trending/Model-Optimizer
1
0
Fork 0
mirror of https://github.com/NVIDIA/Model-Optimizer.git synced 2026-11-13 02:17:13 +08:00
Code Issues Packages Projects Releases Wiki Activity
178 Commits 503 Branches 62 Tags
9e64f81a67dfb84f3151a911554f338a205652a3
Commit Graph
70 Commits
Author SHA1 Message Date
Keval Morabia 9e64f81a67 NAS export refactor + skip conversion on minitron restore (#424) 
Signed-off-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com>
 2025-10-11 10:21:37 +00:00
Jenny Chen 99c44d38ed Sync amax & AWQ-Lite act_scale in context parallel/data parallel [OMNIML-2813] (#359) 
Signed-off-by: Jennifer Chen <jennifchen@nvidia.com>
 2025-10-10 23:16:27 +00:00
h-guo18 40a7d243c3 [Eagle Offline] multinode support for hidden states dumper (#422) 
Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com>
 2025-10-10 15:58:22 -07:00
kaix-nv 08fb23f4a7 Fix the fsdp cmd in llm_sparsity (#417) 
Signed-off-by: Kai Xu <kaix@nvidia.com>
 2025-10-10 19:01:50 +00:00
Keval Morabia 6dffcd0ff1 Add minitron pruning and distillation guidelines in pruning readme (#419) 
Signed-off-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com>
 2025-10-10 10:06:25 +00:00
vishalpandya1990 ee19a7ebd1 Add command-line options in Windows' llm-ptq example for Gather nodes' INT4 ONNX quantization (#418) 
Signed-off-by: vipandya <vipandya@nvidia.com>
 2025-10-10 11:15:05 +05:30
h-guo18 557633c986 Fix: supporting gpt-oss HF eagle (#398) 
Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com>
 2025-10-09 01:14:00 +00:00
Asha Anoosheh 78193fd5db Fix llm_distill example saving issue on new transformers w FSDP2 (#413) 
Signed-off-by: Asha Anoosheh <aanoosheh@nvidia.com>
 2025-10-08 16:34:20 +00:00
h-guo18 abed33c3f8 Feat: TRTLLM Dumper for Eagle Offline Training (#404) 
Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com>
 2025-10-07 20:24:14 -07:00
Huizi MaoandKeval Morabia 1537885f5b [New feature] Support fakequant serve with latest vllm and generalize calib logic (#369) 
Signed-off-by: Huizi Mao <ralphmao95@gmail.com>
Signed-off-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com>
Co-authored-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com>
 2025-10-07 18:17:21 +00:00
Asha Anoosheh d621b27867 Updated Prune-KD NeMo flow (#382) 
Signed-off-by: Asha Anoosheh <aanoosheh@nvidia.com>
 2025-10-07 19:20:41 +02:00
Chenjie Luo 340eb7a756 Remove Qwen tokenizer modification (#390) 
Signed-off-by: Chenjie Luo <chenjiel@nvidia.com>
 2025-10-04 17:41:08 +05:30
Keval Morabia cb44c556d8 Minor pruning update (#397) 
Signed-off-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com>
 2025-10-02 00:35:48 +05:30
ajrasane 0d51156662 [5256037] Automatically infer calibration shapes for per node calibration (#394) 
Signed-off-by: ajrasane <131806219+ajrasane@users.noreply.github.com>
 2025-10-01 10:07:29 -07:00
Chenjie Luo 17439e653d Move phi4_mm warning to above (#389) 
Signed-off-by: Chenjie Luo <108829653+cjluo-nv@users.noreply.github.com>
 2025-09-30 18:24:35 +00:00
realAsma d8ef6f8aec [Bug fix] remove max_memory from from_config init in hf_ptq (#373) 
Signed-off-by: realAsma <akuriparambi@nvidia.com>
 2025-09-30 09:13:54 -07:00
Mohammed Yasin 70abfb416f Fix FLOPs calculation (#388) 
Signed-off-by: Mohammed Yasin <32206511+Y-T-G@users.noreply.github.com>
Signed-off-by: Y-T-G <32206511+Y-T-G@users.noreply.github.com>
 2025-09-30 21:40:51 +05:30
h-guo18 c55bcf0ab8 fix: eagle3 quantized base model (#383) 
Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com>
 2025-09-29 18:30:26 -07:00
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>
 2025-09-26 14:04:01 -07:00
Yuening Li d649fb8375 Support INT8 Weight-Only Quantization (#263) 
Signed-off-by: Yuening Li <62227368+Yuening-wa@users.noreply.github.com>
 2025-09-26 17:46:30 +00:00
Zhiyu 576fbdc61e [Bugfix] Enable essential python files export in llm_ptq examples by adding a customized copy function (#351) 
Signed-off-by: Zhiyu Cheng <zhiyuc@nvidia.com>
 2025-09-25 21:04:33 -07:00
ajrasane a041bbee74 [5532019][AutoCast] Update GraphSanitizer to convert Double to Float32 (#364) 
Signed-off-by: ajrasane <131806219+ajrasane@users.noreply.github.com>
 2025-09-25 15:30:36 -07:00
Keval Morabia b4d6cedb4a Support checkpointing Minitron pruning scores / prune without sorting (#361) 
Signed-off-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com>
 2025-09-26 01:46:22 +05:30
Chenjie Luo 598b9ce6d9 [NVBUG: 5535437] Refactor engine_dir to checkpoint_dir in PTQ examples. (#365) 
Signed-off-by: Chenjie Luo <chenjiel@nvidia.com>
Signed-off-by: Chenjie Luo <108829653+cjluo-nv@users.noreply.github.com>
