mirror of
https://github.com/NVIDIA/Model-Optimizer.git
synced 2026-10-19 05:49:19 +08:00
0.45.0dev
68
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 Chenhan D. YuandClaude Opus 4.6
|
355c6b7883 |
fix: PTQ 1GPU, export PP divisibility, hidden states conversations key (#1293)
## Summary - **megatron_lm_ptq.yaml**: Qwen3-8B PTQ to single GPU for L40 clusters (TP=1, all tasks) - **quantize.sh**: Auto-find largest PP dividing model's `num_hidden_layers` for export step. Qwen3-8B has 36 layers which isn't divisible by 8, causing `AssertionError` on 8-GPU nodes - **compute_hidden_states_trtllm.py**: Use `messages` with `conversations` fallback, matching the HF version. Fixes `KeyError: 'conversations'` when data uses OpenAI `messages` format ## Test plan - [x] Qwen3-8B PTQ runs on single L40 GPU - [x] Export PP auto-selects valid divisor (36 layers → PP=6 on 8 GPUs, PP=4 on 4 GPUs, PP=1 on 1 GPU) - [x] EAGLE3 offline pipeline reads data with `messages` field 🤖 Generated with [Claude Code](https://claude.com/claude-code) <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Dataset input handling now supports multiple field formats for enhanced compatibility. * **Bug Fixes** * Optimized GPU resource allocation during model quantization with improved pipeline parallelism computation. * Updated quantization configuration for more efficient resource utilization. <!-- end of auto-generated comment: release notes by coderabbit.ai --> Signed-off-by: Chenhan Yu <chenhany@nvidia.com> Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> |
||
|
yeyu-nvidiaandClaude Opus 4.6
|
07ae8e7128 |
Add LoRA co-training support for HF EAGLE speculative decoding (#1060)
### What does this PR do?
Type of change: New feature + bug fixes
Adds **LoRA co-training** support for HF EAGLE speculative decoding.
When `eagle_base_lora=True`, HF PEFT LoRA adapters are injected into the
base model and co-trained alongside the EAGLE draft module in a single
online training pass. A preservation loss (KL divergence between the
original frozen base model output and the LoRA-adapted output) prevents
base model drift. LoRA adapter weights are exported in standard peft
format alongside EAGLE draft artifacts.
### Key features
- **LoRA injection**: `peft.inject_adapter_in_model` applied in-place
(no wrapper), keeping the existing `HFEagleModel` structure intact.
- **Preservation loss**: Cross-entropy `H(ref, lora)` — equivalent
gradient to `KL(ref || lora)` since `H(ref)` is constant w.r.t. LoRA
params.
- **Warmup schedule**: `eagle_base_lora_warmup_steps` freezes LoRA for N
steps while the EAGLE head stabilizes, then enables co-training via a
`LoRAWarmupCallback`.
- **Logits detach regularization**: `eagle_base_lora_logits_detach_prob`
stochastically detaches base logits from the EAGLE loss path, preventing
LoRA from degenerating to maximize EAGLE accuracy at the cost of base
model quality.
- **Export**: Standard peft format (`adapter_model.safetensors` +
`adapter_config.json`) alongside EAGLE draft model.
- **Merge script**: `scripts/merge_lora.py` merges LoRA weights into the
base model and restores the original `config.json` (avoids transformers
5.x rewriting `rope_theta` → `rope_parameters` which breaks
vLLM/TRT-LLM).
- **Multinode fix**: `dp_shard_size` now uses `WORLD_SIZE` instead of
local GPU count.
### Config options
```python
mtsp.convert(model, mode=[("eagle", {
"eagle_base_lora": True, # enable LoRA co-training
"eagle_base_lora_rank": 64, # LoRA rank
"eagle_base_lora_alpha": 16.0, # LoRA scaling
"eagle_base_lora_target_modules": ["q_proj", "k_proj", "v_proj", "o_proj"],
"eagle_base_lora_preservation_loss_weight": 0.1, # preservation loss weight
"eagle_base_lora_warmup_steps": 0, # freeze LoRA for N steps
"eagle_base_lora_logits_detach_prob": 0.5, # detach prob (0=never, 1=always)
})])
```
### Experimental results (Qwen3-8B, checkpoint-60000)
Base model quality preserved across detach_prob sweep (lm_eval: IFEval,
ARC-C, Winogrande — results pending final collection).
**Acceptance rate** (mt_bench, draft_length=3, output_length=4096,
temperature=0):
| detach_prob | vLLM AR | TRT-LLM AR |
|---|---|---|
| baseline (no LoRA) | 2.14 | 2.15 |
| 0.5 | 1.45 | 1.44 |
| 0.8 | **3.06** | **3.01** |
| 0.85 | 2.90 | 2.90 |
| 0.9 | 2.76 | 2.77 |
| 0.95 | 2.51 | 2.58 |
| 0.99 | 2.37 | 2.37 |
| 0.999 | 2.30 | 2.27 |
| 0.9999 | 2.31 | 2.26 |
Best AR at `detach_prob=0.8`: ~40% improvement over baseline.
### Testing
`tests/unit/torch/speculative/plugins/test_hf_speculative_lora.py` (5
tests):
- `test_lora_layers_injected` — LoRA layers present after conversion
- `test_trainable_params` — only `lora_*` and `eagle_module` params are
trainable
- `test_forward_returns_loss` — forward returns non-zero scalar loss
- `test_eagle_offline_incompatible` — `eagle_base_lora=True` +
`eagle_offline=True` raises `ValueError`
- `test_export_lora_artifacts` — export produces standard peft adapter
files
### Bug fixes (included in this PR)
1. **`launch_train.sh` case pattern ordering**: glob
`--eagle_base_lora*` was before specific patterns
(`--eagle_base_lora_rank*`, etc.), silently swallowing LoRA args.
2. **LoRA optimizer exclusion during warmup**: warmup freezing excluded
LoRA from the optimizer entirely; fixed with `add_param_group` in the
callback.
3. **`merge_lora.py` config.json**: `save_pretrained()` with
transformers >=5.x rewrites `rope_theta` → `rope_parameters`, breaking
vLLM positional embeddings. Fixed by copying the original base model
config.
4. **Multinode `dp_shard_size`**: used local GPU count instead of
`WORLD_SIZE`.
### Checklist
- [x] Backward compatible (all new config fields have defaults)
- [x] Uses `peft` via lazy imports (no hard dependency)
- [x] Unit tests added
- [x] Online HF training only (`eagle_offline=True` blocked)
---------
Signed-off-by: Ye Yu <yeyu@nvidia.com>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
|
||
|
Chenhan D. YuandClaude Opus 4.6
|
3131195241 |
add: DFlash block diffusion speculative decoding (#1211)
DFlash (Block Diffusion for Flash Speculative Decoding) predicts an entire block of tokens in a single forward pass using masked parallel prediction with KV injection from the target model's hidden states. Key features: - Feature fusion (multi-layer hidden states -> FC + RMSNorm) - KV injection (fused features as K/V in every draft layer with QK-norm) - Random anchor sampling with bidirectional intra-block attention - Logit distillation with exponential loss decay (gamma weighting) - Multi-node DDP training with checkpoint resume - Export to z-lab compatible HF format - Online validation (context-dependent ground truth) Training recipe: modelopt_recipes/general/speculative_decoding/dflash.yaml Results: examples/speculative_decoding/doc/dflash_results.md ### ModelOpt Eval (online validation, osl=512) | Dataset | z-lab | ModelOpt (306K) | Diff | |---------|-------|-----------------|------| | gsm8k | 4.10 | **5.19** | **+1.09** | | MT-Bench | 3.58 | **4.36** | **+0.78** | ### z-lab Official Eval (dflash.benchmark, osl=512) | Dataset | z-lab | ModelOpt (306K) | Diff | |---------|-------|-----------------|------| | gsm8k | **5.00** | 4.08 | -0.92 | | MT-Bench | **3.28** | 2.99 | -0.29 | > z-lab model trained with block_size=16. ModelOpt trained with block_size=8. ## Evaluation Method Impact (gsm8k) | Eval Method | z-lab checkpoint | ModelOpt (306K) | |-------------|-----------------|-----------------| | Fixed GT (ModelOpt eval) | 2.95 | 4.23 | | Online GT (ModelOpt eval) | 4.10 | **5.19** | | z-lab official eval | **5.00** | 4.08 | ### What does this PR do? Type of change: ? <!-- Use one of the following: Bug fix, new feature, new example, new tests, documentation. --> <!-- Details about the change. --> ### Usage ```python # Add a code snippet demonstrating how to use this ``` ### Testing <!-- Mention how have you tested your change if applicable. --> ### Before your PR is "*Ready for review*" Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md) and your commits are signed (`git commit -s -S`). Make sure you read and follow the [Security Best Practices](https://github.com/NVIDIA/Model-Optimizer/blob/main/SECURITY.md#security-coding-practices-for-contributors) (e.g. avoiding hardcoded `trust_remote_code=True`, `torch.load(..., weights_only=False)`, `pickle`, etc.). - Is this change backward compatible?: ✅ / ❌ / N/A <!--- If ❌, explain why. --> - If you copied code from any other sources or added a new PIP dependency, did you follow guidance in `CONTRIBUTING.md`: ✅ / ❌ / N/A <!--- Mandatory --> - Did you write any new necessary tests?: ✅ / ❌ / N/A <!--- Mandatory for new features or examples. --> - Did you update [Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?: ✅ / ❌ / N/A <!--- Only for new features, API changes, critical bug fixes or backward incompatible changes. --> ### Additional Information <!-- E.g. related issue. --> <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Added DFlash speculative decoding mode with parallel block prediction support. * Included training launchers and MT-Bench evaluation scripts for DFlash models. * Added online acceptance rate validation for improved inference verification. * **Documentation** * DFlash quick start guide with configuration parameters and training examples. * Performance results and benchmarks for DFlash-trained models. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Signed-off-by: Chenhan Yu <chenhany@nvidia.com> Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> |
