Create README.md
#2
by
MaziyarPanahi
- opened
README.md
ADDED
@@ -0,0 +1,85 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
language:
|
3 |
+
- fr
|
4 |
+
- en
|
5 |
+
pipeline_tag: text-generation
|
6 |
+
tags:
|
7 |
+
- chat
|
8 |
+
- llama
|
9 |
+
- llama3
|
10 |
+
- finetune
|
11 |
+
- french
|
12 |
+
- legal
|
13 |
+
- loi
|
14 |
+
library_name: transformers
|
15 |
+
inference: false
|
16 |
+
model_creator: MaziyarPanahi
|
17 |
+
quantized_by: MaziyarPanahi
|
18 |
+
base_model: meta-llama/Llama-3.2-3B
|
19 |
+
model_name: calme-3.1-llamaloi-3b
|
20 |
+
datasets:
|
21 |
+
- MaziyarPanahi/calme-legalkit-v0.2
|
22 |
+
license: llama3.2
|
23 |
+
---
|
24 |
+
|
25 |
+
<img src="./calme_3.png" alt="Calme-3 Models" width="800" style="margin-left:'auto' margin-right:'auto' display:'block'"/>
|
26 |
+
|
27 |
+
# MaziyarPanahi/calme-3.1-llamaloi-3b
|
28 |
+
|
29 |
+
This model is an advanced iteration of the powerful `meta-llama/Llama-3.2-3B`, specifically fine-tuned to enhance its capabilities in French Legal domain.
|
30 |
+
|
31 |
+
|
32 |
+
# ⚡ Quantized GGUF
|
33 |
+
|
34 |
+
All GGUF models are available here: [MaziyarPanahi/calme-3.1-llamaloi-3b-GGUF](https://huggingface.co/MaziyarPanahi/calme-3.1-llamaloi-3b-GGUF)
|
35 |
+
|
36 |
+
|
37 |
+
# 🏆 [Open LLM Leaderboard Evaluation Results](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)
|
38 |
+
|
39 |
+
Leaderboard 2 coming soon!
|
40 |
+
|
41 |
+
|
42 |
+
# Prompt Template
|
43 |
+
|
44 |
+
This model uses `ChatML` prompt template:
|
45 |
+
|
46 |
+
```
|
47 |
+
<|im_start|>system
|
48 |
+
{System}
|
49 |
+
<|im_end|>
|
50 |
+
<|im_start|>user
|
51 |
+
{User}
|
52 |
+
<|im_end|>
|
53 |
+
<|im_start|>assistant
|
54 |
+
{Assistant}
|
55 |
+
````
|
56 |
+
|
57 |
+
# How to use
|
58 |
+
|
59 |
+
|
60 |
+
```python
|
61 |
+
|
62 |
+
# Use a pipeline as a high-level helper
|
63 |
+
|
64 |
+
from transformers import pipeline
|
65 |
+
|
66 |
+
messages = [
|
67 |
+
{"role": "user", "content": "Who are you?"},
|
68 |
+
]
|
69 |
+
pipe = pipeline("text-generation", model="MaziyarPanahi/calme-3.1-llamaloi-3b")
|
70 |
+
pipe(messages)
|
71 |
+
|
72 |
+
|
73 |
+
# Load model directly
|
74 |
+
|
75 |
+
from transformers import AutoTokenizer, AutoModelForCausalLM
|
76 |
+
|
77 |
+
tokenizer = AutoTokenizer.from_pretrained("MaziyarPanahi/calme-3.1-llamaloi-3b")
|
78 |
+
model = AutoModelForCausalLM.from_pretrained("MaziyarPanahi/calme-3.1-llamaloi-3b")
|
79 |
+
```
|
80 |
+
|
81 |
+
|
82 |
+
|
83 |
+
# Ethical Considerations
|
84 |
+
|
85 |
+
As with any large language model, users should be aware of potential biases and limitations. We recommend implementing appropriate safeguards and human oversight when deploying this model in production environments.
|