下载速度太慢 python3 -m dfss [email protected]:/ext_model_information/LLM/LLM-TPU/deepseek-r1-distill-qwen-7b.zip

[wcool1]

你好，请问 python3 -m dfss --url=[email protected]:/ext_model_information/LLM/LLM-TPU/deepseek-r1-distill-qwen-7b.zip下载速度太慢如何解决
参考 https://github.com/sophgo/LLM-TPU/tree/main/models/language_model/python_demo

[chuxiaoyi2023]

之前发过网盘，你可以在issue找 #75
#75

[xuyang1102]

之前发过网盘，你可以在issue找 #75 #75

deepseek r1，14b模型有没有长上下文已转换好的，512的太短了，还没回答完就停了

[wcool1]

之前发过网盘，你可以在issue找 #75 #75

谢谢您，我终于下好模型并解压放在了相应的文件夹中，为什么执行ython3 pipeline.py --devid 0 --dir_path ./deepseek-r1-distill-qwen-7b 出现了错误

/data/VAProject/LLM-TPU-main/models/language_model/python_demo$ python3 pipeline.py --devid 0 --dir_path ./deepseek-r1-distill-qwen-7b/

Initializing devices...
Device [ 0 ] loading .....
bmcpu init: skip cpu_user_defined
open usercpu.so, init user_cpu_init
Model[./deepseek-r1-distill-qwen-7b/qwen2_w4bf16_seq2048_20250131_203910.bmodel] loading ....
[BMODEL][ModelCtx:520] FATAL: File[./deepseek-r1-distill-qwen-7b/qwen2_w4bf16_seq2048_20250131_203910.bmodel] is broken ...
Traceback (most recent call last):
File "pipeline.py", line 393, in
main(args)
File "pipeline.py", line 374, in main
model = Model(args)
File "pipeline.py", line 40, in init
self.load_model(args, read_bmodel=True)
File "pipeline.py", line 139, in load_model
self.model.init(self.devices, model_path, read_bmodel) # when read_bmodel = false, not to load weight, reuse weight
RuntimeError: can not load bmodel correctly

[wcool1]

1.5b的运行成功了，7b的解压不了

/data/VAProject/LLM-TPU-main/models/language_model/python_demo$ unzip deepseek-r1-distill-qwen-7b.zip
Archive: deepseek-r1-distill-qwen-7b.zip
End-of-central-directory signature not found. Either this file is not
a zipfile, or it constitutes one disk of a multi-part archive. In the
latter case the central directory and zipfile comment will be found on
the last disk(s) of this archive.
unzip: cannot find zipfile directory in one of deepseek-r1-distill-qwen-7b.zip or
deepseek-r1-distill-qwen-7b.zip.zip, and cannot find deepseek-r1-distill-qwen-7b.zip.ZIP, period.

[stillbanbo]

那试试这个 1.5B是上面的没变，7B是慢慢拉过来的
链接: https://pan.baidu.com/s/1UqkTp-1c9y9pFDIIA35NHQ 提取码: u1iu 复制这段内容后打开百度网盘手机App，操作更方便哦

[wcool1]

那试试这个 1.5B是上面的没变，7B是慢慢拉过来的 链接: https://pan.baidu.com/s/1UqkTp-1c9y9pFDIIA35NHQ 提取码: u1iu 复制这段内容后打开百度网盘手机App，操作更方便哦

好的，谢谢，1.5B是可以跑的，但我还是想请问一下，这种下载速度极慢的问题该怎么解决，因为最近在使用AIbox1684x,很多时候都会遇到这样的问题，比如python3 -m dfss --url=[email protected]:sophon-demo/Whisper/mlir/tpu-mlir_v1.6.135-g12c3f90d8-20240327.tar.gz，也想试一下7B、14B等等别的模型；