Commits · 08fa2e7b01dda8959a930295de9829516f8c77bc · OpenDAS / bitsandbytes

08 Nov, 2022 1 commit
- Fixed bug in cpu quant; faster GPU dequant. · 08fa2e7b
  Tim Dettmers authored Nov 07, 2022
  
  08fa2e7b
07 Nov, 2022 2 commits
- Fixed blockwise test and logic. · e0e697b1
  Tim Dettmers authored Nov 06, 2022
  
  e0e697b1
- Added blocksizes 2048, 1024, and 512 to blockwise quant. · 6bc2b992
  Tim Dettmers authored Nov 06, 2022
  
  6bc2b992
06 Nov, 2022 3 commits
- Added k<256 quantile estimate. · 2f2063ba
  Tim Dettmers authored Nov 06, 2022
  
  2f2063ba
- Added k-bit fp8 map. · 98cbc4bc
  Tim Dettmers authored Nov 06, 2022
  
  98cbc4bc
- Added k-bit linear quantization. · caf18325
  Tim Dettmers authored Nov 06, 2022
  
  caf18325
04 Nov, 2022 1 commit
- Added FP8 quantization map. · 1efb87d8
  Tim Dettmers authored Nov 03, 2022
  
  1efb87d8
25 Oct, 2022 1 commit
- Added CUDA SETUP instruction generator. · a371be30
  Tim Dettmers authored Oct 25, 2022
  
  a371be30
24 Oct, 2022 1 commit
- Isolated CUDASetup logging; all tests green. · df86625a
  Tim Dettmers authored Oct 24, 2022
  
  df86625a
20 Sep, 2022 3 commits
- try fp32 · 76ce9aa6
  justheuristic authored Sep 20, 2022
  
  76ce9aa6
- set threshold · 292a4787
  Tim Dettmers authored Sep 20, 2022
  
  292a4787
- review · a07825ac
  justheuristic authored Sep 20, 2022
  
  a07825ac
17 Sep, 2022 22 commits
- cast device · cff3a715
  justheuristic authored Sep 18, 2022
  
  cff3a715
- cast device · 32a9a88f
  justheuristic authored Sep 18, 2022
  
  32a9a88f
- cast device · 01b4c6a0
  justheuristic authored Sep 18, 2022
  
  01b4c6a0
- cast device · e4086a27
  justheuristic authored Sep 18, 2022
  
  e4086a27
- cast device · 725cc729
  justheuristic authored Sep 18, 2022
  
  725cc729
- cast before allclose · 28a9313d
  justheuristic authored Sep 18, 2022
  
  28a9313d
- cast before allclose · 95dafc64
  justheuristic authored Sep 18, 2022
  
  95dafc64
- debug · 37f805bb
  justheuristic authored Sep 18, 2022
  
  37f805bb
- pre-cast · 6a826c41
  justheuristic authored Sep 18, 2022
  
  6a826c41
- debug · d9b87898
  justheuristic authored Sep 18, 2022
  
  d9b87898
- run backward · 2cd047e3
  justheuristic authored Sep 18, 2022
  
  2cd047e3
- add memory efficient backward · 591f6039
  justheuristic authored Sep 18, 2022
  
  591f6039
- bump threshold to 0.21 · f6670329
  justheuristic authored Sep 18, 2022
  
  f6670329
- more lenient threshold · fa8e07c7
  justheuristic authored Sep 18, 2022
  
  fa8e07c7
- cast properly · e35e2c66
  justheuristic authored Sep 18, 2022
  
  e35e2c66
- un-fuse bias · d9ca0ed9
  justheuristic authored Sep 17, 2022
  
  d9ca0ed9
- copypaste tolerances · 7facedda
  justheuristic authored Sep 17, 2022
  
  7facedda
- clearer assertions · e29c5f5c
  justheuristic authored Sep 17, 2022
  
  e29c5f5c
- check dtypes first · 9379df85
  justheuristic authored Sep 17, 2022
  
  9379df85
- check dtypes first · 140cdbe8
  justheuristic authored Sep 17, 2022
  
  140cdbe8
- cast to half before double_quant · a9c7953e
  justheuristic authored Sep 17, 2022
  
  a9c7953e
- test_bf16 · 469d5a63
  justheuristic authored Sep 17, 2022
  
  469d5a63
13 Sep, 2022 1 commit
- Fixed cpu blockwise quantization for small input tensors. · c05dd42d
  Tim Dettmers authored Sep 13, 2022
  
  c05dd42d
11 Sep, 2022 1 commit
- Fixed 2^31 max size issue for cpu blockwise quant. · 19a7adca
  Tim Dettmers authored Sep 11, 2022
  
  19a7adca
23 Aug, 2022 1 commit
- Some initial code. Needs to be tested. · 7e0fb655
  Tim Dettmers authored Aug 23, 2022
  
  7e0fb655
17 Aug, 2022 1 commit
- Fixed bug in Linear8bitLt, when the bias is None. · 9d60b3c5
  Tim Dettmers authored Aug 17, 2022
  
  9d60b3c5
16 Aug, 2022 2 commits
- Added fused bias to matmullt. · de354f7d
  Tim Dettmers authored Aug 16, 2022
  
  de354f7d
- Added fused bias in dequant_mm. · dede3430
  Tim Dettmers authored Aug 16, 2022
  
  dede3430