issues
search
QwenLM
/
Qwen
The official repo of Qwen (通义千问) chat & pretrained large language model proposed by Alibaba Cloud.
Apache License 2.0
13.59k
stars
1.11k
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
[BUG] <title>qwen1.8b模型lora微调后, 用样例代码来batch推理, batch_size 为1和batch_size 为10 推理结果不一致,已设置num_beam=1, do_sample=False依然失效
#1031
xiaoduozhou
closed
7 months ago
2
[BUG] <title>lora微调14b-chat可以用deepspeed zreo 3 吗?显卡是8 x V100(16G),不管是zero2还是zero3 都存在OOM
#1030
yyyzhao
closed
6 months ago
3
[BUG] <title>qwen model.generate 中添加 max_new_tokens 参数不生效, 新增生成的还是超出了最大长度
#1029
xiaoduozhou
closed
8 months ago
2
During vllm_gptq inference, when tensor_parallel_size is greater than 2, an NCLL error occurs. There are eight A100 GPUs, and I'm not sure what the issue is.
#1028
Leosgp
closed
6 months ago
1
init commit of recipes
#1027
chywang
closed
8 months ago
0
求教!微调后答案不稳定
#1025
wells-Qiang-Chen
closed
7 months ago
16
[BUG] 使用Zero3进行qwen72B 全参数训练报错
#1022
Double-bear
closed
8 months ago
2
[BUG]使用官方profile.py测试qwen-72b-int4模型,速度非常慢,远未达到官方的速度
#1021
ArlanCooper
closed
6 months ago
8
[BUG] <title>72b_int8模型加载问题
#1020
bimver
closed
6 months ago
5
[BUG] RuntimeError: FlashAttention only supports Ampere GPUs or newer.
#1019
dhcn
closed
8 months ago
4
How to fine tune the model to complete information extraction tasks
#1018
jjl001
closed
6 months ago
1
qwen性能测试,针对fp32、bf16和int4不同精度
#1017
MuyeMikeZhang
closed
6 months ago
4
[BUG] <title>peft的p-tuning_v2无法使用,维度没有对齐
#1016
ssgg-code
closed
7 months ago
1
[BUG] Docker: ValueError: Tokenizer class QWenTokenizer does not exist or is not currently imported
#1015
tzdmwwdz
closed
8 months ago
3
请问是否支持增量预训练呢?
#1013
listwebit
closed
6 months ago
1
💡 [REQUEST] - 官方是否用TensorRT-LLM 跑过 Qwen-14b-chat-Int4 模型?效果如何?提速几倍?
#1012
micronetboy
closed
6 months ago
2
[BUG] <title>使用run_gptq.py 量化7b模型后,回答乱码
#1011
liukangjia666
closed
5 months ago
2
How to set model context length in FastChat openai_api_server for Qwen-72B-Chat
#1010
Haxeebraja
closed
8 months ago
4
使用AutoModelForCausalLM加载7B,调用chat_stream报错ValueError: too many values to unpack (expected 2)
#1009
kunzeng-ch
closed
6 months ago
2
[BUG] 运行官方profile.py qwen-72b-int4报错
#1008
ArlanCooper
closed
8 months ago
3
微调之后无法进行多轮对话
#1007
DuBaiSheng
closed
8 months ago
4
[BUG] 你好,请问qwen的tokenizer都是slow吗,有没有fast的tokenizer?
#1006
berooo
closed
8 months ago
1
About Embedding Layer
#1005
OliverHuang1220
closed
8 months ago
2
[BUG] 单卡多机微调1.8bchat卡住
#1004
128Ghe980
closed
8 months ago
5
[BUG] <3090 12G 单卡Lora微调1.4B模型OOM>
#1002
lizhili
closed
8 months ago
3
[BUG] chat_stream时指定generation_config的eos_token_id为列表[198, 151643]时会报错。198是换行符,想让模型遇到换行符就停止生成。
#1001
chenyzh28
closed
8 months ago
2
[BUG] <单机多卡使用finetune_qlora_ds.sh训练生成的/checkpoint-3/global_step3有什么用?/>
#1000
WangxuP
closed
8 months ago
1
求教,如何进行一些否定性质的微调。
#999
cdisk
closed
6 months ago
1
14b模型能微调支持32K吗
#998
Longleaves
closed
7 months ago
1
qwen-14b-chat微调怎么在数据集加入自己的system角色
#997
qianliyx
closed
5 months ago
3
将qwen1.8b导出为onnx后,对话时总是会出现イベ,是为什么呢?
#996
xman1991
closed
5 months ago
4
[BUG] <关于使用finetune_lora_ds.sh显存爆炸的问题>
#995
yoohhwz
closed
5 months ago
2
[BUG] 惊!关于显存的问题<title>
#994
xx-Jiangwen
closed
7 months ago
3
请问通义千问模型能支持外挂知识库吗,如果能该怎么实现,有大量的EXCEL表格数据。
#993
chesp
closed
6 months ago
3
[BUG] Qwen72B-Chat基于llama.cpp推理,答案末尾总是追加[PAD151645]
#992
zhengxingmao
closed
8 months ago
1
Is qwen suitable for sentence embedding for Chinese?[issue] <title>
#990
rocke2020
closed
8 months ago
1
QWen 14B 扩充上下文长度 大概在什么时间能发布
#989
lyman-meng
closed
8 months ago
0
[BUG] 72B-Chat版本微调时lr一直是0?loss不正常。
#988
chenyzh28
closed
8 months ago
1
[72B-Chat LoRA微调报错] AssertionError: {'id': 643, 'status': 'NOT_AVAILABLE', 'numel': 0, 'ds_numel': 0, 'shape': (0,), 'ds_shape': (0,), 'requires_grad': True, 'grad_shape': None, 'persist': True, 'active_sub_modules': {7}, 'ds_tensor.shape': torch.Size([0])}
#987
chenyzh28
closed
8 months ago
0
[BUG] 关于Qwen使用工具的问题
#985
edisonzf2020
closed
5 months ago
2
[BUG] <title>attention_mask改为None后,输出只会是英文
#984
xman1991
closed
6 months ago
1
💡 [REQUEST] - 您好,请教一个问题,微调的时候,一般需要Loss 下降到多少?
#983
micronetboy
closed
5 months ago
1
单块A800推理7B模型和14B模型的速度差异巨大,正常否?不正常改如何改进?[BUG] <title>
#982
OveSteve
closed
8 months ago
1
Qwen-1.8B模型SFT后推理速度变慢
#981
potter2010
closed
6 months ago
1
💡 [REQUEST] - 千问有支持代码链的计划吗,隔壁GLM4的代码链印象深刻
#980
liaoweiguo
closed
6 months ago
3
[BUG] <Qwen最新版本存在causal_mask bug,导致在有kv_cache的情况下,多个tokens的输入会得到不同的结果>
#979
hzjane
closed
7 months ago
5
[BUG] libcudart.so.12: cannot open shared object file: No such file or directory
#978
ArlanCooper
closed
8 months ago
5
[BUG] <title> Qwen的推理速度过于慢了
#977
zrLian
closed
6 months ago
1
[BUG] 当输入内容超过4096后,系统提示词会失效
#976
homelee
closed
6 months ago
1
[BUG] <用vllm_wrapper加载模型时,system指令跟随比较飘忽>
#975
chopin1998
closed
6 months ago
2
Previous
Next