Files changed (1) hide show
  1. README.md +100 -89
README.md CHANGED
@@ -1,90 +1,101 @@
1
- ---
2
- license: other
3
- license_name: qwen-research
4
- license_link: https://huggingface.co/Qwen/Qwen2.5-3B/blob/main/LICENSE
5
- language:
6
- - fr
7
- - en
8
- pipeline_tag: text-generation
9
- tags:
10
- - chat
11
- - qwen
12
- - qwen2.5
13
- - finetune
14
- - french
15
- - english
16
- library_name: transformers
17
- inference: false
18
- model_creator: MaziyarPanahi
19
- quantized_by: MaziyarPanahi
20
- base_model: Qwen/Qwen2.5-3B
21
- model_name: calme-3.1-instruct-3b
22
- datasets:
23
- - MaziyarPanahi/french_instruct_sharegpt
24
- - arcee-ai/EvolKit-20k
25
- ---
26
-
27
- <img src="./calme_3.png" alt="Calme-3 Models" width="800" style="margin-left:'auto' margin-right:'auto' display:'block'"/>
28
-
29
- > [!TIP]
30
- > This is avery small model, so it might not perform well for some prompts and may be sensitive to hyper parameters. I would appreciate any feedback to see if I can fix any issues in the next iteration. ❤️
31
-
32
- # MaziyarPanahi/calme-3.1-instruct-3b
33
-
34
- This model is an advanced iteration of the powerful `Qwen/Qwen2.5-3B`, specifically fine-tuned to enhance its capabilities in generic domains.
35
-
36
-
37
- # ⚡ Quantized GGUF
38
-
39
- All GGUF models are available here: [MaziyarPanahi/calme-3.1-instruct-3b-GGUF](https://huggingface.co/MaziyarPanahi/calme-3.1-instruct-3b-GGUF)
40
-
41
-
42
- # 🏆 [Open LLM Leaderboard Evaluation Results](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)
43
-
44
- Leaderboard 2 coming soon!
45
-
46
-
47
- # Prompt Template
48
-
49
- This model uses `ChatML` prompt template:
50
-
51
- ```
52
- <|im_start|>system
53
- {System}
54
- <|im_end|>
55
- <|im_start|>user
56
- {User}
57
- <|im_end|>
58
- <|im_start|>assistant
59
- {Assistant}
60
- ````
61
-
62
- # How to use
63
-
64
-
65
- ```python
66
-
67
- # Use a pipeline as a high-level helper
68
-
69
- from transformers import pipeline
70
-
71
- messages = [
72
- {"role": "user", "content": "Who are you?"},
73
- ]
74
- pipe = pipeline("text-generation", model="MaziyarPanahi/calme-3.1-instruct-3b")
75
- pipe(messages)
76
-
77
-
78
- # Load model directly
79
-
80
- from transformers import AutoTokenizer, AutoModelForCausalLM
81
-
82
- tokenizer = AutoTokenizer.from_pretrained("MaziyarPanahi/calme-3.1-instruct-3b")
83
- model = AutoModelForCausalLM.from_pretrained("MaziyarPanahi/calme-3.1-instruct-3b")
84
- ```
85
-
86
-
87
-
88
- # Ethical Considerations
89
-
 
 
 
 
 
 
 
 
 
 
 
90
  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.
 
1
+ ---
2
+ license: other
3
+ license_name: qwen-research
4
+ license_link: https://huggingface.co/Qwen/Qwen2.5-3B/blob/main/LICENSE
5
+ language:
6
+ - zho
7
+ - eng
8
+ - fra
9
+ - spa
10
+ - por
11
+ - deu
12
+ - ita
13
+ - rus
14
+ - jpn
15
+ - kor
16
+ - vie
17
+ - tha
18
+ - ara
19
+ pipeline_tag: text-generation
20
+ tags:
21
+ - chat
22
+ - qwen
23
+ - qwen2.5
24
+ - finetune
25
+ - french
26
+ - english
27
+ library_name: transformers
28
+ inference: false
29
+ model_creator: MaziyarPanahi
30
+ quantized_by: MaziyarPanahi
31
+ base_model: Qwen/Qwen2.5-3B
32
+ model_name: calme-3.1-instruct-3b
33
+ datasets:
34
+ - MaziyarPanahi/french_instruct_sharegpt
35
+ - arcee-ai/EvolKit-20k
36
+ ---
37
+
38
+ <img src="./calme_3.png" alt="Calme-3 Models" width="800" style="margin-left:'auto' margin-right:'auto' display:'block'"/>
39
+
40
+ > [!TIP]
41
+ > This is avery small model, so it might not perform well for some prompts and may be sensitive to hyper parameters. I would appreciate any feedback to see if I can fix any issues in the next iteration. ❤️
42
+
43
+ # MaziyarPanahi/calme-3.1-instruct-3b
44
+
45
+ This model is an advanced iteration of the powerful `Qwen/Qwen2.5-3B`, specifically fine-tuned to enhance its capabilities in generic domains.
46
+
47
+
48
+ # ⚡ Quantized GGUF
49
+
50
+ All GGUF models are available here: [MaziyarPanahi/calme-3.1-instruct-3b-GGUF](https://huggingface.co/MaziyarPanahi/calme-3.1-instruct-3b-GGUF)
51
+
52
+
53
+ # 🏆 [Open LLM Leaderboard Evaluation Results](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)
54
+
55
+ Leaderboard 2 coming soon!
56
+
57
+
58
+ # Prompt Template
59
+
60
+ This model uses `ChatML` prompt template:
61
+
62
+ ```
63
+ <|im_start|>system
64
+ {System}
65
+ <|im_end|>
66
+ <|im_start|>user
67
+ {User}
68
+ <|im_end|>
69
+ <|im_start|>assistant
70
+ {Assistant}
71
+ ````
72
+
73
+ # How to use
74
+
75
+
76
+ ```python
77
+
78
+ # Use a pipeline as a high-level helper
79
+
80
+ from transformers import pipeline
81
+
82
+ messages = [
83
+ {"role": "user", "content": "Who are you?"},
84
+ ]
85
+ pipe = pipeline("text-generation", model="MaziyarPanahi/calme-3.1-instruct-3b")
86
+ pipe(messages)
87
+
88
+
89
+ # Load model directly
90
+
91
+ from transformers import AutoTokenizer, AutoModelForCausalLM
92
+
93
+ tokenizer = AutoTokenizer.from_pretrained("MaziyarPanahi/calme-3.1-instruct-3b")
94
+ model = AutoModelForCausalLM.from_pretrained("MaziyarPanahi/calme-3.1-instruct-3b")
95
+ ```
96
+
97
+
98
+
99
+ # Ethical Considerations
100
+
101
  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.