XavierSpycy
commited on
Commit
•
916093b
1
Parent(s):
e02e8a8
Update model card
Browse files
README.md
CHANGED
|
@@ -2,7 +2,7 @@
|
|
| 2 |
license: apache-2.0
|
| 3 |
---
|
| 4 |
|
| 5 |
-
# Meta-Llama-3-8B-Instruct-zh-10k
|
| 6 |
|
| 7 |
## Model Details / 模型细节
|
| 8 |
This model, <u>`Meta-Llama-3-8B-Instruct-zh-10k`</u>, was fine-tuned from the original [Meta-Llama-3-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct) due to its underperformance in Chinese. Utilizing the LoRa technology within the [LLaMA-Factory](https://github.com/hiyouga/LLaMA-Factory) utilities, this model was adapted to better handle Chinese through three epochs on three corpora: `alpaca_zh`, `alpaca_gpt4_zh`, and `oaast_sft_zh`, amounting to approximately 10,000 examples. This is reflected in the `10k` in its name.
|
|
@@ -23,12 +23,22 @@ Additional fine-tuning configurations are avaiable at [Hands-On LoRa](https://gi
|
|
| 23 |
|
| 24 |
更多微调配置可以在我的个人仓库 [Hands-On LoRa](https://github.com/XavierSpycy/hands-on-lora) 或 [Llama3Ops](https://github.com/XavierSpycy/llama-ops) 获得。
|
| 25 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 26 |
### Model Developer / 模型开发者
|
| 27 |
- **Pretraining**: Meta
|
| 28 |
-
- **Fine-tuning**: [XavierSpycy @
|
| 29 |
|
| 30 |
- **预训练**: Meta
|
| 31 |
-
- **微调**: [XavierSpycy @
|
| 32 |
|
| 33 |
|
| 34 |
### Usage / 用法
|
|
@@ -37,6 +47,9 @@ This model can be utilized like the original <u>Meta-Llama3</u> but offers enhan
|
|
| 37 |
我们能够像原版的<u>Meta-Llama3</u>一样使用该模型，而它提供了提升后的中文能力。
|
| 38 |
|
| 39 |
```python
|
|
|
|
|
|
|
|
|
|
| 40 |
from transformers import AutoTokenizer, AutoModelForCausalLM
|
| 41 |
|
| 42 |
model_id = "XavierSpycy/Meta-Llama-3-8B-Instruct-zh-10k"
|
|
@@ -66,7 +79,8 @@ outputs = model.generate(
|
|
| 66 |
response = outputs[0][input_ids.shape[-1]:]
|
| 67 |
|
| 68 |
print(tokenizer.decode(response, skip_special_tokens=True))
|
| 69 |
-
#
|
|
|
|
| 70 |
```
|
| 71 |
|
| 72 |
Further details about the deployment are available in the GitHub repository [Llama3Ops: From LoRa to Deployment with Llama3](https://github.com/XavierSpycy/llama-ops).
|
|
|
|
| 2 |
license: apache-2.0
|
| 3 |
---
|
| 4 |
|
| 5 |
+
# Meta-Llama-3-8B-Instruct-zh-10k: A Llama🦙 which speaks Chinese / 一只说中文的羊驼🦙
|
| 6 |
|
| 7 |
## Model Details / 模型细节
|
| 8 |
This model, <u>`Meta-Llama-3-8B-Instruct-zh-10k`</u>, was fine-tuned from the original [Meta-Llama-3-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct) due to its underperformance in Chinese. Utilizing the LoRa technology within the [LLaMA-Factory](https://github.com/hiyouga/LLaMA-Factory) utilities, this model was adapted to better handle Chinese through three epochs on three corpora: `alpaca_zh`, `alpaca_gpt4_zh`, and `oaast_sft_zh`, amounting to approximately 10,000 examples. This is reflected in the `10k` in its name.
