-
### Reminder
- [X] I have read the README and searched the existing issues.
### Reproduction
Here is my Training & Eval & Prediction script
The training for LoRA is done with Quantization …
-
After saving the fine tuned model using the finetune.py script, running the inference script gives the above error on the line
```python
model=model.merge_and_unload()
```
Inference script:
``…
-
Hi, I see the https://github.com/lvwerra/trl/tree/main/examples/stack_llama/scripts and found it a good RLHF tutorial. However, there are some steps I can't figure out.
The first step is "Supervise…
-
Hello, I am reproducing your project. I have followed the instructions and installed the required environment correctly using conda.
My configuration file is as follows.
```yml
ps_ip: 10.128.…
-
## Environment info
- `transformers` version:
- Platform:
- Python version:
- PyTorch version (GPU?):
- Tensorflow version (GPU?):
- Using GPU in script?:
- Using distributed or parallel se…
-
### 🐛 Describe the bug
Meet the error while trying function backward() in IRM.
`RuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation: [torch.F…
-
### 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…
-
### 🐛 Bug
Int8 finetuning throwing a type error
I'm trying to finetune EleutherAI/pythia-2.8b-deduped model with oasst dataset on a machine with 8 V100 GPUs.
Only the following parameters…
-
### Is your feature request related to a problem? Please describe.
The functions defined in `optimize/_differentiable_functions.py` all cast the inputs to float64 before evaluation. This is a probl…
-
Just reviving the question/issue in #165 to address some doubts of mine:
- how can the `eg_loss` be negative? i have some occurences during training and reading the paper i could not find an explan…