Update README.md
Browse files
README.md
CHANGED
@@ -15,6 +15,30 @@ DPOには[Low-Rank Adaptation (LoRA)](https://huggingface.co/docs/peft/conceptua
|
|
15 |
## Requirements, Usage, Chat Template
|
16 |
|
17 |
[cyberagent/calm2-7b-chat](https://huggingface.co/cyberagent/calm2-7b-chat)と同様です。
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
18 |
|
19 |
## 実験結果
|
20 |
|
@@ -46,6 +70,9 @@ DPOには[Low-Rank Adaptation (LoRA)](https://huggingface.co/docs/peft/conceptua
|
|
46 |
| stem | 6.3 | 6.2 |
|
47 |
| writing | 7.7 | 9.1 |
|
48 |
|
|
|
|
|
|
|
49 |
|
50 |
## Author
|
51 |
|
|
|
15 |
## Requirements, Usage, Chat Template
|
16 |
|
17 |
[cyberagent/calm2-7b-chat](https://huggingface.co/cyberagent/calm2-7b-chat)と同様です。
|
18 |
+
同様のコード・プロンプトで動かすことができます。
|
19 |
+
|
20 |
+
```python
|
21 |
+
import transformers
|
22 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer, TextStreamer
|
23 |
+
|
24 |
+
assert transformers.__version__ >= "4.34.1"
|
25 |
+
|
26 |
+
model = AutoModelForCausalLM.from_pretrained("cyberagent/calm2-7b-chat-dpo-experimental", device_map="auto", torch_dtype="auto")
|
27 |
+
tokenizer = AutoTokenizer.from_pretrained("cyberagent/calm2-7b-chat-dpo-experimental")
|
28 |
+
streamer = TextStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)
|
29 |
+
|
30 |
+
prompt = """USER: AIによって私達の暮らしはどのように変わりますか?
|
31 |
+
ASSISTANT: """
|
32 |
+
|
33 |
+
token_ids = tokenizer.encode(prompt, return_tensors="pt")
|
34 |
+
output_ids = model.generate(
|
35 |
+
input_ids=token_ids.to(model.device),
|
36 |
+
max_new_tokens=300,
|
37 |
+
do_sample=True,
|
38 |
+
temperature=0.8,
|
39 |
+
streamer=streamer,
|
40 |
+
)
|
41 |
+
```
|
42 |
|
43 |
## 実験結果
|
44 |
|
|
|
70 |
| stem | 6.3 | 6.2 |
|
71 |
| writing | 7.7 | 9.1 |
|
72 |
|
73 |
+
## Releases
|
74 |
+
|
75 |
+
1.0: v1 release (Jan 24, 2024)
|
76 |
|
77 |
## Author
|
78 |
|