Ecosyste.ms: Issues

An open API service for providing issue and pull request metadata for open source projects.

GitHub / facebookresearch/LLM-QAT issues and pull requests

#32 - Does LLM-QAT support group-wise quantization?

Issue - State: open - Opened by mxjmtxrm 3 months ago

#31 - 运行报错elif self.deepspeed:

Issue - State: open - Opened by LSB0798 4 months ago

#30 - How to run inference

Issue - State: open - Opened by wangkuiyi 4 months ago

#29 - Looks like an Incorrect READ ME file

Issue - State: closed - Opened by gdsaikrishna 5 months ago - 1 comment

#29 - Looks like an Incorrect READ ME file

Issue - State: closed - Opened by gdsaikrishna 5 months ago - 1 comment

#28 - Question about the training cost

Issue - State: closed - Opened by KimythAnly 5 months ago - 1 comment

#28 - Question about the training cost

Issue - State: closed - Opened by KimythAnly 5 months ago - 1 comment

#27 - Does this method support chat models as well as Llama-2 models?

Issue - State: open - Opened by Saoyu99 8 months ago - 1 comment

#26 - How long will it take to train

Issue - State: open - Opened by XA23i 12 months ago - 1 comment

#25 - Accuracy

Issue - State: open - Opened by yileijin about 1 year ago

#24 - Suggest change the README

Issue - State: closed - Opened by jingyao-zhang about 1 year ago - 1 comment

#23 - FileNotFoundError: [Errno 2] No such file or directory: 'wiki2.jsonl'

Issue - State: open - Opened by StiphyJay about 1 year ago - 1 comment

#22 - Training is not working.

Issue - State: open - Opened by XinnuoXu about 1 year ago

#21 - Hi, 可以开源你们生成的训练数据吗，感谢！

Issue - State: open - Opened by Xingrun-Xing about 1 year ago - 1 comment

#20 - docs: blockquote cite article format README

Pull Request - State: closed - Opened by guspan-tanadi about 1 year ago
Labels: cla signed

#19 - Inconsistent results with LLM.int8() and SmoothQuant papers

Issue - State: closed - Opened by fxmarty about 1 year ago - 1 comment

#18 - Questions about the valid_dataset format

Issue - State: closed - Opened by TravisL24 about 1 year ago - 1 comment

#17 - Questions about the valid_dataset format

Issue - State: closed - Opened by TravisL24 about 1 year ago

#16 - run run_train.sh, CUDA out of memory

Issue - State: closed - Opened by priscilla-pan about 1 year ago

#15 - no smoothquant in QuantizeLinear

Issue - State: closed - Opened by priscilla-pan about 1 year ago - 5 comments

#14 - Is there an efficient way to generate data?

Issue - State: closed - Opened by benyang0506 over 1 year ago - 3 comments

#13 - The choice of kd_loss_scale

Issue - State: closed - Opened by zhanlaoban over 1 year ago - 1 comment

#12 - can you provide inference example for QuantizeLinear in 8 8 8

Issue - State: closed - Opened by jackzhou121 over 1 year ago - 2 comments

#11 - How should I save the 8 bit model？

Issue - State: closed - Opened by liguodongiot over 1 year ago - 2 comments

#10 - If 4-8-8 is used to do QAT, how to process weight in inference?

Issue - State: closed - Opened by jackzhou121 over 1 year ago - 1 comment

#9 - No randomization operation for the first token in data generation phrase.

Issue - State: closed - Opened by xingyueye over 1 year ago - 1 comment

#8 - Why use clip_tensor[-2.0, 2.0] in the backward?

Issue - State: closed - Opened by jackzhou121 over 1 year ago - 1 comment

#7 - APEX and FSDP can not run

Issue - State: closed - Opened by jackzhou121 over 1 year ago - 1 comment

#6 - why generated data was not used

Issue - State: closed - Opened by jackzhou121 over 1 year ago - 1 comment

#5 - Does this method support Bloom？

Issue - State: closed - Opened by 18140663659 over 1 year ago - 1 comment

#4 - Expects full precision but got torch.bfloat16 error

Issue - State: open - Opened by liguodongiot over 1 year ago - 1 comment

#3 - Harcoded train paths and configuration for table in readme

Issue - State: closed - Opened by aitorormazabal over 1 year ago - 1 comment

#2 - Why do smoothquant dynamically in the forward() function of QuantizeLinear layer

Issue - State: closed - Opened by Starmys over 1 year ago - 2 comments

#1 - Change teaser image to relative path to properlly display on Github

Pull Request - State: closed - Opened by Lyken17 over 1 year ago
Labels: cla signed