version": 228, + "versionNonce": 460195775, + "isDeleted": false, + "boundElements": [], + "updated": 1766376548390, + "link": null, + "locked": false, + "text": "job级别\nckpt, config, log等", + "fontSize": 13.377846272786437, + "fontFamily": 5, + "textAlign": "center", + "verticalAlign": "top", + "containerId": null, + "originalText": "job级别\nckpt, config, log等", + "autoResize": true, + "lineHeight": 1.25 + }, + { + "id": "kyEqz0rZKz404SUTYZRai", + "type": "text", + "x": 931.766707611084, + "y": 557.3056093851726, + "width": 74.01625061035156, + "height": 33.44461568196609, + "angle": 0, + "strokeColor": "#e03131", + "backgroundColor": "transparent", + "fillStyle": "solid", + "strokeWidth": 2, + "strokeStyle": "solid", + "roughness": 1, + "opacity": 100, + "groupIds": [], + "frameId": null, + "index": "b0d", + "roundness": null, + "seed": 1742834719, + "version": 152, + "versionNonce": 2136750623, + "isDeleted": false, + "boundElements": [], + "updated": 1766376569303, + "link": null, + "locked": false, + "text": "系统级\nsession日志", + "fontSize": 13.377846272786437, + "fontFamily": 5, + "textAlign": "center", + "verticalAlign": "top", + "containerId": null, + "originalText": "系统级\nsession日志", + "autoResize": true, + "lineHeight": 1.25 + } + ], + "appState": { + "gridSize": 20, + "gridStep": 5, + "gridModeEnabled": false, + "viewBackgroundColor": "#ffffff" + }, + "files": {} +} \ No newline at end of file