 2025-09-25 12:07:06 -07:00
h-guo18 4ff8fc9022 Example: add offline eagle training commands to README (#366) 
Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com>
 2025-09-24 16:41:42 -07:00
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>
 2025-09-23 19:26:20 +00:00
yueshen2016 7d5f636419 [Bug fix 5528642] [Bug fix 5528695] VLM NVBug fix (#355) 
Signed-off-by: Yue <yueshen@nvidia.com>
 2025-09-23 09:02:59 -07:00
ynankani-nv 8b1cedf2b7 [5506930]Add support in ModelOpt for generating mixed-precision (INT4… (#310) 
Signed-off-by: unknown <ynankani@nvidia.com>
 2025-09-23 12:09:32 +00:00
Benjamin Chislettandh-guo18 add61dbbde Feature: Offline training for EAGLE3 (#300) 
Signed-off-by: Benjamin Chislett <benjamin.chislett@centml.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>
 2025-09-22 23:54:09 +00:00
Keval Morabia c0590b0255 Deprecate ModelOpt custom docker and directly use TRT-LLM / PyTorch / TRT docker (#346) 
Signed-off-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com>
 2025-09-23 01:43:25 +05:30
Keval Morabia c60baae5fa Add Megatron-LM pruning example link (#344) 
Signed-off-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com>
 2025-09-19 23:36:42 +05:30
Chenjie Luo 5a3fd29fb1 Add int8_sq back to auto_quant support list (#345) 
Signed-off-by: Chenjie Luo <108829653+cjluo-nv@users.noreply.github.com>
 2025-09-19 06:04:19 -07:00
Keval Morabia 4c36abe536 Add llm_ptq PR test, Cleanup dependency installation in modelopt docker (#338) 
Signed-off-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com>
 2025-09-19 10:48:44 +05:30
Asha Anoosheh b895dc5c7c Disable KD mode from saving problematic state (#320) 
Signed-off-by: Asha Anoosheh <aanoosheh@nvidia.com>
 2025-09-19 02:31:20 +05:30
ajrasane 27a68219ee Specify engine path for TRT evaluation (#313) 
Signed-off-by: ajrasane <131806219+ajrasane@users.noreply.github.com>
 2025-09-18 13:09:53 -07:00
Jenny Chen 1c7c16e0c8 Slurm support for QAT Simplified Flow + Qwen3-8B recipe (#285) 
Signed-off-by: Jennifer Chen <jennifchen@nvidia.com>
 2025-09-18 11:24:39 -07:00
Chenjie Luo b7ed8cd7e6 [NVBug: 5525758] Update VLM-PTQ readme (#339) 
Signed-off-by: Chenjie Luo <chenjiel@nvidia.com>
 2025-09-18 16:51:00 +00:00
Chenjie Luo 912f3dcbd5 Reinstate int8_sq support for vlm_example. (#333) 
Signed-off-by: Chenjie Luo <chenjiel@nvidia.com>
 2025-09-18 10:43:01 +05:30
Chenjie LuoChenjie LuoKeval Morabia
 		146d1d9d40 Deprecate TRTLLM-build in examples (#297) 
Signed-off-by: Chenjie Luo <chenjiel@nvidia.com>
Signed-off-by: Chenjie Luo <108829653+cjluo-nv@users.noreply.github.com>
Signed-off-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com>
Co-authored-by: Chenjie Luo <chenjie@omniml.ai>
Co-authored-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com>
 2025-09-18 03:22:37 +05:30
h-guo18 461980e6df update eagle example notebook (#314) 
Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com>
 2025-09-17 14:17:50 -07:00
realAsma 35247328d7 [1/N] QATTrainer training workflow fixes and clean up; Added backend specific unitests; (#318) 
Signed-off-by: realAsma <akuriparambi@nvidia.com>
 2025-09-17 01:43:44 +00:00
Farshad Ghodsian 76e8ce21bf Add QAT Walkthrough Notebook example (#278) 
Signed-off-by: Farshad Ghodsian <47931571+farshadghodsian@users.noreply.github.com>
 2025-09-11 04:12:26 +00:00
jingyu-ml 358b0c61b4 Fixed the CICD for diffusers (#312) 
Signed-off-by: jingyu <jingyu@omniml.ai>
Signed-off-by: Jingyu Xin <jingyux@nvidia.com>
 2025-09-10 17:03:03 -05:00
Riyad Islam 85b309f15e Qwen quantize and hf export support in examples (#311) 
Signed-off-by: Riyad Islam <rislam@nvidia.com>
 2025-09-10 07:11:10 +00:00
Riyad Islam b913290dfd Making stronglyTyped default for modelopt evaluation (#287) 
Signed-off-by: Riyad Islam <rislam@nvidia.com>
 2025-09-10 02:19:26 +00:00
ajrasane 0adb1b6772 Upgrade to ONNX 1.19.0 (#289) 
Signed-off-by: ajrasane <131806219+ajrasane@users.noreply.github.com>
 2025-09-10 00:40:39 +00:00
Riyad Islam 669ae05a0e CUDNN lib path fix in tensorrt docker (#306) 
Signed-off-by: Riyad Islam <rislam@nvidia.com>
 2025-09-09 23:18:28 +00:00
h-guo18 a6fa34cda4 Feat: update eagle3 example; add export (#293) 
Signed-off-by: h-guo18 <67671475+h-guo18@users.noreply.github.com>
 2025-09-08 18:02:41 -07:00
Keval Morabia cf6f1d4e23 Reduce eagle example test memory usage from 28 to 1 GB (#299) 
Signed-off-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com>
 2025-09-08 21:35:15 +05:30
Zhiyu 78008ac59c Make vlm detection more robust in ptq workflow (#286) 
Signed-off-by: Zhiyu Cheng <zhiyuc@nvidia.com>
 2025-09-04 19:53:17 -07:00