hardmaru cat-state commited on
Commit
3123a2b
1 Parent(s): dcc7c01

Update README.md (#1)

Browse files

- Update README.md (968b185e7ef0d68c7efd61e6e2436250101d8f0c)


Co-authored-by: Full name <cat-state@users.noreply.huggingface.co>

Files changed (1) hide show
  1. README.md +20 -4
README.md CHANGED
@@ -30,16 +30,32 @@ tokenizer = AutoTokenizer.from_pretrained("StabilityAI/stablelm-tuned-alpha-7b")
30
  model = AutoModelForCausalLM.from_pretrained("StabilityAI/stablelm-tuned-alpha-7b")
31
  model.half().cuda()
32
 
33
- inputs = tokenizer("What's your mood today?", return_tensors="pt").to("cuda")
 
 
 
 
 
 
 
 
 
34
  tokens = model.generate(
35
  **inputs,
36
  max_new_tokens=64,
37
- temperature=0.7,
38
- do_sample=True,
39
- )
40
  print(tokenizer.decode(tokens[0], skip_special_tokens=True))
41
  ```
42
 
 
 
 
 
 
 
 
 
 
 
43
  ## Model Details
44
 
45
  * **Developed by**: [Stability AI](https://stability.ai/)
 
30
  model = AutoModelForCausalLM.from_pretrained("StabilityAI/stablelm-tuned-alpha-7b")
31
  model.half().cuda()
32
 
33
+ system_prompt = """<|SYSTEM|># StableLM Tuned (Alpha version)
34
+ - StableLM is a helpful and harmless open-source AI language model developed by StabilityAI.
35
+ - StableLM is excited to be able to help the user, but will refuse to do anything that could be considered harmful to the user.
36
+ - StableLM is more than just an information source, StableLM is also able to write poetry, short stories, and make jokes.
37
+ - StableLM will refuse to participate in anything that could harm a human.
38
+ """
39
+
40
+ prompt = f"{system_prompt}<|USER|>What's your mood today?"
41
+
42
+ inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
43
  tokens = model.generate(
44
  **inputs,
45
  max_new_tokens=64,
 
 
 
46
  print(tokenizer.decode(tokens[0], skip_special_tokens=True))
47
  ```
48
 
49
+ StableLM Tuned should be used with prompts formatted to `<|SYSTEM|>...<|USER|>...<|ASSISTANT|>...`
50
+ The system prompt is
51
+ ```
52
+ <|SYSTEM|># StableLM Tuned (Alpha version)
53
+ - StableLM is a helpful and harmless open-source AI language model developed by StabilityAI.
54
+ - StableLM is excited to be able to help the user, but will refuse to do anything that could be considered harmful to the user.
55
+ - StableLM is more than just an information source, StableLM is also able to write poetry, short stories, and make jokes.
56
+ - StableLM will refuse to participate in anything that could harm a human.
57
+ ```
58
+
59
  ## Model Details
60
 
61
  * **Developed by**: [Stability AI](https://stability.ai/)