||
|
h-guo18
|
6403389eb0 |
Feat: Configurable Eagle ROPE scaling during export (#1238)
### What does this PR do? JIRA ticket: https://jirasw.nvidia.com/browse/OMNIML-3469 Type of change: New feature Decouple EAGLE training rope configuration from export rope configuration, enabling separate YaRN rope scaling injection at export time for long-context inference. #### Changes **Configurable export rope scaling (`EagleConfig`)** - Add `eagle_export_rope_scaling` field to `EagleConfig` with default YaRN config (`factor=32.0`, `original_max_position_embeddings=2048`) - Set to `{}` to disable rope scaling injection at export **Simplified training defaults (`default_config.py`)** - Change default training rope from `llama3` (theta=500k) to `default` (theta=10k) — models now train with simple positional embeddings; rope scaling is applied only at export - Add `rope_theta` inside `rope_scaling` dict for transformers 5.x cross-version compatibility **Move config validation/rewriting into `EagleConfig` (`config.py`)** - `_derive_eagle_offline`: derives `eagle_offline` from `data_args.offline_data_path` via validation context, removing manual assignment in `main.py` - `_check_rope_scaling_consistency`: rejects configs where `eagle_export_rope_scaling` is set but training `rope_type` is not `"default"` - `_warn_rope_vs_training_seq_len`: warns when `original_max_position_embeddings` differs from `training_seq_len` **Export rope injection (`hf_spec_export.py`)** - Inject `eagle_export_rope_scaling` into the exported HF config when training rope_type is `"default"` - Fall back `rope_theta` from `rope_scaling` dict for transformers 5.x compatibility **Fix Megatron RotaryEmbedding crash (`megatron_eagle.py`)** - `dict_to_config()` set `rope_scaling=True` whenever the `rope_scaling` key existed, even without a `"factor"` — causing `RotaryEmbedding` to divide by `None` - Now only enables `rope_scaling` when the dict actually contains a `"factor"` key ### Usage Configure in YAML config (or use defaults from `eagle3.yaml`): ```yaml eagle: eagle_export_rope_scaling: rope_type: yarn factor: 32.0 original_max_position_embeddings: 2048 ``` Set to empty dict to disable export rope injection: ```yaml eagle: eagle_export_rope_scaling: {} ``` ### Testing - New unit tests: `tests/unit/torch/speculative/test_eagle_config.py` — rope consistency validator, seq_len warning, context-derived `eagle_offline` - New unit tests: `tests/unit/torch/export/test_hf_spec_rope_export.py` — export rope injection, fallback, and empty-config cases ### Before your PR is "*Ready for review*" Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md) and your commits are signed (`git commit -s -S`). Make sure you read and follow the [Security Best Practices](https://github.com/NVIDIA/Model-Optimizer/blob/main/SECURITY.md#security-coding-practices-for-contributors) (e.g. avoiding hardcoded `trust_remote_code=True`, `torch.load(..., weights_only=False)`, `pickle`, etc.). - Is this change backward compatible?: ✅ (new field has sensible default; existing configs work unchanged) - If you copied code from any other sources or added a new PIP dependency, did you follow guidance in `CONTRIBUTING.md`: N/A - Did you write any new necessary tests?: ✅ - Did you update [Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?: ❌ (should be added if merging as a feature) <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Add export-time rope-scaling configuration for EAGLE models. * **Improvements** * Stronger validation and context-aware reconciliation between training and export configs. * Export now injects rope-scaling and rope-theta when appropriate. * Default rope-scaling values updated for EAGLE variants. * Model instances now expose export rope-scaling for downstream use. * **Tests** * Added unit tests covering rope-scaling export behavior and configuration validators. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com> |
||
|
yeyu-nvidiaandClaude Sonnet 4.6
|
9050188034 |
Fix test_collect_hidden_states: use synthetic short conversations (#1234)
## Summary - `test_collect_hidden_states` was using real daring-anteater conversations (typically 1000+ tokens) but the tiny test model has `max_position_embeddings=32`. Both sampled conversations exceeded the default `--max-seq-len 3072` filter, producing zero `.pt` files and failing the assertion. - Added a `tiny_conversations_path` fixture with synthetic short single-turn conversations that tokenize within `max_position_embeddings=32`. - Changed `test_collect_hidden_states` to use this fixture with `--max-seq-len 32`. - Added a `None` guard for `tokenizer.chat_template.replace(...)` to avoid `AttributeError` when the tokenizer has no chat template. ## Test plan - [ ] `pytest tests/examples/speculative_decoding/test_eagle_offline_ptq.py::test_collect_hidden_states` passes - [ ] CI `speculative_decoding` job passes 🤖 Generated with [Claude Code](https://claude.com/claude-code) <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Bug Fixes** * Resolved compatibility issues when tokenizers do not have a chat template configuration by adding proper error handling. * Standardized tokenization input extraction logic across different transformer library versions for consistent behavior. * **Tests** * Enhanced test infrastructure with new conversation data fixtures and improved sequence length validation for speculative decoding examples. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Signed-off-by: Ye Yu <yeyu@nvidia.com> Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com> |
||
|
yeyu-nvidiaandClaude Sonnet 4.6
|
c901814ff9 |
Fix compute_hidden_states_hf.py: handle BatchEncoding from apply_chat_template (#1225)
## Summary
- `apply_chat_template(..., return_tensors="pt")` returns a
`BatchEncoding` in transformers 4.46+, which no longer subclasses `dict`
- The old guard `isinstance(tokenized, dict)` evaluates to `False` for
`BatchEncoding`, so `input_ids` was set to the whole `BatchEncoding`
object
- Calling `.shape[1]` on a `BatchEncoding` triggers
`__getattr__("shape")` → `AttributeError`
- Fix: check `isinstance(tokenized, torch.Tensor)` instead, which
correctly handles both old transformers (plain tensor) and new
transformers (BatchEncoding)
This is causing `test_collect_hidden_states` to fail in the speculative
decoding CI for all open PRs (#1207, #1210, #1221).
## Test plan
- [ ] `torch-pr (speculative_decoding, 26.01)` CI passes
- [ ] Verify fix handles both `torch.Tensor` return (old transformers)
and `BatchEncoding` return (new transformers 4.46+)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Signed-off-by: Ye Yu <yeyu@nvidia.com>
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
|
||
|
Keval Morabia
|
04cd596d79 |
Add experimental support for transformers>=5.0 + min torch 2.8 (#975)
### What does this PR do? - Add experimental support for transformers >=5.0 and remove deprecated usages: https://github.com/huggingface/transformers/blob/main/MIGRATION_GUIDE_V5.md - ⚠️ For accelerate examples that used `--warmup-ratio: float` (deprecated in 5.x), we now change it to `--warmup-steps: float | int` which works as ratio if float but only for 5.x. For 4.x, it will error out if float and prompt user to change back to `--warmup-ratio` or pass an int absolute step count. - ⚠️ Unified Hugging Face checkpoint export for quantized checkpoints may not work for some models with transformers>=5.0 yet as it requires a lot of fixes (e.g. change in how MoE experts are organized) - ~Add Workaround for TRT-LLM's import of deprecated transformers functions so trt-llm based gpu unit tests work fine. Still deployment for models needs proper fixes directly in TRT-LLM hence llm/vlm ptq example tests still run with transformers 4.57~ - Everything except PTQ and Export (mainly MoE) should work fine with transformers>=5.0 - Bump min torch to 2.8 and enable 2.11 cicd testing - NOTE: Upcoming Nemo:26.04 container comes with transformers 5.3 ### Testing <!-- Mention how have you tested your change if applicable. --> - [x] CI/CD tests passing - [x] Manually tested unit tests, gpu tests with transformers 4.56 and 5.4 - [x] Manually tested example tests (except trt-llm container tests) with transformers 4.56 and 5.4 - [x] 2-gpu nightly CICD tests manually triggered and passing: [gpu tests](https://github.com/NVIDIA/Model-Optimizer/actions/runs/23867257540), [example tests](https://github.com/NVIDIA/Model-Optimizer/actions/runs/23867260643) ### Before your PR is "*Ready for review*" Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md) and your commits are signed (`git commit -s -S`). Make sure you read and follow the [Security Best Practices](https://github.com/NVIDIA/Model-Optimizer/blob/main/SECURITY.md#security-coding-practices-for-contributors) (e.g. avoiding hardcoded `trust_remote_code=True`, using `torch.load(..., weights_only=True)`, avoiding `pickle`, etc.). - Is this change backward compatible?: ✅ <!--- If ❌, explain why. --> - If you copied code from any other source, did you follow IP policy in [CONTRIBUTING.md](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md#-copying-code-from-other-sources)?: N/A <!--- Mandatory --> - Did you write any new necessary tests?: ✅ <!--- Mandatory for new features or examples. --> - Did you update [Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?: ✅ <!--- Only for new features, API changes, critical bug fixes or backward incompatible changes. --> <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Make remote-code usage opt-in via a configurable --trust_remote_code flag across examples and tools. * **Bug Fixes** * Improve checkpoint/resume detection and related training guidance to avoid erroneous errors. * **Refactor** * Consolidate dtype/config naming, switch warmup settings from ratio → steps, and unify tokenizer invocation patterns. * **Documentation** * Simplify changelog title and add misc notes for release 0.44. * **Chores** * Remove scheduled PR-branch cleanup workflow and relax/remove several transformers version pins. * **Tests** * Adjust test gates, skips, and structures to align with updated deps and behaviors. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Signed-off-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com> |
||
