423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-12-14 19:06:26 +08:00
Code Issues Packages Projects Releases Wiki Activity

[inference] fix stop token for object detection (#6624)

Browse Source 
* fix stop token

* update minicpm data pipeline

* fix npu qlora examples

Former-commit-id: e3e2c8c689
This commit is contained in:
hoshi-hiyouga
2025-01-13 21:34:20 +08:00
committed by GitHub
parent 089c7d5e51
commit d8cba9464f
15 changed files with 101 additions and 45 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
examples/README.md
View File

@@ -109,6 +109,12 @@ USE_RAY=1 llamafactory-cli train examples/train_full/llama3_lora_sft_ray.yaml
llamafactory-cli train examples/train_qlora/llama3_lora_sft_otfq.yaml
```
#### Supervised Fine-Tuning with 4-bit Bitsandbytes Quantization on Ascend NPU
```bash
llamafactory-cli train examples/train_qlora/llama3_lora_sft_bnb_npu.yaml
```
#### Supervised Fine-Tuning with 4/8-bit GPTQ Quantization
```bash

6
examples/README_zh.md
View File

@@ -109,6 +109,12 @@ USE_RAY=1 llamafactory-cli train examples/train_full/llama3_lora_sft_ray.yaml
llamafactory-cli train examples/train_qlora/llama3_lora_sft_otfq.yaml
```
#### 在 NPU 上基于 4 比特 Bitsandbytes 量化进行指令监督微调
```bash
llamafactory-cli train examples/train_qlora/llama3_lora_sft_bnb_npu.yaml
```
#### 基于 4/8 比特 GPTQ 量化进行指令监督微调
```bash

2
examples/train_qlora/llama3_lora_sft_otfq_npu.yaml → examples/train_qlora/llama3_lora_sft_bnb_npu.yaml
View File

@@ -1,7 +1,7 @@
### model
model_name_or_path: meta-llama/Meta-Llama-3-8B-Instruct
quantization_bit: 4
quantization_method: bitsandbytes # choices: [bitsandbytes (4/8), hqq (2/3/4/5/6/8), eetq (8)]
quantization_method: bitsandbytes
double_quantization: false
trust_remote_code: true