Commits · 46415f5aa5b47aed5c638f286b89c4496903cceb · OpenDAS / AutoAWQ

04 Jan, 2024 1 commit
- Add CPU-loaded multi-GPU quantization (#289) · 46415f5a
  Blake Wyatt authored Jan 04, 2024
  
  46415f5a
22 Dec, 2023 1 commit
- Mixtral: Mixture of Experts quantization (#251) · 5b9f3c47
  Casper authored Dec 22, 2023
  
  5b9f3c47
16 Dec, 2023 1 commit
- Fix quantization issue with transformers >= 4.36.0 (#264) · 2350a4d0
  Younes Belkada authored Dec 16, 2023
  
  2350a4d0
11 Dec, 2023 2 commits
- FEAT: Add possibility of skipping modules when quantizing (#248) · 9c3dfa07
  Younes Belkada authored Dec 11, 2023
  
  9c3dfa07
- [`core`] Fix quantization issues with transformers==4.36.0 (#249) · 78b59d73
  Younes Belkada authored Dec 11, 2023
  
  78b59d73
28 Nov, 2023 1 commit
- New scaling to improve perplexity (#216) · d1112e1c
  Casper authored Nov 28, 2023
  
  d1112e1c
20 Nov, 2023 1 commit
- Robust quantization for Catcher (#209) · 63d2aaec
  Casper authored Nov 20, 2023
  
  63d2aaec
16 Nov, 2023 1 commit
- Fixed multi-GPU quantization (#196) · 6f516b8d
  Casper authored Nov 16, 2023
  
  6f516b8d
02 Nov, 2023 1 commit
- Fix init quant (#136) · 89f3feae
  Casper authored Nov 02, 2023
  
  89f3feae
29 Oct, 2023 1 commit
- Pseudo dequantize function (#127) · f1a2e56e
  Casper authored Oct 29, 2023
  
  f1a2e56e
28 Oct, 2023 1 commit
- Fixing starcoder based models with 15B (#118) · 6c05f669
  Sebastian Bodza authored Oct 28, 2023
  
  6c05f669
24 Sep, 2023 1 commit
- Fix typing classes · 4abfefc9
  Vik Paruchuri authored Sep 23, 2023
  
  4abfefc9
20 Sep, 2023 11 commits
- Remove cosine loss · 90bf52c7
  Casper Hansen authored Sep 20, 2023
  
  90bf52c7
- Correct comments · 9848b6a4
  Casper Hansen authored Sep 20, 2023
  
  9848b6a4
- Better comments. Implement cosine similarity · c9d01ac3
  Casper Hansen authored Sep 20, 2023
  
  c9d01ac3
- Change filename · 2290bdb2
  Casper Hansen authored Sep 20, 2023
  
  2290bdb2
- Fix comment · 44b3187a
  Casper Hansen authored Sep 20, 2023
  
  44b3187a
- Fix small performance regression · 0f39807b
  Casper Hansen authored Sep 20, 2023
  
  0f39807b
- Remove old quantization code · d3550fec
  Casper Hansen authored Sep 20, 2023
  
  d3550fec
- Working for OPT · 724bda58
  Casper Hansen authored Sep 20, 2023
  
  724bda58
- Make code cleaner · 356cbc92
  Casper authored Sep 20, 2023
  
  356cbc92
- Add desc · 9204d6e8
  Casper authored Sep 20, 2023
  
  9204d6e8
- Initial quantization refactoring · 45c22ee5
  Casper authored Sep 20, 2023
  
  45c22ee5
19 Aug, 2023 1 commit
- Rename q_config -> quant_config. Include w_bit in quant_config. Save quant_config.json. · d35ade75
  Casper Hansen authored Aug 19, 2023
  
  d35ade75
17 Aug, 2023 4 commits
- Cleanup code · 35ac58c7
  Casper Hansen authored Aug 17, 2023
  
  35ac58c7
- Refactor entry.py [WIP] · 5d4ab5dc
  Casper Hansen authored Aug 17, 2023
  
  5d4ab5dc
- Refactored pre_quant into base · 290f45e7
  Casper Hansen authored Aug 17, 2023
  
  290f45e7
- Initial refactor [WIP]. · 7e091fb1
  Casper authored Aug 17, 2023
  
  7e091fb1
26 Jul, 2023 1 commit
- [Minor] accelerate loading quantized model · f22c2a35
  Jiaming Tang authored Jul 26, 2023
  
  f22c2a35
19 Jul, 2023 1 commit
- Memory optimization · ba01560f
  Abhinav Kulkarni authored Jul 11, 2023
  
  ba01560f
04 Jul, 2023 1 commit
- [Minor] Added model dispatch to GPU logic · df0c600c
  Abhinav Kulkarni authored Jul 04, 2023
  
  df0c600c
01 Jul, 2023 1 commit
- [Major] Add CPU offloading support for apply_scale, apply_clip,... · 95cd9c2d
  Abhinav Kulkarni authored Jul 01, 2023
```
[Major] Add CPU offloading support for apply_scale, apply_clip, pseudo_quantize_model_weight, real_quantize_model_weight
```
  95cd9c2d
25 Jun, 2023 1 commit
- [Major] Add support for BLOOM, MPT and Falcon. · 71d8e68d
  Haotian Tang authored Jun 25, 2023
  
  71d8e68d
01 Jun, 2023 1 commit
- first commit · 460d6d45
  Ji Lin authored May 31, 2023
  
  460d6d45