|
yeyu-nvidiaandClaude Sonnet 4.6
|
cccfded8a9 |
Add support for offline speculative decoding model PTQ (#883)
## What does this PR do? **Type of change:** new feature **Overview:** This PR enables loading in a ModelOpt pretrained offline speculative decoding model (e.g., EAGLE3) and performs PTQ on it and export. ## Usage Follow the speculative_decoding examples to train an offline speculative decoding model first. Then follow the command below to quantize and export it: ```bash python hf_ptq.py --pyt_ckpt_path <dir_of_offline_specdec_model> --specdec_offline_dataset <dir_of_dataset> ``` ## Testing <!-- Mention how have you tested your change if applicable. --> ## Before your PR is "*Ready for review*" <!-- If you haven't finished some of the above items you can still open `Draft` PR. --> - **Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md)** and your commits are signed. - **Is this change backward compatible?**: Yes/No <!--- If No, explain why. --> - **Did you write any new necessary tests?**: Yes/No - **Did you add or update any necessary documentation?**: Yes/No - **Did you update [Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?**: Yes/No <!--- Only for new features, API changes, critical bug fixes or bw breaking changes. --> ## Additional Information <!-- E.g. related issue. --> <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Offline speculative decoding workflow: support loading a local dataset for calibration, generation, and export; new CLI option to specify the offline dataset. * **Improvements** * Export and quantization paths now accept and propagate offline speculative-decoding inputs. * Offline data loading honors a sample-size limit and enforces safe batch sizing for calibration. * **Bug Fixes** * Better handling of model/config mismatches and varied batch types in offline flows. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Signed-off-by: Ye Yu <yeyu@nvidia.com> Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com> |
||
|
Keval Morabia
|
ba4f42df1c |
Minor fix for example tests
Signed-off-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com> |
||
|
Keval Morabia
|
ebc534d765 |
Update code-copying guidelines in CONTRIBUTING.md
Signed-off-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com> |
||
|
Chenhan D. Yu
|
0246041b01 |
feat(speculative): add vLLM data synthesis pipeline and Nemotron dataset preparation scripts (#1176)
### What does this PR do? Type of change: New feature, new example, bug fix Adds a vLLM-based synthetic data generation pipeline for speculative decoding draft model training, along with dataset preparation scripts for NVIDIA's Nemotron Post-Training dataset collections. **Data synthesis pipeline** (`tools/launcher/common/vllm/query.sh` + `common/query.py`): - Launch a vLLM server and run multi-turn inference to synthesize training data from input conversation skeletons - Fork-safe OpenAI client: reinitializes HTTP connection pool after `datasets.map()` forks worker processes, preventing 400 errors from corrupted connections - Clear Docker `ENTRYPOINT` so vLLM containers (which default to `vllm serve`) work correctly under NeMo Run's executor - `--max-tokens` argument to bound generation length - Local file loading support (`--data /path/to/file.jsonl`) - Re-raise connection errors so `datasets.map()` halts the shard instead of silently producing empty rows - Map `developer` role to `system` (OpenAI format compatibility) **Multi-turn reasoning trace handling** (`common/query.py`): - Strip `<think>...</think>` blocks from intermediate assistant turns before re-feeding to the model; preserve the full trace only on the final turn **Nemotron dataset preparation** (`examples/dataset/`): - `make_nemotron_ptv2_dataset.py` — prepares [nvidia/Nemotron-Post-Training-Dataset-v2](https://huggingface.co/datasets/nvidia/Nemotron-Post-Training-Dataset-v2) (~3.3M rows generate, ~1.9M rows train) - `make_nemotron_ptv3_dataset.py` — prepares the [Nemotron PTv3 collection](https://huggingface.co/collections/nvidia/nemotron-post-training-v3) of 16 datasets (~3.4M rows generate, ~3.9M rows train) - Both support `generate` mode (strips assistant turns for synthesis input) and `train` mode (normalizes to clean OpenAI format for SFT) - `conversation_utils.py` — shared utilities: `strip_assistant_turns`, `normalize_messages`, `make_augment_fn`, `AugmentationSpec` - `augmentations.yaml` — 12 language-redirect variants + style/format hints, cycled across dataset rows - Scripts live in `examples/dataset/` (not under `speculative_decoding/`) to signal reusability beyond speculative decoding **Bug fixes**: - `strip_assistant_turns()`: return `{"messages": []}` when no user turns remain (system-only rows were previously passed through instead of being filtered) - `concatenate_datasets()`: guard against empty parts list - SSH tunnel user precedence: explicit `user` arg now correctly overrides `slurm_config.user` ### Usage ```bash # Prepare PTv3 input conversations for synthesis (~3.4M rows): python examples/dataset/make_nemotron_ptv3_dataset.py --output-dir /tmp/ptv3_gen # Launch vLLM server + synthesize responses: bash tools/launcher/common/vllm/query.sh \ --model /path/to/model \ --tensor-parallel-size 4 \ -- \ --data /tmp/ptv3_gen/default.jsonl \ --save /tmp/ptv3_responses \ --num-shards 10 --num-proc 4 --max-tokens 4096 # Prepare PTv2 for direct SFT training (~1.9M rows): python examples/dataset/make_nemotron_ptv2_dataset.py --mode train --output-dir /tmp/ptv2_train ``` ### Testing Tested end-to-end on an NVIDIA GB10 node (119 GiB GPU memory) with `vllm/vllm-openai:qwen3_5-cu130` container and `Qwen/Qwen3.5-4B`: - vLLM server starts correctly with cleared Docker entrypoint - `datasets.map(num_proc=4)` runs without connection errors (fork-safe client) - Multi-turn synthesis produces correct assistant responses with thinking traces handled ### Before your PR is "*Ready for review*" Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md) and your commits are signed (`git commit -s -S`). Make sure you read and follow the [Security Best Practices](https://github.com/NVIDIA/Model-Optimizer/blob/main/SECURITY.md#security-coding-practices-for-contributors) (e.g. avoiding hardcoded `trust_remote_code=True`, `torch.load(..., weights_only=False)`, `pickle`, etc.). - Is this change backward compatible?: ✅ - If you copied code from any other sources or added a new PIP dependency, did you follow guidance in `CONTRIBUTING.md`: N/A - Did you write any new necessary tests?: N/A (data synthesis scripts; tested manually) - Did you update [Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?: N/A <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Added dataset generation and augmentation capabilities for Nemotron post-training datasets (v2 and v3) * Enhanced query functionality with thinking-block filtering and improved client management for robust parallel processing * Added support for local dataset file paths alongside HuggingFace Hub datasets * **Bug Fixes** * Fixed SLURM executor user resolution and Docker container entrypoint configuration * Improved error handling for connection failures during dataset synthesis * **Documentation** * Updated dataset preparation guide with new generation modes and augmentation configuration details <!-- end of auto-generated comment: release notes by coderabbit.ai --> Signed-off-by: Chenhan D. Yu <5185878+ChenhanYu@users.noreply.github.com> Signed-off-by: Chenhan Yu <chenhany@nvidia.com> |
||
|
h-guo18
|
82d96a635f |
[Speculative Decoding] Refactor EAGLE3 training to YAML-based config and recipe system (#1134)
## What does this PR do?
Refactors EAGLE3 training to use a single base YAML config with
OmegaConf dotlist overrides.
**Type of change:** Refactor
## Changes
- Single base config
`modelopt_recipes/speculative_decoding/_base_eagle3.yaml` for all EAGLE3
training; removed per-model child YAMLs.
- `launch_train.sh` accepts `--config <yaml>` plus dotlist overrides
(e.g. `model.model_name_or_path=xxx`).
- Removed `__base__` YAML inheritance logic from `main.py`.
- `dp_shard_size` default changed from `0` sentinel to `None` for
clarity.
- Removed `eagle_config.json` and `fsdp_config.json`; architecture
config is now nested under `eagle.eagle_architecture_config` in YAML.
- `train_eagle3_and_export.sh` now uses base YAML + dotlist instead of
generating a temporary YAML.
- Updated README and tests accordingly.
