bitsandbytes-foundation/bitsandbytes

Issues

• Cannot install multi backend (ROCm) Wheel version does not match filename

  #1473 opened 4 days ago
  0

• asking for CUDA 124 when only CUDA 126 exists

  #1470 opened 7 days ago
  0

• [BUG] AMD GPU compilation from source

  #1468 opened 9 days ago
  6

• AttributeError: 'NoneType' object has no attribute 'cquantize_blockwise_fp16_fp4'

  #1467 opened 11 days ago
  0

• i am trying to install forge UI but its getting stuck at this

  #1466 opened 13 days ago
  0

• [Ascend][Bug] Fail to use 4-bit quant model to inference

  #1465 opened 17 days ago
  1

• [ROCm] [Stable diffusion LoRA training, sd-scripts] Error invalid device function at line 224 in file /src/csrc/ops.hip

  #1464 opened 20 days ago
  0

• Deleting StableEmbedding still occupies GPU RAM

  #1462 opened 23 days ago
  0

• [ROCm] Allowing for more flexibility in matching ROCm-specific PyTorch wheels to the installed ROCm version

  #1461 opened 25 days ago
  0

• bitsandbytes for macos M1,M2,M3 chips

  #1460 opened 25 days ago
  1

• Link to code for reproducing table found in Multi-backend support (non-CUDA backends) documentation?

  #1456 opened a month ago
  0

• Could not load bitsandbytes native library: libcusparse.so.11: cannot open shared object file: No such file or directory

  #1452 opened a month ago
  2

• LoRA + deepspeed zero3 finetuing using 8bit quantization of base weights results in increased loss

  #1451 opened a month ago
  0

• AttributeError: 'MatmulLtState' object has no attribute 'memory_efficient_backward'

  #1450 opened a month ago
  2

• CUDA Setup failed despite GPU being available.

  #1449 opened a month ago
  3

• Lookahead in forward

  #1447 opened a month ago
  3

• CUDA Setup failed despite GPU being available.

  #1446 opened a month ago
  1

• No access to rocminfo in a production environment - ability to manually set GPU arch.

  #1444 opened a month ago
  1

• CUDA Setup failed despite GPU being available.

  #1443 opened 20 days ago
  2