Commits · d5bb4ec8c883e617297117178ab3055aa77ce48d · OpenDAS / AutoAWQ

02 Oct, 2023 3 commits
- Mistral fused modules · d5bb4ec8
  Casper Hansen authored Oct 02, 2023
  
  d5bb4ec8
- Fix Falcon n_kv_heads parameter · 80eea43e
  Casper Hansen authored Oct 02, 2023
  
  80eea43e
- Fix unexpected keyword · d016c513
  Casper Hansen authored Oct 02, 2023
  
  d016c513
01 Oct, 2023 2 commits
- Merge pull request #84 from casper-hansen/fix_ghwf · 3fa74007
  Casper authored Oct 01, 2023
```
Faster build, fix "no space left".
```
  3fa74007
- Faster build, fix "no space left". · 8f05eaef
  Casper authored Oct 01, 2023
  
  8f05eaef
27 Sep, 2023 15 commits
- Add badges · 20cd7981
  Casper authored Sep 27, 2023
  
  20cd7981
- Bump to v0.1.1 · 6bdcd786
  Casper authored Sep 27, 2023
  
  6bdcd786
- Merge pull request #77 from s4rduk4r/offloading · f220ccf1
  Casper authored Sep 27, 2023
```
Offloading to cpu and disk
```
  f220ccf1
- Merge pull request #80 from casper-hansen/low_cpu_mem_example · 8793a9f7
  Casper authored Sep 27, 2023
```
Add low_cpu_mem_usage=True in example
```
  8793a9f7
- Add low_cpu_mem_usage=True in example · aa6497cd
  Casper Hansen authored Sep 27, 2023
  
  aa6497cd
- Merge pull request #79 from casper-hansen/mistral · 1c5ccc79
  Casper authored Sep 27, 2023
```
Mistral support
```
  1c5ccc79
- Add exception handling · 1ac3f976
  Casper Hansen authored Sep 27, 2023
  
  1ac3f976
- Mistral support · bd56c301
  Casper Hansen authored Sep 27, 2023
  
  bd56c301
- Fix apply_rotary_emb() to have both tensors on the same device · 841a2313
  s4rduk4r authored Sep 27, 2023
  
  841a2313
- Rely on accelerate.dispatch_model() only · ffaaa259
  s4rduk4r authored Sep 27, 2023
  
  ffaaa259
- Merge pull request #75 from casper-hansen/fix_runtime · c57da6b8
  Casper authored Sep 27, 2023
```
Fix KV cache shapes error
```
  c57da6b8
- Clean code · cba9a28c
  Casper Hansen authored Sep 27, 2023
  
  cba9a28c
- Remove unused imports · 707315a6
  Casper Hansen authored Sep 27, 2023
  
  707315a6
- Move cache initialization back to init · 8c80b3e0
  Casper Hansen authored Sep 27, 2023
  
  8c80b3e0
- Roll cache efficiently and correctly · c2cd6535
  Casper Hansen authored Sep 27, 2023
  
  c2cd6535
26 Sep, 2023 6 commits
- Clean · 91de3bfa
  s4rduk4r authored Sep 27, 2023
  
  91de3bfa
- Offload to cpu · 4f76ecb2
  s4rduk4r authored Sep 27, 2023
  
  4f76ecb2
- Fix edge case · 0392a823
  Casper Hansen authored Sep 26, 2023
  
  0392a823
- Add compute_memory_used_pct function · 3bb4a9f6
  Casper Hansen authored Sep 26, 2023
  
  3bb4a9f6
- Automatically reset/increase cache · 202c6d6a
  Casper Hansen authored Sep 26, 2023
  
  202c6d6a
- Merge pull request #69 from VikParuchuri/main · 8eb26eb2
  Casper authored Sep 26, 2023
```
Use typing classes over base types
```
  8eb26eb2
24 Sep, 2023 1 commit
- Fix typing classes · 4abfefc9
  Vik Paruchuri authored Sep 23, 2023
  
  4abfefc9
21 Sep, 2023 13 commits
- Merge pull request #61 from casper-hansen/bigcode · 386fede8
  Casper authored Sep 21, 2023
```
Add GPT BigCode support (StarCoder)
```
  386fede8
- Remove imports · 9060b980
  Casper Hansen authored Sep 21, 2023
  
  9060b980
- Remove fused modules (later) · 37db5759
  Casper Hansen authored Sep 21, 2023
  
  37db5759
- Fix usage of engine · 133dd7a7
  Casper authored Sep 21, 2023
  
  133dd7a7
- Fix module import · fbeea40b
  Casper Hansen authored Sep 21, 2023
  
  fbeea40b
- Bump to v0.1.0 · 872387fc
  Casper authored Sep 21, 2023
  
  872387fc
- Merge pull request #66 from casper-hansen/improve_loading · 6b82cac7
  Casper authored Sep 21, 2023
```
Improve model loading
```
  6b82cac7
- Merge pull request #53 from qwopqwop200/main · d62aebfe
  Casper authored Sep 21, 2023
```
support windows 
```
  d62aebfe
- Include arch by default, disable with INCLUDE_ARCH=0 · 14d4f8cb
  Casper authored Sep 21, 2023
  
  14d4f8cb
- Fix argument · 7c48033a
  Casper Hansen authored Sep 21, 2023
  
  7c48033a
- Improve model loading · 2dada8f8
  Casper Hansen authored Sep 21, 2023
  
  2dada8f8
- remove extra_compile_args for support windows · b621c142
  qwopqwop200 authored Sep 21, 2023
  
  b621c142
- Merge branch 'main' into pr/53 · 81fd57c3
  Casper authored Sep 21, 2023
  
  81fd57c3