## Usage
```bash
# Online training
./launch_train.sh \
--config ../../modelopt_recipes/speculative_decoding/_base_eagle3.yaml \
model.model_name_or_path=meta-llama/Llama-3.2-1B \
data.data_path=input_conversations/train.jsonl \
training.output_dir=ckpts/llama-3.2-1b-online
# Offline training
./launch_train.sh \
--config ../../modelopt_recipes/speculative_decoding/_base_eagle3.yaml \
model.model_name_or_path=meta-llama/Llama-3.2-1B \
data.offline_data_path=$HIDDEN_STATES_DIR \
training.output_dir=ckpts/llama-3.2-1b-offline
```
---------
Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com>
|
||
|
Keval MorabiaandRinZ27
|
5dc17dfd15 |
[Security] Enable torch.load(weights_only=True) for secure checkpoint loading + trust_remote_code fix (#1181)
### What does this PR do? - Add secure checkpoint loading support using `torch.serialization.add_safe_globals([cls])`. This also removes 1 existing pickle usage. - Remove hard-coded `trust_remote_code=True` - Replaces https://github.com/NVIDIA/Model-Optimizer/pull/1056 by @RinZ27 ### Testing <!-- Mention how have you tested your change if applicable. --> CICD tests ran ### Before your PR is "*Ready for review*" Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md) and your commits are signed (`git commit -s -S`). Make sure you read and follow the [Security Best Practices](https://github.com/NVIDIA/Model-Optimizer/blob/main/SECURITY.md#security-coding-practices-for-contributors) (e.g. avoiding hardcoded `trust_remote_code=True`, `torch.load(..., weights_only=False)`, `pickle`, etc.). - Is this change backward compatible?: ✅ <!--- If ❌, explain why. --> - If you copied code from any other sources or added a new PIP dependency, did you follow guidance in `CONTRIBUTING.md`: ✅ <!--- Mandatory --> - Did you write any new necessary tests?: ✅ <!--- Mandatory for new features or examples. --> - Did you update [Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?: ✅ <!--- Only for new features, API changes, critical bug fixes or backward incompatible changes. --> ### Additional Information NVBug: 5999336 <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Added safe checkpoint save/load helpers and a --trust_remote_code CLI flag in examples to control remote-code loading. * **Bug Fixes** * Checkpoint loading now defaults to safer, weights-only semantics to reduce arbitrary-code exposure. * **Documentation** * CHANGELOG updated with security guidance and opt-in procedure for unsafe checkpoint loading. * **Tests** * New unit tests validating the safe-load behavior. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Signed-off-by: RinZ27 <222222878+RinZ27@users.noreply.github.com> Signed-off-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com> Co-authored-by: RinZ27 <222222878+RinZ27@users.noreply.github.com> |
||
|
Keval Morabiaandh-guo18
|
80d2f02a2d |
Fix spec dec example tests (#1183)
### What does this PR do? Type of change: Test fix <!-- Use one of the following: Bug fix, new feature, new example, new tests, documentation. --> - Fix `tests/examples/speculative_decoding` - previously silently skipped - Avoid pulling nemotron-post-training-dataset-v2 in tests to reduce chances of HF loading timeout in CICD - Make slow and redundant tests manual to speed up CICD ### Testing <!-- Mention how have you tested your change if applicable. --> - Tests passing ### Before your PR is "*Ready for review*" Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md) and your commits are signed (`git commit -s -S`). Make sure you read and follow the [Security Best Practices](https://github.com/NVIDIA/Model-Optimizer/blob/main/SECURITY.md#security-coding-practices-for-contributors) (e.g. avoiding hardcoded `trust_remote_code=True`, `torch.load(..., weights_only=False)`, `pickle`, etc.). - Is this change backward compatible?: ✅ <!--- If ❌, explain why. --> - If you copied code from any other sources or added a new PIP dependency, did you follow guidance in `CONTRIBUTING.md`: N/A <!--- Mandatory --> - Did you write any new necessary tests?: ✅ <!--- Mandatory for new features or examples. --> - Did you update [Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?: N/A <!--- Only for new features, API changes, critical bug fixes or backward incompatible changes. --> <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Chores** * Removed git‑LFS install step from CI and deleted an automated branch‑cleanup workflow * Trimmed example environment dependencies and relaxed transformers compatibility; added an optional tokenization dependency * **Tests** * Switched tests to generate datasets dynamically and improved fixture handling * Standardized PTQ test parameters (explicit calibration dataset) and refined GPU/test selection * **Bug Fixes** * Improved device-awareness and numeric handling in speculative decoding attention paths <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Signed-off-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com> Co-authored-by: h-guo18 <67671475+h-guo18@users.noreply.github.com> |
||
|
h-guo18
|
7f5fd65003 |
[Feat]FakeBaseModel for offline eagle; Kimi-K2.5 fixes; (#1052)
### What does this PR do? Adds `FakeBaseModel` for offline EAGLE training and several Kimi-K2.5 compatibility fixes. - **New**: `FakeBaseModel` — lightweight model that loads only `lm_head` and `embed_tokens` from a local checkpoint, avoiding full model weight loading during offline training. Configured via `FakeBaseArguments` and integrated into `load_vlm_or_llm`. - **Fix**: `_find_base_model_parts` — support Kimi-K2.5 VLM layout (`language_model.model` path) - **Fix**: offline mode lm_head access and CompressedTensors ignore path - **Fix**: Kimi-K2.5 decoder `past_key_value`/`past_key_values` argument mismatch - **Fix**: `rglob` for `.pt` discovery in nested offline data dirs; single-node GPU count respects `CUDA_VISIBLE_DEVICES` Type of change: Bug fix, new feature ### Testing Tested offline EAGLE training for Kimi-K2.5 end-to-end. ### Before your PR is "*Ready for review*" Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md) and your commits are signed (`git commit -s -S`). Make sure you read and follow the [Security Best Practices](https://github.com/NVIDIA/Model-Optimizer/blob/main/SECURITY.md#security-coding-practices-for-contributors) (e.g. avoiding hardcoded `trust_remote_code=True`, `torch.load(..., weights_only=False)`, `pickle`, etc.). - Is this change backward compatible?: ✅ - If you copied code from any other sources or added a new PIP dependency, did you follow guidance in `CONTRIBUTING.md`: N/A - Did you write any new necessary tests?: ❌ - Did you update [Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?: ❌ ### Additional Information <!-- E.g. related issue. --> <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Lightweight fake-base model support for offline speculative-decoding training * **Improvements** * Added CLI flags: --use_fake_base_for_offline, --trust_remote_code, and --fsdp * Expanded offline .pt discovery to include nested subdirectories * Better GPU detection with explicit single-node logging; FSDP enabled only when requested * Model loading and launch tooling now honor offline and trust-remote-code flags * **Bug Fixes** * Improved compatibility with legacy transformer / Kimi-K2 call signatures * **Tests** * Added tests covering fake-base loading and offline training workflows <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com> |
||
|
Keval Morabia
|
d0bf0bef96 |
Remove deprecated Nemo 2.0 references / examples (#1098)
### What does this PR do? - Remove `examples/nemo_run` and other deprecated Nemo 2.0 references - Add Megatron-Bridge example links where missing <!-- Details about the change. --> ### Testing <!-- Mention how have you tested your change if applicable. --> ### Before your PR is "*Ready for review*" Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md) and your commits are signed (`git commit -s -S`). Make sure you read and follow the [Security Best Practices](https://github.com/NVIDIA/Model-Optimizer/blob/main/SECURITY.md#security-coding-practices-for-contributors) (e.g. avoiding hardcoded `trust_remote_code=True`, `torch.load(..., weights_only=False)`, `pickle`, etc.). - Is this change backward compatible?: ✅ <!--- If ❌, explain why. --> - If you copied code from any other sources or added a new PIP dependency, did you follow guidance in `CONTRIBUTING.md`: N/A <!--- Mandatory --> - Did you write any new necessary tests?: N/A <!--- Mandatory for new features or examples. --> - Did you update [Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?: ✅ <!--- Only for new features, API changes, critical bug fixes or backward incompatible changes. --> <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Deprecations** * Removed deprecated NeMo 2.0 support and related example flows and utilities. * **Documentation** * Updated docs and examples to emphasize Megatron-Bridge / Megatron-LM and refreshed technique/deployment guidance and links. * **New Features** * Added CLI options for additional parallelism (context/expert tensor/expert model) in Megatron-Bridge distillation. * **Chores** * Removed legacy CI configs and refreshed container image tags across examples and docs. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Signed-off-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com> |
||
|
Benjamin Chislettandh-guo18
|
547d8fe935 |
[EAGLE] Optimize EAGLE Training (#1044)
### What does this PR do? Type of change: Optimization <!-- Details about the change. --> Changes: - Precompute base_model_logits.argmax() and base_model_logits.softmax() instead of recomputing in every call to _eagle_loss - Calculate per-prediction accuracy on the GPU and synchronize it to the host after running all TTT steps, to avoid cpu/gpu synchronization inside the TTT step loop. - Apply torch.compile to performance-critical training functions: prepare inputs, eagle forward, and eagle loss calculation. Omitted from target model in online case for now, as it may not be natively compatible with all architectures. ### Usage No changes to external interfaces ### Testing Ran training commands for benchmarking. Did not do a full training run, did not validate correctness. ### Before your PR is "*Ready for review*" Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md) and your commits are signed (`git commit -s -S`). Make sure you read and follow the [Security Best Practices](https://github.com/NVIDIA/Model-Optimizer/blob/main/SECURITY.md#security-coding-practices-for-contributors) (e.g. avoiding hardcoded `trust_remote_code=True`, `torch.load(..., weights_only=False)`, `pickle`, etc.). - Is this change backward compatible?: ✅ - If you copied code from any other sources or added a new PIP dependency, did you follow guidance in `CONTRIBUTING.md`: N/A <!--- Mandatory --> - Did you write any new necessary tests?: N/A - Did you update [Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?: ✅ / ❌ / N/A <!--- Only for new features, API changes, critical bug fixes or backward incompatible changes. --> ### Additional Information <!-- E.g. related issue. --> <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Added runtime profiling ranges for speculative decoding to enable finer performance tracing. * **Improvements** * Lazy initialization for rotary embeddings in llama-style decoders for more reliable startup. * Speculative decoding now uses base-model predicted tokens and softmax probabilities for sampling and loss, improving stability and accuracy reporting. * Parallelism configuration is now conditional, avoiding unnecessary setup unless required. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Signed-off-by: Benjamin Chislett <bchislett@nvidia.com> Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com> Co-authored-by: h-guo18 <67671475+h-guo18@users.noreply.github.com> |
||
