zxdu20 (Zhengxiao Du) – Community Activity

New activity in THUDM/chatglm2-6b 6 months ago

有pro版本开源了吗？

#95 opened 6 months ago by

LycheeX

New activity in THUDM/chatglm2-6b 7 months ago

fix when use_cache = False，inference 乱码

#85 opened 7 months ago by

ShiJueXiaofei

New activity in THUDM/chatglm2-6b 10 months ago

MacBook使用本地下载模型为什么会提示OSError: Can't load the configuration of 'your local path'.

3

#36 opened 10 months ago by

Amy0829

Update README.md

#41 opened 10 months ago by

Azzeddine-1

longfish

#40 opened 10 months ago by

zphilip48

Request: DOI

#42 opened 10 months ago by

tradequant

Ggml version please

1

#44 opened 10 months ago by

Hoioi

New activity in THUDM/chatglm-6b 10 months ago

Update README.md for typo

#87 opened 10 months ago by

Bojun-Feng

New activity in THUDM/chatglm2-6b 10 months ago

It is so faaaaaast!

#30 opened 10 months ago by

AIReach

Update README.md

#37 opened 10 months ago by

linoUCV

我做SFT微调时，发现tokenizer save的问题，需要改下

2

#13 opened 10 months ago by

shibing624

RuntimeError: Internal: src/sentencepiece_processor.cc(1101) [model_proto->ParseFromArray(serialized.data(), serialized.size())]

1

#25 opened 10 months ago by

charles2030

这个模型太强大了 , 感受到清华的威力,赞赞赞

#17 opened 10 months ago by

yuerski

请教一下推理使用Multi-Query Attention 是需要在训练的时候就要使用Multi-Query Attention训练么

1

#21 opened 10 months ago by

zhuhai123

0.0.216 版本 , 引入不了from langchain.document_loaders , 报错

2

#18 opened 10 months ago by

yuerski

add default model_type

#10 opened 10 months ago by

songyouwei

可以处理更长的上下文，那么max_length应该设置更长？

3

#15 opened 10 months ago by

chaochaoli

how to suppress warning Setting `pad_token_id` to `eos_token_id`:2 for open-end generation.

4

#5 opened 10 months ago by

huashiyiqike

add missing `model_type`

#9 opened 10 months ago by

songyouwei

Add task to model card for discoverability

#6 opened 10 months ago by

osanseviero

some problem with cpu

2

#3 opened 10 months ago by

zhangbo2008

fix(Tokenizer): fix save_pretrained error

1

#2 opened 10 months ago by

sharpbai

peft微调不兼容

1

#7 opened 10 months ago by

JaheimLee

New activity in THUDM/chatglm-6b about 1 year ago

完整模型参数下载问题

2

#50 opened about 1 year ago by

Yuchangz

New activity in THUDM/chatglm-6b-int4 about 1 year ago

Avoid Loading CPU kernel if User Have a GPU and Cuda Environment

3

#8 opened about 1 year ago by

rdo4920

New activity in THUDM/chatglm-6b about 1 year ago

Why cant chinese people just write comments in English ?

5

#47 opened about 1 year ago by

praff1234

Update modeling_chatglm.py for inputs_embeds

2

#45 opened about 1 year ago by

Xipotzzz

A slim version of chatglm-6b. Image tokens are removed to save memory and computation.

#8 opened about 1 year ago by

silver

Error in loding models

1

#10 opened about 1 year ago by

huvee

[bug] get_position_ids

#11 opened about 1 year ago by

kebo

离线下载了8个model文件

1

#23 opened about 1 year ago by

q113

"Local variable ‘context_length’ referenced before assignment" when tokenizing a sentence

1

#39 opened about 1 year ago by

liliang

Update README.md

1

#25 opened about 1 year ago by

baby12

skip_init causes bugs when finetuning using deepspeed.

1

#37 opened about 1 year ago by

sunworshipper

Can you combine ChatGLM-6B with ChatRWKV and Open-Assistant ?

1

#36 opened about 1 year ago by

anon4364

Create handler.py

1

#18 opened about 1 year ago by

wensun

Cool Project.

#3 opened about 1 year ago by

kanseaveg

这个问题有人遇到过吗？请问一下如何解决，昨天运行的好好的，今天不知道为啥突然维度不对应了

1

#30 opened about 1 year ago by

zxjyes

Fix LogitsProcessor using slim checkpoint

#29 opened about 1 year ago by

bcol

Rename README.md to hey can

#31 opened about 1 year ago by

RavyaloMark

Update slim checkpoint

#28 opened about 1 year ago by

zxdu20

fix typo in use_gmask

#21 opened about 1 year ago by

fzhang

感谢研究团队

1

#22 opened about 1 year ago by

kuroneko5943

Support single integer or empty list as input to decode

#7 opened about 1 year ago by

peakji

fix GLM6BBlock name typo

#20 opened about 1 year ago by

richbrain

Zhengxiao Du

AI & ML interests

Organizations

zxdu20's activity

有pro版本开源了吗？

fix when use_cache = False，inference 乱码

MacBook使用本地下载模型为什么会提示OSError: Can't load the configuration of 'your local path'.

Update README.md

longfish

Request: DOI

Ggml version please

Update README.md for typo

It is so faaaaaast!

Update README.md

我做SFT微调时，发现tokenizer save的问题，需要改下

RuntimeError: Internal: src/sentencepiece_processor.cc(1101) [model_proto->ParseFromArray(serialized.data(), serialized.size())]

这个模型太强大了 , 感受到清华的威力,赞赞赞

请教一下推理使用Multi-Query Attention 是需要在训练的时候就要使用Multi-Query Attention训练么

0.0.216 版本 , 引入不了from langchain.document_loaders , 报错

add default model_type

可以处理更长的上下文，那么max_length应该设置更长？

how to suppress warning Setting `pad_token_id` to `eos_token_id`:2 for open-end generation.

add missing `model_type`

Add task to model card for discoverability

some problem with cpu

fix(Tokenizer): fix save_pretrained error

peft微调不兼容

完整模型参数下载问题

Avoid Loading CPU kernel if User Have a GPU and Cuda Environment

Why cant chinese people just write comments in English ?

Update modeling_chatglm.py for inputs_embeds

A slim version of chatglm-6b. Image tokens are removed to save memory and computation.

Error in loding models

[bug] get_position_ids

离线下载了8个model文件

"Local variable ‘context_length’ referenced before assignment" when tokenizing a sentence

Update README.md

skip_init causes bugs when finetuning using deepspeed.

Can you combine ChatGLM-6B with ChatRWKV and Open-Assistant ?

Create handler.py

Cool Project.

这个问题有人遇到过吗？请问一下如何解决，昨天运行的好好的，今天不知道为啥突然维度不对应了

Fix LogitsProcessor using slim checkpoint

Rename README.md to hey can

Update slim checkpoint

fix typo in use_gmask

感谢研究团队

最新的commit好像有bug，但是不确定该怎么修复

Support single integer or empty list as input to decode

fix GLM6BBlock name typo