Issues

• About collect training data. Encounter error about Bus

  FMInference/DejaVu#37 · Filter by author Phan1000-7 opened on Jun 7, 2025

• lm_head.weight in convert_opt_checkpoint.py

  FMInference/DejaVu#36 · Filter by author quaternior opened on Oct 31, 2024

• How to calculate Activated Neurons

  FMInference/DejaVu#34 · Filter by author LiuJinzhe-Keepgoing opened on Sep 20, 2024

• No such file ./trace_json/xxxxx

  FMInference/DejaVu#33 · Filter by author MathsCode opened on Sep 4, 2024

• How to test the sparsity of the model？

  FMInference/DejaVu#32 · Filter by author lw20191012 opened on Aug 21, 2024
  2

• Training of predictors based on MOE model

  FMInference/DejaVu#31 · Filter by author mailonghua opened on Jul 23, 2024

• Question about downsteam tasks accuracy

  FMInference/DejaVu#30 · Filter by author W-215 opened on Jul 10, 2024

• the issue of parameter settings

  FMInference/DejaVu#29 · Filter by author W-215 opened on Jun 7, 2024

• How to run on a single GPU?

  FMInference/DejaVu#28 · Filter by author xbzjsj opened on Apr 17, 2024
  7

• PyTorch 1.12 and flash-attn==0.2.8 are not compatible.

  FMInference/DejaVu#27 · Filter by author heheda12345 opened on Apr 10, 2024

• the code about collect activations while inferencing

  FMInference/DejaVu#25 · Filter by author mli-tian opened on Apr 9, 2024
  1

• How to run the code with smaller ckpt like opt-6.7B

  FMInference/DejaVu#24 · Filter by author chenzq-98 opened on Apr 6, 2024
  14

• A question about attention block sparsity

  FMInference/DejaVu#23 · Filter by author imh966 opened on Mar 22, 2024

• Miss csrc folder when building docker images

  FMInference/DejaVu#19 · Filter by author Mi5sssss opened on Feb 2, 2024
  3

• Does anyone know if this work is implemented on llama?

  FMInference/DejaVu#18 · Filter by author wyxscir opened on Jan 25, 2024
  6

• Please give a instruction to run Hardware-efficient Implementation dejaVu model inferenc,I Can't find the moudle src.models.gpt_sparse

  FMInference/DejaVu#17 · Filter by author wangzhenyu32 opened on Jan 23, 2024

• Question about OPT-30B

  FMInference/DejaVu#16 · Filter by author Jimskns opened on Jan 4, 2024

• Missing full.pt when running latency benchmark

  FMInference/DejaVu#15 · Filter by author PaulWang0513 opened on Dec 23, 2023
  1

• LLama 2 compatability

  FMInference/DejaVu#14 · Filter by author yomanFX opened on Nov 6, 2023

• Some questions about implementation details

  FMInference/DejaVu#13 · Filter by author SUDA-HLT-ywfang opened on Oct 30, 2023
  3

• Failed to collect training data for sparsity predictor

  FMInference/DejaVu#11 · Filter by author PaulWang0513 opened on Oct 7, 2023
  9

• Compare the sparse model on downstream tasks

  FMInference/DejaVu#10 · Filter by author iidsample opened on Oct 2, 2023
  2

• Missing lm_eval harness for downstream task prediction

  FMInference/DejaVu#9 · Filter by author iidsample opened on Oct 2, 2023

• Inconsisten File Names

  FMInference/DejaVu#8 · Filter by author iidsample opened on Oct 2, 2023

• Missing Argument and Filtering

  FMInference/DejaVu#7 · Filter by author iidsample opened on Sep 30, 2023