|
Benjamin Chislett
|
c76633ac9d |
[EAGLE] Configurable number of TTT steps (#1042)
### What does this PR do? Type of change: new CLI option for existing option <!-- Details about the change. --> - Added num_ttt_steps CLI flag - Changed num_ttt_steps default from 4 to 3 for consistency. Num_spec_tokens == 3 or == 7 are most common in practice, so rounding down to 3 and allowing users to increment higher on-demand. Will also improve training efficiency for the OOTB experience. ### Usage Users can now pass `--num_ttt_steps 7` to `launch_train.sh` when training an EAGLE3 model for extended speculation lengths. ### Testing N/A ### Before your PR is "*Ready for review*" Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md) and your commits are signed (`git commit -s -S`). Make sure you read and follow the [Security Best Practices](https://github.com/NVIDIA/Model-Optimizer/blob/main/SECURITY.md#security-coding-practices-for-contributors) (e.g. avoiding hardcoded `trust_remote_code=True`, `torch.load(..., weights_only=False)`, `pickle`, etc.). - Is this change backward compatible?: ✅ - If you copied code from any other sources or added a new PIP dependency, did you follow guidance in `CONTRIBUTING.md`: N/A - Did you write any new necessary tests?: N/A - Did you update [Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?: N/A <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Added ability to configure train-time-test steps for speculative decoding training via command-line argument. * Updated default train-time-test steps value from 4 to 3. <!-- end of auto-generated comment: release notes by coderabbit.ai --> Signed-off-by: Benjamin Chislett <bchislett@nvidia.com> |
||
|
Benjamin Chislett
|
4292505512 |
Refactor: Clean up EAGLE training dataset preparation (#684)
## What does this PR do? **Type of change:** Refactor **Overview:** - Consolidate input dataset preparation into `make_dataset.py` - Read dataset mix spec from a YAML file - - Can now specify how many samples to take from each split - - Can no longer easily split a dataset into train/test sections. I don't think this feature was really useful to begin with. Most datasets can already be separated into train/val/test at the split level, and those that can't are usually going to be splitted by the training FW anyways. - Add support for a few new dataset types, magpie 300k/500k/1M, nemotron post-training dataset v2. ## Usage See README for detailed example ## Testing Ran it locally on all dataset modes, works successfully and output looks good. Checked shuffling, conversation IDs, and output contents were all unique and usable. - **Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md)** and your commits are signed. - **Is this change backward compatible?**: Yes - **Did you write any new necessary tests?**: No - **Did you add or update any necessary documentation?**: Yes - **Did you update [Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?**: No <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Documentation** * Updated speculative decoding example documentation with new dataset references and standardized file paths. * **New Features** * Introduced configuration-driven dataset preparation supporting multiple dataset sources with centralized configuration files. * **Refactor** * Simplified dataset preparation workflow with unified tooling and updated default data paths throughout the training pipeline. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Signed-off-by: Benjamin Chislett <bchislett@nvidia.com> |
||
|
yeyu-nvidia
|
5d0e012751 |
inplement mix hidden_states for eagle3; deprecate eagle1 (#946)
## What does this PR do? new feature **Overview:** Enable mix hidden_states in eagle3 training. Deprecate eagle1 ## Usage Add --mix_hidden_states True to launch_train.sh ```python # Add a code snippet demonstrating how to use this ``` ## Testing <!-- Mention how have you tested your change if applicable. --> ## Before your PR is "*Ready for review*" <!-- If you haven't finished some of the above items you can still open `Draft` PR. --> - **Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md)** and your commits are signed. - **Is this change backward compatible?**: Yes/No <!--- If No, explain why. --> - **Did you write any new necessary tests?**: Yes/No - **Did you add or update any necessary documentation?**: Yes/No - **Did you update [Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?**: Yes/No <!--- Only for new features, API changes, critical bug fixes or bw breaking changes. --> ## Additional Information <!-- E.g. related issue. --> <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Added --mix_hidden_states option to enable optional hidden-state mixing during training. * Added eagle_ttt_steps setting to control speculative multi-step iterations. * **Chores** * Consolidated speculative decoding to EAGLE3 only; legacy Medusa/EAGLE1 paths removed. * Unified configuration handling so models and plugins accept a single config object. * **Tests** * Updated and expanded tests for hidden-state mixing and EAGLE3-only scenarios. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Signed-off-by: Ye Yu <yeyu@nvidia.com> |
||
|
skierat
|
dd16a96fda |
example demonstrating how to train CosmosReason2 Eagle3 (#965)
## What does this PR do? **Type of change:** New example **Overview:** Adds examples/speculative_decoding/guides/train_eagle_head_cosmos_reason2.ipynb, a step-by-step Jupyter notebook that walks through the full EAGLE3 draft-head training workflow for nvidia/Cosmos-Reason2-8B. The notebook covers: 1. Installing dependencies 2. Authenticating with Hugging Face 3. Preparing training data from the Nemotron-Post-Training-Dataset-v2 (chat split) using a curated row-selection mapping (guides/nemotron_mapping.csv) 4. Inspecting the bundled EAGLE3 config (guides/CR2_eagle_config.json) tuned for Cosmos-Reason2 (YaRN RoPE, FlexAttention, reduced draft vocabulary) 5. (Optional) Calibrating the draft vocabulary to 32k tokens for faster training and inference 6. Launching training via launch_train.sh with FSDP2 multi-GPU support 7. Exporting the checkpoint to HF format and serving with vLLM Also includes guides/nemotron_mapping.csv and guides/CR2_eagle_config.json as companion files. ## Usage Open and run examples/speculative_decoding/guides/train_eagle_head_cosmos_reason2.ipynb cell by cell. After training, serve the exported checkpoint with: ``` vllm serve nvidia/Cosmos-Reason2-8B \ --host 0.0.0.0 \ --port 8000 \ --speculative-model export/cosmos-reason2-8b-eagle3 \ --num-speculative-tokens 3 \ --dtype bfloat16 ``` ## Testing Tested end-to-end on a 4xB100 GPUs. The exported checkpoint was validated with specdec_bench ## Before your PR is "*Ready for review*" <!-- If you haven't finished some of the above items you can still open `Draft` PR. --> - **Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md)** and your commits are signed. - **Is this change backward compatible?**: Yes - **Did you write any new necessary tests?**: No - **Did you add or update any necessary documentation?**: Yes (the notebook is self-documenting) - **Did you update [Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?**: No ## Additional Information Cosmos-Reason2-8B requires at least one 80 GB GPU (H100/A100) <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Added a speculative-decoding configuration for draft model and rotary/attention behavior. * Added an end-to-end training notebook demonstrating training, export, and deployment of a speculative-decoding draft head on Cosmos-Reason2. * Added a data-preparation tool to download, normalize, and convert Nemotron chat conversations into a standardized conversation format. * **Documentation** * Notebook documents environment setup, data prep, training/validation cadence, export, and deployment steps. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Signed-off-by: Slawek Kierat <skierat@nvidia.com> |
||
|
h-guo18
|
a34d613d3c |
Feat: Speculatice Decoding export with quantization support (#913)
## What does this PR do? **Type of change:** ? <!-- Use one of the following: Bug fix, new feature, new example, new tests, documentation. --> **Overview:** Main changes: - Refactored speculative decoding export logics into `class EagleExporter` to improve cohesion; - Separated speculative decoding export entrance with quantization export (`export_hf_checkpoint()`) due to their fundamental differences: - Quantization export base model's state_dict and config, while speculative decoding only export drafter's. - Most of the model-specific logics of quantization export (e.g. diffusers, vlms) are not needed for speculative decoding export. - Quantization export produce different format than speculative decoding checkpoint. (The former produce tokenizer config, generation config, e.t.c, while the later does not need. ) ## Usage <!-- You can potentially add a usage example below. --> To export an regular bf16 eagle checkpoint without quantization, the commands are the same: ```python python scripts/export_hf_checkpoint.py --model_path <x> --export_path <x> ``` To run PTQ on online-trained eagle checkpoint and export it: ```python python hf_ptq.py --pyt_ckpt_path <x> --qformat fp8 --export_path <x> ``` The above two commands will produce drafter ckpt for deployment, in the same foramt. ## Testing <!-- Mention how have you tested your change if applicable. --> Tested setting: - Base model: llama3.1-8b - Algorithms: eagle - Export path tested: - (Unquantized online ckpt) `python scripts/export_hf_checkpoint.py --model_path <x> --export_path <x>` - (PTQ) export `python hf_ptq.py --pyt_ckpt_path <x> --qformat fp8 --export_path <x>` - Tested deployment on vllm. Got normal AR. ## Before your PR is "*Ready for review*" <!-- If you haven't finished some of the above items you can still open `Draft` PR. --> - **Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md)** and your commits are signed. - **Is this change backward compatible?**: Yes/No <!--- If No, explain why. --> - **Did you write any new necessary tests?**: Yes/No - **Did you add or update any necessary documentation?**: Yes/No - **Did you update [Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?**: Yes/No <!--- Only for new features, API changes, critical bug fixes or bw breaking changes. --> ## Additional Information <!-- E.g. related issue. --> <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit ## Release Notes * **New Features** * Added export functionality for speculative decoding-optimized models * Support for multiple speculative decoding architectures with pre-configured deployment templates * Enhanced model export detection and automatic routing for optimized models * **Tests** * Updated export validation tests for speculative decoding models <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com> |
||
|
Chenhan D. Yu
|
edde087bbd |
