macadeliccc commited on
Commit
1ff8d2c
1 Parent(s): b171572

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +63 -1
README.md CHANGED
@@ -13,4 +13,66 @@ The model is a merge of models that are capable of Chinese and Japanese output.
13
  + oshizo/japanese-e5-mistral-7b_slerp
14
  + cognitivecomputations/dolphin-2.6-mistral-7b-dpo-laser
15
  + s3nh/Mistral-7B-Evol-Instruct-Chinese
16
-
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
13
  + oshizo/japanese-e5-mistral-7b_slerp
14
  + cognitivecomputations/dolphin-2.6-mistral-7b-dpo-laser
15
  + s3nh/Mistral-7B-Evol-Instruct-Chinese
16
+
17
+
18
+ # Code Example
19
+
20
+ ```python
21
+ # Import necessary libraries
22
+ from transformers import AutoTokenizer, AutoModelForCausalLM
23
+
24
+ # Load tokenizer and model
25
+ tokenizer = AutoTokenizer.from_pretrained("macadeliccc/laser-dolphin-mixtral-2x7b-dpo")
26
+ model = AutoModelForCausalLM.from_pretrained("macadeliccc/laser-dolphin-mixtral-2x7b-dpo")
27
+
28
+ def generate_response(prompt, max_length=50, num_return_sequences=1, temperature=1.0, top_k=50, top_p=1.0):
29
+ """
30
+ Generate a response from the model based on the input prompt and hyperparameters.
31
+
32
+ Args:
33
+ prompt (str): Prompt for the model.
34
+ max_length (int): Maximum length of the model's response.
35
+ num_return_sequences (int): Number of response sequences to generate.
36
+ temperature (float): Sampling temperature for model generation.
37
+ top_k (int): The number of highest probability vocabulary tokens to keep for top-k filtering.
38
+ top_p (float): If set to float < 1, only the most probable tokens with probabilities that add up to top_p or higher are kept for generation.
39
+
40
+ Returns:
41
+ str: The generated response from the model.
42
+ """
43
+ messages = [
44
+ {"role": "system", "content": "You are Dolphin, an AI assistant."},
45
+ {"role": "user", "content": prompt}
46
+ ]
47
+
48
+ # Apply chat template to input messages
49
+ gen_input = tokenizer.apply_chat_template(messages, return_tensors="pt")
50
+
51
+ # Generate a response
52
+ output = model.generate(**gen_input,
53
+ max_length=max_length,
54
+ num_return_sequences=num_return_sequences,
55
+ temperature=temperature,
56
+ top_k=top_k,
57
+ top_p=top_p)
58
+
59
+ # Decode the generated tokens to a string
60
+ response = tokenizer.decode(output[0], skip_special_tokens=True)
61
+
62
+ return response
63
+
64
+ # Example prompts in different languages
65
+ english_prompt = "Write a quicksort algorithm in python"
66
+ chinese_prompt = "用Python写一个快速排序算法"
67
+ japanese_prompt = "Pythonでクイックソートアルゴリズムを書いてください"
68
+
69
+ # Generate and print responses for each language
70
+ print("English Response:")
71
+ print(generate_response(english_prompt, max_length=100, temperature=0.8), "\n")
72
+
73
+ print("Chinese Response:")
74
+ print(generate_response(chinese_prompt, max_length=100, temperature=0.8), "\n")
75
+
76
+ print("Japanese Response:")
77
+ print(generate_response(japanese_prompt, max_length=100, temperature=0.8), "\n")
78
+ ```