|
|
|
|
| 23 |
|
| 24 |
更多微调配置可以在我的个人仓库 [Hands-On LoRa](https://github.com/XavierSpycy/hands-on-lora) 或 [Llama3Ops](https://github.com/XavierSpycy/llama-ops) 获得。
|
| 25 |
|
| 26 |
+
### Other Models / 其他模型
|
| 27 |
+
- <u>llama.cpp</u>
|
| 28 |
+
- [Meta-Llama-3-8B-Instruct-zh-10k-GGUF](https://huggingface.co/XavierSpycy/Meta-Llama-3-8B-Instruct-zh-10k-GGUF)
|
| 29 |
+
|
| 30 |
+
- <u>AutoAWQ</u>
|
| 31 |
+
- [Meta-Llama-3-8B-Instruct-zh-10k-AWQ](https://huggingface.co/XavierSpycy/Meta-Llama-3-8B-Instruct-zh-10k-AWQ)
|
| 32 |
+
|
| 33 |
+
- <u>AutoGPTQ</u>
|
| 34 |
+
- [Meta-Llama-3-8B-Instruct-zh-10k-GPTQ](https://huggingface.co/XavierSpycy/Meta-Llama-3-8B-Instruct-zh-10k-GPTQ)
|
| 35 |
+
|
| 36 |
### Model Developer / 模型开发者
|
| 37 |
- **Pretraining**: Meta
|
| 38 |
+
- **Fine-tuning**: [XavierSpycy @ GitHub ](https://github.com/XavierSpycy) | [XavierSpycy @ 🤗](https://huggingface.co/XavierSpycy)
|
| 39 |
|
| 40 |
- **预训练**: Meta
|
| 41 |
+
- **微调**: [XavierSpycy @ GitHub](https://github.com/XavierSpycy) | [XavierSpycy @ 🤗 ](https://huggingface.co/XavierSpycy)
|
| 42 |
|
| 43 |
|
| 44 |
### Usage / 用法
|
|
|
|
| 47 |
我们能够像原版的<u>Meta-Llama3</u>一样使用该模型，而它提供了提升后的中文能力。
|
| 48 |
|
| 49 |
```python
|
| 50 |
+
# !pip install accelerate
|
| 51 |
+
|
| 52 |
+
import torch
|
| 53 |
from transformers import AutoTokenizer, AutoModelForCausalLM
|
| 54 |
|
| 55 |
model_id = "XavierSpycy/Meta-Llama-3-8B-Instruct-zh-10k"
|
|
|
|
| 79 |
response = outputs[0][input_ids.shape[-1]:]
|
| 80 |
|
| 81 |
print(tokenizer.decode(response, skip_special_tokens=True))
|
| 82 |
+
# 我是一个人工智能助手，旨在帮助用户解决问题和完成任务。
|
| 83 |
+
# 我是一个虚拟的人工智能助手，能够通过自然语言处理技术理解用户的需求并为用户提供帮助。
|
| 84 |
```
|
| 85 |
|
| 86 |
Further details about the deployment are available in the GitHub repository [Llama3Ops: From LoRa to Deployment with Llama3](https://github.com/XavierSpycy/llama-ops).
|
meta-llama-3-8b-instruct-zh-10k.Q2_K.gguf
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:5634aeaeff78e940e10671bec07e99d1acb15c8f230388df3972e44fa3f7015e
|
| 3 |
-
size 3179131456
|
|
|
|
|
|
|
|
|
|
|
|
meta-llama-3-8b-instruct-zh-10k.Q8_0.gguf
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:e675330f4b8e5916823427938b23a18f810183c7f1eb5188430efa104554c04e
|
| 3 |
-
size 8540770880
|
|
|
|
|
|
|
|
|
|
|
|
meta-llama-3-8b-instruct-zh-10k.gguf
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:d317b766a878d9c2ed765a412dc043f543f49a9795bd47ab47e7f18fa53b64ac
|
| 3 |
-
size 16068891200
|
|
|
|
|
|
|
|
|
|
|
|