Adding a special list to handle HF models that require (#950)
## What does this PR do? **Type of change:** ? <!-- Use one of the following: Bug fix, new feature, new example, new tests, documentation. --> Bug fix **Overview:** ? 1. Some models check `layer_types` and our offline overwrites `num_hidden_layers=0` which will requires passing `layer_types=[]`. It is also possible that the model will raise unexpected arguments when passing `layer_types=[]`. So the current WAR is to use a list to handle by checking the architectures. 2. Fix relative path issue in `launch_train.sh` ## Usage <!-- You can potentially add a usage example below. --> ```python # Add a code snippet demonstrating how to use this ``` ## Testing <!-- Mention how have you tested your change if applicable. --> ## Before your PR is "*Ready for review*" <!-- If you haven't finished some of the above items you can still open `Draft` PR. --> - **Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md)** and your commits are signed. - **Is this change backward compatible?**: Yes/No <!--- If No, explain why. --> - **Did you write any new necessary tests?**: Yes/No - **Did you add or update any necessary documentation?**: Yes/No - **Did you update [Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?**: Yes/No <!--- Only for new features, API changes, critical bug fixes or bw breaking changes. --> ## Additional Information <!-- E.g. related issue. --> <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Bug Fixes** * Enhanced model loading robustness for specific layer configurations. * **Chores** * Improved example script path handling and GPU initialization tracking. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Signed-off-by: Chenhan Yu <chenhany@nvidia.com> |
||
|
yeyu-nvidia
|
4a11486f6c |
Enable multinode training for HF speculative decoding (#922)
## What does this PR do? **Type of change:** New example **Overview:** Modify launch_train.sh script to enable multi-node training. Provide a slurm template script. ## Usage Add required fields in slurm.sh. Then use the command below to submit multi-node job: ```bash bash slurm.sh ``` ## Testing <!-- Mention how have you tested your change if applicable. --> ## Before your PR is "*Ready for review*" <!-- If you haven't finished some of the above items you can still open `Draft` PR. --> - **Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md)** and your commits are signed. - **Is this change backward compatible?**: Yes/No <!--- If No, explain why. --> - **Did you write any new necessary tests?**: Yes/No - **Did you add or update any necessary documentation?**: Yes/No - **Did you update [Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?**: Yes/No <!--- Only for new features, API changes, critical bug fixes or bw breaking changes. --> ## Additional Information <!-- E.g. related issue. --> <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Added multi-node training support with new configuration options for distributed setups. * Introduced Slurm batch script for streamlined job submission to cluster environments. * **Improvements** * Enhanced GPU resource management for multi-GPU and distributed training configurations. * Updated speculative decoding model support and validation handling. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Signed-off-by: Ye Yu <yeyu@nvidia.com> |
||
|
h-guo18
|
eb99488da1 |
Fix: restore requires_grad in transformers5 reloading (#907)
## What does this PR do? **Type of change:** ? <!-- Use one of the following: Bug fix, new feature, new example, new tests, documentation. --> **Overview:** Patch transformers 5.x parameter loading to preserve original `requires_grad` settings. In transformers v5.x, loading a checkpoint forcibly sets parameters' requires_grad, which unintentionally unfreeze frozen parameters (e.g. Base model in eagle training). This leads to optimizer initialization error since the restored optimizer expected more parameter than the checkpoint. This monkey-patch restores the original`requires_grad` after loading parameters. Reference: https://github.com/huggingface/transformers/blob/v5.0.0.rc1-release/src/transformers/core_model_loading.py#L640 ## Usage <!-- You can potentially add a usage example below. --> ```python # Add a code snippet demonstrating how to use this ``` ## Testing <!-- Mention how have you tested your change if applicable. --> ## Before your PR is "*Ready for review*" <!-- If you haven't finished some of the above items you can still open `Draft` PR. --> - **Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md)** and your commits are signed. - **Is this change backward compatible?**: Yes/No <!--- If No, explain why. --> - **Did you write any new necessary tests?**: Yes/No - **Did you add or update any necessary documentation?**: Yes/No - **Did you update [Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?**: Yes/No <!--- Only for new features, API changes, critical bug fixes or bw breaking changes. --> ## Additional Information <!-- E.g. related issue. --> <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Bug Fixes** * Fixed model parameter loading in speculative decoding to properly preserve gradient requirements for each parameter when using HuggingFace Transformers 5.x, ensuring correct behavior during checkpoint resumption and model initialization. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com> |
||
|
h-guo18
|
b8a4586702 |
Refactor: Eagle data loading (#668)
## What does this PR do? **Type of change:** Refactor <!-- Use one of the following: Bug fix, new feature, new example, new tests, documentation. --> **Overview:** Jira ticket: https://jirasw.nvidia.com/browse/OMNIML-2955 Main changes : - Consolidate Eagle data loading with @ChenhanYu 's implementation of `transformers_dataset.py` - Refactor: baked the following logics from `example/main.py` to `modelopt/torch` for cleaner example entrance: - default config selecting and merging with custom config - tokenizer post-processor (chat template and pad_tok_id) - d2t loading - Implementation refactor: In HF workflow, reuse base modfel's input hidden states as input_embedding, instead of calculating from input_ids. This has two main benefits: - Easier VLM support, which has various embedding processing logics. - Training effieicy. - Deprecating eagle1 from the example. It is still available by setting custom config. - Other minor fixes and readme updates. ## Usage <!-- You can potentially add a usage example below. --> ```python # Add a code snippet demonstrating how to use this ``` ## Testing <!-- Mention how have you tested your change if applicable. --> Tested that training curves after changes (both online&offline) is identical with original branch: <img width="1073" height="634" alt="image" src="https://github.com/user-attachments/assets/abfd7bea-c82c-48a7-8181-68c5a9e4da8d" /> ## Before your PR is "*Ready for review*" <!-- If you haven't finished some of the above items you can still open `Draft` PR. --> - **Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md)** and your commits are signed. - **Is this change backward compatible?**: Yes/No <!--- If No, explain why. --> - **Did you write any new necessary tests?**: Yes/No - **Did you add or update any necessary documentation?**: Yes/No - **Did you update [Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?**: Yes/No <!--- Only for new features, API changes, critical bug fixes or bw breaking changes. --> ## Additional Information <!-- E.g. related issue. --> <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit ## Release Notes * **New Features** * Added draft vocabulary cache support for EAGLE model training, enabling runtime vocabulary customization via `--draft_vocab_cache` parameter * Introduced new data loading utilities with sharding, streaming, and tokenization support for large-scale training * Added optional `--log_steps` configuration to training launcher * **Documentation** * Updated EAGLE configuration guides with draft vocabulary cache setup instructions and examples * **Refactor** * Restructured data pipeline for offline training with improved dataset handling and batching * Updated command-line arguments across training scripts (`--input-data` replaces `--input-file`) <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com> |
||
|
yeyu-nvidia
|
a8f5314c93 |
fix the path change in torch v2.10 for spec dec (#863)
## What does this PR do? **Type of change:** bug fix **Overview:** torch v2.10 changes the path for _SDPAMerger. will need to use the new path for import ## Usage <!-- You can potentially add a usage example below. --> ```python # Add a code snippet demonstrating how to use this ``` ## Testing <!-- Mention how have you tested your change if applicable. --> ## Before your PR is "*Ready for review*" <!-- If you haven't finished some of the above items you can still open `Draft` PR. --> - **Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md)** and your commits are signed. - **Is this change backward compatible?**: Yes/No <!--- If No, explain why. --> - **Did you write any new necessary tests?**: Yes/No - **Did you add or update any necessary documentation?**: Yes/No - **Did you update [Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?**: Yes/No <!--- Only for new features, API changes, critical bug fixes or bw breaking changes. --> ## Additional Information <!-- E.g. related issue. --> <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Chores** * Updated internal import references to reflect organizational changes in dependencies. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Signed-off-by: Ye Yu <yeyu@nvidia.com> |
||
|
yeyu-nvidia
|
81b67ddf06 |
Context parallelism for Megatron core models (#818)
## What does this PR do? New feature **Overview:** This PR implements the context manager which injects attn_mask as attn_bias to TEDotProductAttention so that we can enable EAGLE training with arbitrary mask. ## Usage set CP>1 in https://github.com/NVIDIA/Megatron-LM/blob/main/examples/post_training/modelopt/finetune.sh ```python # Add a code snippet demonstrating how to use this ``` ## Testing Tested on DSR1 Llama 8B. CP1->CP2 38854MB->28050MB MTbench AL 2.26->2.31 ## Before your PR is "*Ready for review*" <!-- If you haven't finished some of the above items you can still open `Draft` PR. --> - **Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md)** and your commits are signed. - **Is this change backward compatible?**: Yes/No <!--- If No, explain why. --> - **Did you write any new necessary tests?**: Yes/No - **Did you add or update any necessary documentation?**: Yes/No - **Did you update [Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?**: Yes/No <!--- Only for new features, API changes, critical bug fixes or bw breaking changes. --> ## Additional Information <!-- E.g. related issue. --> <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit ## New Features * Context parallelism support added for Eagle speculative decoding with HuggingFace and Megatron Core models. * Model checkpoint loading enhanced to enable remote code execution capabilities when required. <sub>✏️ Tip: You can customize this high-level summary in your review settings.</sub> <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Signed-off-by: Ye Yu <yeyu@nvidia.com> |
||
|
h-guo18
|
3036a9ea9f |
Feat: Context Parallel for Eagle3 Training (#745)
