-
微博内容精选
-
On a M1 Max 32GB I can load mistral or phi2 but when I look at RAM usage I see strange things:
- Mistral 7B occupies ~14GB as it should 👍
- Phi2 ~20GB. Why? 👎
Also when trying to save a model…
-
### Describe the bug
Whenever I try to train a LoRA model, I get following error:
```
Traceback (most recent call last):
File “C:\Users\vstil\Downloads\oobabooga_windows\oobabooga_windows\text…
-
Hi, I'm trying to quantize my mixtral model after finetuning.
After the training session, i do model.merge_and_unload, then save it and reload it with AutoAWQForCausalLM
I then follow the procedure …
-
### 🐛 Describe the bug
Error messages generated
> Traceback (most recent call last):
File "C:\Users\zynas\AppData\Local\Programs\Python\Python39\lib\runpy.py", line 197, in _run_module_as_main
…
-
### 🐛 Bug
I trained a model based on circulus/Llama-2-7b-orca-v1 and exported:
- Model
- Validation predictions
- Logs
However, I'm currently having troubles reproducing **the exact output*…
-
UPDATE (08/09/2023):
We have done major performance overhaul in the past few months, and now I'm happy to share the latest results:
- SOTA performance on CUDA: https://github.com/mlc-ai/llm-perf-b…
-
### Search before asking
- [X] I have searched the YOLOv8 [issues](https://github.com/ultralytics/ultralytics/issues) and [discussions](https://github.com/ultralytics/ultralytics/discussions) and fou…
-
https://github.com/modelscope/swift/blob/main/docs/source_en/Multi-Modal/glm4v-best-practice.md
使用的是文档中的命令,eval_acc一直没有变化
# Experimental environment: A100
# 40GB GPU memory
CUDA_VISIBLE_DEVICE…
-
### Issue type
Bug
### Have you reproduced the bug with TensorFlow Nightly?
Yes
### Source
source
### TensorFlow version
2.13.0
### Custom code
Yes
### OS platform and distribution
Ubuntu …