-
code
```python
'''
conda activate beyond_scale_2_unsloth
'''
import torch
from datasets import load_dataset
from trl import SFTConfig, SFTTrainer
from unsloth import FastLanguageModel
from tr…
-
Hi,
I encountered an issue after updating to unsloth=="2024.11.6". When training the `Qwen2.5-0.5B-Instruct` model without PEFT, I observed that the model's gradient norm is 0, resulting in no weig…
-
I tried to write a custom trainer based on SFTTrainer and trained with unsloth. The code snippets is:
```
# customize from SFTTrainer
class CustomTrainer(SFTTrainer):
def compute_loss(self,…
-
Here is the code it gives .... EVERY NOTEBOOK I tried.
---------------------------------------------------------------------------
NameError Traceback (most recent…
-
I want to do only training (lora is fine) for the head of the network, how do I do that? I get this error:
```bash
(beyond_scale_2_unsloth) brando9@ampere1~/beyond-scale-2-alignment-coeff $ python /…
-
I am performing a Mega Merge using LLaMA 3.2 3B, both the base model and fine-tuning/instruction tuning, with the DARE linear method. Following the successful completion of the initial merge, I encoun…
-
运行资源：
模型：Qwen2.5-32B-Instruct
数据集：自定义数据集
单卡运行脚本：
微调方式：Qlora
CUDA_VISIBLE_DEVICES=0 \
swift sft \
--model_type qwen2_5-32b-instruct \
--model_id_or_path /hy-tmp/model/Qwen/Qwen2.5-32B-I…
-
Unsloth: Merging 4bit and LoRA weights to 16bit...
Unsloth: Will use up to 23.73 out of 50.99 RAM for saving.
100%|██████████| 32/32 [00:19 4 if True: model.push_to_hub_gguf("mINE", tokenizer, quant…
-
### ⚠️ Please check that this feature request hasn't been suggested before.
- [X] I searched previous [Ideas in Discussions](https://github.com/OpenAccess-AI-Collective/axolotl/discussions/categories…
-
To repro:
Install the latest versions of unsloth and transformers
```
!pip uninstall unsloth -y && pip install --upgrade --no-cache-dir "unsloth[colab-new] @ git+https://github.com/unslothai/unslot…