## What does this PR do?
**Type of change:** New Feature <!-- Use one of the following: Bug fix,
new feature, new example, new tests, documentation. -->
**Overview:**
- Supported Context Parallel by patching torch ring attention;
- Require following libirary version for stable cp:
- torch2.8.0
- transformers5.0.0
- accelrate1.12.0
- Move to FSDP2
- Removed unused arguments in training script (`--multi_gpu`,
`fsdp_wrap_layer`)
- Bump CI container to `nvcr.io/nvidia/pytorch:25.08-py3`
## Usage
<!-- You can potentially add a usage example below. -->
```bash
./launch_train.sh --model $MODEL \
--output_dir $OUTPUT_DIR \
--data $DATA \
--num_epochs 0.1 \
--train_bs 1 \
--eagle_config eagle_config.json \
--training_seq_len 1024 \
--cp_size 2 #newly added
```
## Testing
- SDPA level correctness: tested TTT attention with/without CP, diff <
1%
```
=== Compare context-parallel (CP) outputs and grads with non-CP ===
Forward output comparison (CP vs Non-CP):
Absolute diff (adiff) cp_out vs out: 0.001953125
Relative diff (rdiff) cp_out vs out: 0.00182342529296875
WQ (query proj) grad comparison (CP vs Non-CP):
Absolute diff (adiff) cp_wq_grad vs wq_grad: 0.0078125
Relative diff (rdiff) cp_wq_grad vs wq_grad: 0.00347900390625
WK (key proj) grad comparison (CP vs Non-CP):
Absolute diff (adiff) cp_wk_grad vs wk_grad: 0.0078125
Relative diff (rdiff) cp_wk_grad vs wk_grad: 0.002471923828125
WV (value proj) grad comparison (CP vs Non-CP):
Absolute diff (adiff) cp_wv_grad vs wv_grad: 0.25
Relative diff (rdiff) cp_wv_grad vs wv_grad: 0.0069580078125
==============================================================
```
- E2E Training Acc
(Llama3.1-8B, Unsynthesized magpie)
<img width="911" height="630" alt="image"
src="https://github.com/user-attachments/assets/1ecacc7f-c720-494c-9c1b-b60e7ced7baa"
/>
- Peak Mem Reserved
(llama3.1-8B, 8xH100, train_length=4k)
| cp_size | max_memory_allocated(MB) |max_memory_reserved (MB) |
|----|--------------------------|--------------------------|
| 1 | 65040.20 |79018.00
| 2 | 50409.17 |73098.00
| 4 | 45120.92 |72052.00
| 8 | 38882.12 |66484.00
- Max Training Length test
(llama3.1-8B, H100)
| cp_size | 6k | 12k | 24k | 48k |
|--------------------|-----|-----|-----|-----|
| 1 | ✅ | OOM | OOM | OOM |
|2 | ✅ | ✅ | OOM | OOM |
| 4 | ✅ | ✅ | ✅ | OOM |
| 8 | ✅ | ✅ | ✅ | ✅ |
## Before your PR is "*Ready for review*"
<!-- If you haven't finished some of the above items you can still open
`Draft` PR. -->
- **Make sure you read and follow [Contributor
guidelines](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md)**
and your commits are signed.
- **Is this change backward compatible?**: Yes/No <!--- If No, explain
why. -->
- **Did you write any new necessary tests?**: Yes/No
- **Did you add or update any necessary documentation?**: Yes/No
- **Did you update
[Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?**:
Yes/No <!--- Only for new features, API changes, critical bug fixes or
bw breaking changes. -->
## Additional Information
<!-- E.g. related issue. -->
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Added context parallelism (CP) and data parallelism shard size
configuration parameters to training arguments.
* **Enhancements**
* Improved TTT attention masking support for speculative decoding
workflows.
* Enhanced training launch script with improved parallelism
configuration handling.
* **Chores**
* Updated core dependencies: torch, transformers, accelerate, and wandb.
* Added FSDP configuration file for distributed training setup.
<sub>✏️ Tip: You can customize this high-level summary in your review
settings.</sub>
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com>
|
||
|
h-guo18
|
3fd8b804f4 |
Fix:add vllm dump script (#710)
## What does this PR do? **Type of change:** ? <!-- Use one of the following: Bug fix, new feature, new example, new tests, documentation. --> **Overview:** ? Missing one file in last PR:https://github.com/NVIDIA/Model-Optimizer/pull/689/changes ## Usage <!-- You can potentially add a usage example below. --> ```python # Add a code snippet demonstrating how to use this ``` ## Testing <!-- Mention how have you tested your change if applicable. --> ## Before your PR is "*Ready for review*" <!-- If you haven't finished some of the above items you can still open `Draft` PR. --> - **Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md)** and your commits are signed. - **Is this change backward compatible?**: Yes/No <!--- If No, explain why. --> - **Did you write any new necessary tests?**: Yes/No - **Did you add or update any necessary documentation?**: Yes/No - **Did you update [Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?**: Yes/No <!--- Only for new features, API changes, critical bug fixes or bw breaking changes. --> ## Additional Information <!-- E.g. related issue. --> Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com> |
||
|
Ben Shoham Ofir
|
2a51bbd7cb |
Optimize calibrate_draft_vocab to read only required lines when calib… (#618)
Optimize calibrate_draft_vocab to read only required lines when
calibrate_size is set
## What does this PR do?
**Type of change:** Performance improvement
**Overview:**
This PR optimizes the
[calibrate_draft_vocab.py](cci:7://file:///Users/obenshoham/PycharmProjects/TensorRT-Model-Optimizer/examples/speculative_decoding/scripts/calibrate_draft_vocab.py:0:0-0:0)
script to improve memory efficiency and I/O performance when using the
`--calibrate_size` parameter. Previously, the script would read all
lines from the data file into memory before slicing to the specified
`calibrate_size`, causing unnecessary resource usage for large datasets.
The optimization uses `itertools.islice` to read only the required
number of lines when `calibrate_size` is specified.
## Usage
The script usage remains unchanged. When using `--calibrate_size`, the
script now only reads the specified number of lines instead of loading
the entire dataset:
```bash
# Only reads first 1000 lines from the dataset (optimized)
python scripts/calibrate_draft_vocab.py \
--model meta-llama/Llama-3.2-1B-Instruct \
--data input_conversations/daring-anteater.jsonl \
--draft_vocab_size 32000 \
--calibrate_size 1000 \
--save_dir draft_vocab_cache
Signed-off-by: Ofir Ben Shoham <ofir_benshoham@intuit.com>
|
||
|
h-guo18andyeyu-nvidia
|
bdd10c2dbe |
Feat: MLA eagle (#689)
## What does this PR do?
**Type of change:** New Feature <!-- Use one of the following: Bug fix,
new feature, new example, new tests, documentation. -->
**Overview:**
- Add MLA Eagle support
- Add new argument "eagle_decoder_type" to switch between llama and
kimik2 eagle;
- Add patches to load from kimik2 model implementations dynamically;
- new default config for kimi k2;
- Refactor eagle export to support multilayer/multitype eagle export
concisely;
- Rename some modules for simplified export logic;
- Other minor improvements;
## Usage
<!-- You can potentially add a usage example below. -->
```python
# Add a code snippet demonstrating how to use this
```
## Testing
<!-- Mention how have you tested your change if applicable. -->
- Tested that kimi k2 thinking works with eagle_type=kimik2:
<img width="1068" height="636" alt="image"
src="https://github.com/user-attachments/assets/5557ef87-c719-4fb1-be18-30435f6b3885"
/>
- Tested that llama 3.2 1b works with eagle_type=llama:
<img width="1066" height="634" alt="image"
src="https://github.com/user-attachments/assets/633c575c-cc79-43af-aed3-0378a303ebc7"
/>
## Before your PR is "*Ready for review*"
<!-- If you haven't finished some of the above items you can still open
`Draft` PR. -->
- **Make sure you read and follow [Contributor
guidelines](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md)**
and your commits are signed.
