facebookresearch/bitsandbytes

Issues

• Support for Tesla Architecture

  #4 opened 3 years ago by nbroad1881
  8

• OSError: We couldn't connect to 'https://huggingface.co' to load this file

  #34 opened a year ago by xwpaul
  0

• is it work on Win platform?

  #7 opened 3 years ago by Nehc
  17

• Can I quantize the tensors to other data type rather than 8bit tensor?

  #33 opened a year ago by kaneyxx
  0

• Quantization functions test fail on Pascal

  #3 opened 3 years ago by TimDettmers
  1

• python setup.py install error

  #18 opened 3 years ago by mathpopo
  11

• Llama?

  #32 opened 2 years ago by generic-username0718
  0

• 'NoneType' object has no attribute 'cdequantize_blockwise_cpu_fp32'

  #31 opened 2 years ago by HumzaSami00
  0

• Unable to install via Pypi and source

  #28 opened 2 years ago by kpriyanshu256
  10

• 8-bit optimizer crashes when fine-tuning gpt2-large

  #26 opened 2 years ago by rationalism
  0

• bfloat16 grads are not supported

  #25 opened 2 years ago by kurumuz
  0

• Did you ever try MNMT systems?

  #15 opened 2 years ago by SirRob1997
  9

• no difference in memory usage

  #22 opened 2 years ago by ofrimasad
  3

• import bitsandbytes as bnb 错误

  #21 opened 2 years ago by zhishui3
  4

• Is AdamW8bit compatible with OSS in fairscale?

  #24 opened 3 years ago by LeeDoYup
  1

• undefined symbol: __fatbinwrap_38_cuda_device_runtime_compute_75_cpp1_ii_8b1a5d37

  #12 opened 3 years ago by SparkJiao
  6

• [Question] Usage of bnb.nn.Embedding with existing classes from other libraries

  #19 opened 3 years ago by LSinev
  2

• bnb.optim.AdamW

  #10 opened 3 years ago by patrickvonplaten
  2

• How to use it in huggingface AdamW optimizer?

  #13 opened 3 years ago by ztl-35
  4

• Optimizer2State: Unsafe use of eval() in __init__

  #8 opened 3 years ago by bcoles
  1

• sparse is not passed in __init__ function of StableEmbedding

  #11 opened 3 years ago by SirRob1997
  1

• errors when training to the third epoch. everytime.

  #20 opened 3 years ago by Dootmaan
  2

• The code uses more GPU memory with Multi-scale Vision Transformers

  #16 opened 3 years ago by JunweiLiang
  2

• Ok

  #17 opened 3 years ago by unusali
  0

• Feature request: Please, add implementation for Novograd algorithm

  #6 opened 3 years ago by LSinev
  1

• undefined symbol: __fatbinwrap_38

  #5 opened 3 years ago by TimDettmers
  2