| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
chore: enforce license (#1171) | 4 个月前 | |
fix: add missing pre-commit check file (#1173) | 4 个月前 | |
chore(deps): upgrade runtime dependencies and CI workflow (#1206) * chore(deps): upgrade runtime dependencies and CI workflow Upgrade megatron-core, sglang, vllm (0.19.1), transformers, and related packages. Pin deepep/deepgemm commits and lock trackio version for reproducibility. Key changes: - Upgrade sglang, vllm, transformers, megatron-core versions - Pin vllm to 0.19.1, fix compatibility across backends - Optimize Dockerfile and slim Docker image - Add uv_sync.sh install script, fix uv install on Linux - Fix Archon Qwen3.5 precision and port range issues - Remove integration tests from PR CI (moved to nightly) - Sync GRPO integration test config Refs: upgrade-deps branch * chore: remove duplicated test | 4 个月前 | |
gov: enforce 2-approval merge policy on main (#1307) * gov: enforce 2-approval merge policy on main Configure repository governance so that pull requests targeting main require two maintainer approvals (or one administrator bypass), with code-owner review and signed commits enforced via GitHub branch protection. Key changes: - Add .github/ruleset.json defining the active branch protection: 2 approving reviews, code-owner review, last-push approval, linear history, signed commits, pre-commit status check, admin bypass via PR only. - Reshuffle .github/CODEOWNERS so every path has at least two owners — the rule degrades gracefully when a single owner is unavailable. - Reconcile GOVERNANCE.md with the mechanical enforcement: 2 maintainer approvals required for all PRs; the lead maintainer may bypass for trivial or hotfix changes with disclosure in the PR description. - Add areal/tools/format_codeowners.py and a corresponding pre-commit hook that aligns owner columns, strips trailing whitespace, validates owner syntax, errors on duplicate paths, and warns on single-owner rules. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * gov: remove required_signatures rule from main branch ruleset Commit signature verification is no longer enforced, allowing contributors without GPG/SSH signing setup to merge into main. --------- Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com> | 4 个月前 | |
perf: reduce Megatron training memory peaks (#1555) * perf: reduce Megatron training memory peaks Add an SFT profiling workflow and use its memory snapshots to remove full-sequence vocabulary and optimizer gradient peaks from Megatron training. Key changes: - Add rank-aware kernel and memory profiling for packed SFT workloads - Fuse FP32 vocab-parallel logprob storage with LM head backward - Add optional true chunked LM head loss with recomputed backward - Configure precision-aware optimizer fields before Megatron validation - Cover BF16/FP32 numerical parity and distributed TP/SP behavior * fix(models): avoid private storage identity checks Track the LM head output tensor weakly and compare storage through the public data_ptr API. This preserves allocator-address reuse protection without depending on PyTorch's private storage _cdata field. * test: compare parameter storage without object identity Parameter.data may return a fresh Tensor wrapper on each access. Verify that replicated parameters retain their data pointer and storage offset instead of comparing transient Python objects. * test: make recycled CUDA storage check deterministic Construct the replacement tensor from the original storage instead of relying on the caching allocator to immediately reuse a freed address after the full CI suite. * fix(engine): guard AReaL LM Head storage reuse Keep entropy differentiable unless its gradients are disabled, and make the optimized LM Head path opt-in. Warn when destructive storage reuse makes entropy non-differentiable, while rejecting unsupported NPU and tree-training combinations. * fix(engine): export standard FSDP LoRA adapter keys PEFT keeps the adapter name in live parameter FQNs, while serving engines expect serialized LoRA keys without it. Normalize per-parameter FSDP exports and align the SGLang best-effort assertion with its load behavior. * feat(engine): support chunked logits for padded models Enable chunked LM Head loss for text-only padded BSHD models such as Qwen3.5 and rename the public toggle to enable_chunked_logits so the configuration reflects its behavior. Key changes: - Add padded label construction and output repacking - Add Qwen3.5 and updated Qwen3 MoE profile recipes - Update CLI docs, validation, and regression coverage * fix(engine): configure logprob chunking explicitly Replace the profile-only environment override with a validated train-engine option so FSDP, Megatron, Archon, and tree paths use the same explicit value. Key changes: - add and document TrainEngineConfig.logprobs_chunk_size - pass the setting through every engine logprob path - translate the profile guide and remove out-of-scope FSDP LoRA changes - add config, launcher, and explicit chunk-size tests Refs: #1555 | 1 个月前 | |
chore: enforce license (#1171) | 4 个月前 | |
chore: enforce license (#1171) | 4 个月前 | |
chore: enforce license (#1171) | 4 个月前 | |
chore: enforce license (#1171) | 4 个月前 | |
chore(deps): upgrade runtime dependencies and CI workflow (#1206) * chore(deps): upgrade runtime dependencies and CI workflow Upgrade megatron-core, sglang, vllm (0.19.1), transformers, and related packages. Pin deepep/deepgemm commits and lock trackio version for reproducibility. Key changes: - Upgrade sglang, vllm, transformers, megatron-core versions - Pin vllm to 0.19.1, fix compatibility across backends - Optimize Dockerfile and slim Docker image - Add uv_sync.sh install script, fix uv install on Linux - Fix Archon Qwen3.5 precision and port range issues - Remove integration tests from PR CI (moved to nightly) - Sync GRPO integration test config Refs: upgrade-deps branch * chore: remove duplicated test | 4 个月前 | |
chore: enforce license (#1171) | 4 个月前 | |
chore: enforce license (#1171) | 4 个月前 |
| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
| 4 个月前 | ||
| 4 个月前 | ||
| 4 个月前 | ||
| 4 个月前 | ||
| 1 个月前 | ||
| 4 个月前 | ||
| 4 个月前 | ||
| 4 个月前 | ||
| 4 个月前 | ||
| 4 个月前 | ||
| 4 个月前 | ||
| 4 个月前 |