- **Is this change backward compatible?**: Yes/No <!--- If No, explain
why. -->
- **Did you write any new necessary tests?**: Yes/No
- **Did you add or update any necessary documentation?**: Yes/No
- **Did you update
[Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?**:
Yes/No <!--- Only for new features, API changes, critical bug fixes or
bw breaking changes. -->
## Additional Information
<!-- E.g. related issue. -->
---------
Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com>
Signed-off-by: yeyu-nvidia <yeyu@nvidia.com>
Co-authored-by: yeyu-nvidia <yeyu@nvidia.com>
|
||
|
yeyu-nvidia
|
8c6de51e26 |
Yeyu/hf eagle medusa (#664)
## What does this PR do? new feature, **Overview:** This PR implements HF parallel draft by combining eagle and medusa. In training, multiple medusa heads are added and trained together with eagle. In inference, medusa heads are used to generate draft tokens after all eagle tokens. ## Usage Set parallel_draft_step>1 in eagle_config to enable parallel draft. ```python # Add a code snippet demonstrating how to use this ``` ## Testing <!-- Mention how have you tested your change if applicable. --> ## Before your PR is "*Ready for review*" <!-- If you haven't finished some of the above items you can still open `Draft` PR. --> - **Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/Model-Optimizer/blob/main/CONTRIBUTING.md)** and your commits are signed. - **Is this change backward compatible?**: Yes/No <!--- If No, explain why. --> - **Did you write any new necessary tests?**: Yes/No - **Did you add or update any necessary documentation?**: Yes/No - **Did you update [Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?**: Yes/No <!--- Only for new features, API changes, critical bug fixes or bw breaking changes. --> ## Additional Information <!-- E.g. related issue. --> --------- Signed-off-by: Ye Yu <yeyu@nvidia.com> |
||
|
Keval Morabia
|
53a2ddebab |
Product Rename: TensorRT Model Optimizer to Model Optimizer (#583)
- [x] Product Rename: TensorRT Model Optimizer to Model Optimizer (OMNIML-3033) - [x] Mention in Latest News section with date on the date of merging this PR (12/08) Signed-off-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com> |
||
|
h-guo18
|
5a4242faf4 |
Fix: update file path in eagle3 scripts (#654)
## What does this PR do? **Type of change:** Bug fix <!-- Use one of the following: Bug fix, new feature, new example, new tests, documentation. --> **Overview:** Update file path ## Usage <!-- You can potentially add a usage example below. --> ```python # Add a code snippet demonstrating how to use this ``` ## Testing <!-- Mention how have you tested your change if applicable. --> ## Before your PR is "*Ready for review*" <!-- If you haven't finished some of the above items you can still open `Draft` PR. --> - **Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/TensorRT-Model-Optimizer/blob/main/CONTRIBUTING.md)** and your commits are signed. - **Is this change backward compatible?**: Yes/No <!--- If No, explain why. --> - **Did you write any new necessary tests?**: Yes/No - **Did you add or update any necessary documentation?**: Yes/No - **Did you update [Changelog](https://github.com/NVIDIA/TensorRT-Model-Optimizer/blob/main/CHANGELOG.rst)?**: Yes/No <!--- Only for new features, API changes, critical bug fixes or bw breaking changes. --> ## Additional Information <!-- E.g. related issue. --> Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com> |
||
|
Keval Morabia
|
6702f99fec |
Bump TRT-LLM docker to 1.2.0rc4 (CUDA 13) (#578)
## What does this PR do? Bump TRT-LLM docker from 1.1.0rc2.post2 (CUDA 12) to 1.2.0rc4 (CUDA 13) ## Testing <!-- Mention how have you tested your change if applicable. --> - [ ] 2-gpu tests for all examples: ? ## Before your PR is "*Ready for review*" <!-- If you haven't finished some of the above items you can still open `Draft` PR. --> - **Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/TensorRT-Model-Optimizer/blob/main/CONTRIBUTING.md)** and your commits are signed. - **Is this change backward compatible?**: Yes <!--- If No, explain why. --> - **Did you write any new necessary tests?**: No - **Did you add or update any necessary documentation?**: Yes - **Did you update [Changelog](https://github.com/NVIDIA/TensorRT-Model-Optimizer/blob/main/CHANGELOG.rst)?**: Yes <!--- Only for new features, API changes, critical bug fixes or bw breaking changes. --> --------- Signed-off-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com> |
||
|
h-guo18
|
bc52b6cf12 |
Feat: Support VLLM one-model eagle ckpt; Add unit tests; (#573)
## What does this PR do? **Type of change:** New feature, new tests; <!-- Use one of the following: Bug fix, new feature, new example, new tests, documentation. --> **Overview:** - Add conversion scripts for eagle3 llm-compressor style checkpoint - Jira Ticket: https://jirasw.nvidia.com/browse/OMNIML-2866 - Add unit tests for `ar_validate.py`, `export_hf_checkpoint.py`, and `convert_to_vllm_ckpt.py`. ## Usage <!-- You can potentially add a usage example below. --> ```python python scripts/convert_to_vllm_ckpt.py --input <eagle3 ckpt> --verifier <base model> --output <path> ``` ## Testing <!-- Mention how have you tested your change if applicable. --> ## Before your PR is "*Ready for review*" <!-- If you haven't finished some of the above items you can still open `Draft` PR. --> - **Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/TensorRT-Model-Optimizer/blob/main/CONTRIBUTING.md)** and your commits are signed. - **Is this change backward compatible?**: Yes/No <!--- If No, explain why. --> - **Did you write any new necessary tests?**: Yes/No - **Did you add or update any necessary documentation?**: Yes/No - **Did you update [Changelog](https://github.com/NVIDIA/TensorRT-Model-Optimizer/blob/main/CHANGELOG.rst)?**: Yes/No <!--- Only for new features, API changes, critical bug fixes or bw breaking changes. --> ## Additional Information <!-- E.g. related issue. --> --------- Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com> |
||
|
h-guo18
|
fd79188d49 |
Refactor: eagle3 training loop & loss mask (#548)
## What does this PR do? **Type of change:** ? <!-- Use one of the following: Bug fix, new feature, new example, new tests, documentation. --> **Overview:** A few refactors to eagle3 training code: - Put first eagle step into TTT loop; - Support TTT mask beyond 4; - Remove answer-only loss mask in data loading. ## Usage <!-- You can potentially add a usage example below. --> ```python # Add a code snippet demonstrating how to use this ``` ## Testing <!-- Mention how have you tested your change if applicable. --> ## Before your PR is "*Ready for review*" <!-- If you haven't finished some of the above items you can still open `Draft` PR. --> - **Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/TensorRT-Model-Optimizer/blob/main/CONTRIBUTING.md)** and your commits are signed. - **Is this change backward compatible?**: Yes/No <!--- If No, explain why. --> - **Did you write any new necessary tests?**: Yes/No - **Did you add or update any necessary documentation?**: Yes/No - **Did you update [Changelog](https://github.com/NVIDIA/TensorRT-Model-Optimizer/blob/main/CHANGELOG.rst)?**: Yes/No <!--- Only for new features, API changes, critical bug fixes or bw breaking changes. --> ## Additional Information <!-- E.g. related issue. --> --------- Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com> |
||
|
h-guo18
|
1a89a8842c |
Feat: Eagle3 HF Online - support nemotron and VLMs (#463)
## What does this PR do?
**Type of change:** New feature <!-- Use one of the following: Bug fix,
new feature, new example, new tests, documentation. -->
**Overview:**
- Support the nano and nano-VL in eagle3 online mode:
- Added submodule path detection for `base model`, `lm_head`, and
`embeddings` to adapt different base model naming structure;
- Refactored data loading/preprocessing to support VLM;
- Attn backend improvement:
- Added option of `sdpa` in case `flex_attn` doesn't work.
- Added a unified TTT mask function that produce either `BlockMask` for
flex_attn or tensor masks for regular attn.
- Logging improvements:
- Added estimated AR validation during training. This is available for
both online and offline.
- Plot estimated AR and training acc to wandb for better training
visualization;
- Fix: PTQ on speculative decoding model.
## Usage
<!-- You can potentially add a usage example below. -->
For VLM as base model, pass in extra arguments `--vlm_processor
<hf_model_path> --vlm_img_dir <path to images>` in original launching
commands. Other usage unchanged.
E.g.
```bash
./launch_train.sh --model $MODEL \
--output_dir $OUTPUT_DIR \
--data $DATA \
--num_gpu 1 \
--num_epochs 2 \
--train_bs 2 \
--lr 3e-5 \
--eagle_config eagle_config.json \
--training_seq_len 4096 \
--vlm_processor $MODEL \
--vlm_img_dir <path to images>
```
## Testing
<!-- Mention how have you tested your change if applicable. -->
Tested short training with HF Online training on following models:
- `llama-3.2-1b` - data: daring-anteater
- The new nano (Hyrbid LLM) - data: daring-anteater
- The nano-VL - data: Llama-Nemotron-VLM-Dataset-v1/ocr_1
See loss decreasing and AR > 1.
## Before your PR is "*Ready for review*"
<!-- If you haven't finished some of the above items you can still open
`Draft` PR. -->
- **Make sure you read and follow [Contributor
guidelines](https://github.com/NVIDIA/TensorRT-Model-Optimizer/blob/main/CONTRIBUTING.md)**
and your commits are signed.
- **Is this change backward compatible?**: Yes/No <!--- If No, explain
why. -->
- **Did you write any new necessary tests?**: Yes/No
- **Did you add or update any necessary documentation?**: Yes/No
- **Did you update
[Changelog](https://github.com/NVIDIA/TensorRT-Model-Optimizer/blob/main/CHANGELOG.rst)?**:
Yes/No <!--- Only for new features, API changes, critical bug fixes or
bw breaking changes. -->
## Additional Information
<!-- E.g. related issue. -->
---------
Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com>
|
||
|
Izzy PuttermanandKeval Morabia
|
5adb9ba0fc |
Add Spec dec Bench example (#474)
## What does this PR do? **Type of change:** New feature <!-- Use one of the following: Bug fix, new feature, new example, new tests, documentation. --> **Overview:** Specdec bench example ## Usage <!-- You can potentially add a usage example below. --> ```python # Add a code snippet demonstrating how to use this ``` ## Testing <!-- Mention how have you tested your change if applicable. --> ## Before your PR is "*Ready for review*" <!-- If you haven't finished some of the above items you can still open `Draft` PR. --> - **Make sure you read and follow [Contributor guidelines](https://github.com/NVIDIA/TensorRT-Model-Optimizer/blob/main/CONTRIBUTING.md)** and your commits are signed. - **Is this change backward compatible?**: Yes <!--- If No, explain why. --> - **Did you write any new necessary tests?**: No - **Did you add or update any necessary documentation?**: Yes - **Did you update [Changelog](https://github.com/NVIDIA/TensorRT-Model-Optimizer/blob/main/CHANGELOG.rst)?**: Yes <!--- Only for new features, API changes, critical bug fixes or bw breaking changes. --> ## Additional Information <!-- E.g. related issue. --> --------- Signed-off-by: Izzy Putterman <iputterman@nvidia.com> Signed-off-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com> Co-authored-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com> |
||
|
Keval Morabia
|
90b1e68fcf |
Update HF nvidia collection links in docs (#475)
Signed-off-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com> |
||
|
Keval Morabia
|
547be00ace |
Fix LICENSE pre-commit hook
Signed-off-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com> |
||
|
h-guo18
|
ff8a1ed126 |
fix:eagle3 offline (#456)
Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com> |
||
|
h-guo18
|
40a7d243c3 |
[Eagle Offline] multinode support for hidden states dumper (#422)
Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com> |
||
|
h-guo18
|
557633c986 |
Fix: supporting gpt-oss HF eagle (#398)
Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com> |
||
|
h-guo18
|
abed33c3f8 |
Feat: TRTLLM Dumper for Eagle Offline Training (#404)
Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com> |
||
|
h-guo18
|
c55bcf0ab8 |
fix: eagle3 quantized base model (#383)
Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com> |
||
|
h-guo18
|
615f3c01f1 |
Efficient Eagle3 training with eagle KV cache and flex attention (#350)
Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com> |
||
|
h-guo18
|
4ff8fc9022 |
Example: add offline eagle training commands to README (#366)
Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com> |
||
|
h-guo18
|
26c203abde |
fix: update tests_eagle.py; disable eval for offline (#358)
Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com> |