taoki commited on
Commit
926cdc1
ยท
verified ยท
1 Parent(s): 3305ed3

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +17 -22
README.md CHANGED
@@ -27,46 +27,41 @@ base_model: tokyotech-llm/Swallow-MS-7b-v0.1
27
  # Usage
28
 
29
  ```python
30
- from transformers import AutoTokenizer, AutoModelForCausalLM
31
  import torch
 
32
 
33
  tokenizer = AutoTokenizer.from_pretrained(
34
- "taoki/Swallow-MS-7b-v0.1-qlora-amenokaku-code"
35
  )
36
  model = AutoModelForCausalLM.from_pretrained(
37
- "taoki/Swallow-MS-7b-v0.1-qlora-amenokaku-code"
38
  )
39
 
40
  if torch.cuda.is_available():
41
- model = model.to("cuda")
42
 
43
- prompt="""### Instruction:
44
- ๅ…‰ใฎไธ‰ๅŽŸ่‰ฒใฏ๏ผŸ
45
- ### Response:
46
- """
47
 
48
  input_ids = tokenizer(prompt, return_tensors="pt").to(model.device)
49
  outputs = model.generate(
50
- **input_ids,
51
- max_new_tokens=512,
52
- do_sample=True,
53
- top_p=0.95,
54
- temperature=0.1,
55
- repetition_penalty=1.0,
56
  )
57
  print(tokenizer.decode(outputs[0]))
58
  ```
59
 
60
  # Output
61
  ````
62
- <s>### Instruction:
63
- ๅ…‰ใฎไธ‰ๅŽŸ่‰ฒใฏ๏ผŸ
64
- ### Response:
65
- ```python
66
- print('่ตค')
67
- print('็ท‘')
68
- print('้’')
69
- ```</s>
70
  ````
71
 
72
  This llama model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library.
 
27
  # Usage
28
 
29
  ```python
 
30
  import torch
31
+ from transformers import AutoTokenizer, AutoModelForCausalLM
32
 
33
  tokenizer = AutoTokenizer.from_pretrained(
34
+ "taoki/Swallow-MS-7b-v0.1-qlora-oaast1-jmulti-dolly-amenokaku"
35
  )
36
  model = AutoModelForCausalLM.from_pretrained(
37
+ "taoki/Swallow-MS-7b-v0.1-qlora-oaast1-jmulti-dolly-amenokaku"
38
  )
39
 
40
  if torch.cuda.is_available():
41
+ model = model.to("cuda")
42
 
43
+ prompt="[INST] ไปŠๆ—ฅใฏ4/1ใชใฎใงใ™ใŒใ€ๅ‘จใ‚Šใฎ็š†ใ•ใ‚“ใŒ็ชๆ‹ๅญใ‚‚ใชใ„ใ“ใจใ‚’่จ€ใฃใฆใ„ใฆๅ›ฐๆƒ‘ใ—ใฆใ„ใพใ™ใ€‚ไธ€ไฝ“ไฝ•ใŒ่ตทใ“ใฃใฆใ„ใ‚‹ใฎใงใ—ใ‚‡ใ†ใ‹๏ผŸ [/INST]\n"
 
 
 
44
 
45
  input_ids = tokenizer(prompt, return_tensors="pt").to(model.device)
46
  outputs = model.generate(
47
+ **input_ids,
48
+ max_new_tokens=512,
49
+ do_sample=True,
50
+ top_p=0.95,
51
+ temperature=0.1,
52
+ repetition_penalty=1.1,
53
  )
54
  print(tokenizer.decode(outputs[0]))
55
  ```
56
 
57
  # Output
58
  ````
59
+ <s> [INST] ไปŠๆ—ฅใฏ4/1ใชใฎใงใ™ใŒใ€ๅ‘จใ‚Šใฎ็š†ใ•ใ‚“ใŒ็ชๆ‹ๅญใ‚‚ใชใ„ใ“ใจใ‚’่จ€ใฃใฆใ„ใฆๅ›ฐๆƒ‘ใ—ใฆใ„ใพใ™ใ€‚ไธ€ไฝ“ไฝ•ใŒ่ตทใ“ใฃใฆใ„ใ‚‹ใฎใงใ—ใ‚‡ใ†ใ‹๏ผŸ [/INST]
60
+ 4ๆœˆ1ๆ—ฅใฏใ‚จใ‚คใƒ—ใƒชใƒซใƒ•ใƒผใƒซใงใ€ไบบใ€…ใฏๅ†—่ซ‡ใ‚„ใ„ใŸใšใ‚‰ใ‚’่จ€ใฃใฆๆฅฝใ—ใ‚€ๆ—ฅใจใ•ใ‚Œใฆใ„ใพใ™ใ€‚ใ“ใฎ็ฟ’ๆ…ฃใฏใ€1564ๅนดใซใƒ•ใƒฉใƒณใ‚นใฎใ‚ทใƒฃใƒซใƒซ9ไธ–ใŒ4ๆœˆ1ๆ—ฅใซ็ตๅฉšใ—ใŸใ“ใจใ‹ใ‚‰ๅง‹ใพใฃใŸใจ่จ€ใ‚ใ‚Œใฆใ„ใ‚‹ใ€‚
61
+
62
+ ใ—ใ‹ใ—ใ€ใ‚ใชใŸใŒๅ›ฐๆƒ‘ใ—ใฆใ„ใ‚‹ใฎใชใ‚‰ใ€ใใ‚ŒใฏใŠใใ‚‰ใใ€ใ‚ใชใŸใŒๅ†—่ซ‡ใ‚„ใ„ใŸใšใ‚‰ใ‚’่จ€ใฃใฆใ„ใ‚‹ไบบใŸใกใŒใ€ใ‚ใชใŸใŒใใฎใ‚ˆใ†ใชใ‚‚ใฎใ ใจๆ€ใฃใฆใ„ใชใ„ใ“ใจใ‚’่จ€ใฃใฆใ„ใ‚‹ใ‹ใ‚‰ใ ใจๆ€ใ„ใพใ™ใ€‚ใ“ใ‚Œใฏใ€ๅฝผใ‚‰ใŒใ‚ใชใŸใ‚’้ฉšใ‹ใ›ใŸใ‚Šใ€็ฌ‘ใ‚ใ›ใŸใ‚Šใ™ใ‚‹ใŸใ‚ใซๅ˜˜ใ‚’ใคใ„ใฆใ„ใ‚‹ๅฏ่ƒฝๆ€งใŒใ‚ใ‚‹ใ“ใจใ‚’ๆ„ๅ‘ณใ—ใพใ™ใ€‚
63
+
64
+ ใ‚‚ใ—่ชฐใ‹ใŒใ‚ใชใŸใ‚’ๅ›ฐๆƒ‘ใ•ใ›ใŸใ‚Šใ€ไธๅฟซใซใ•ใ›ใŸใ‚Šใ™ใ‚‹ใ‚ˆใ†ใชใ“ใจใ‚’่จ€ใฃใŸๅ ดๅˆใฏใ€ๆฐ—ใซใ›ใš็„ก่ฆ–ใ—ใฆใใ ใ•ใ„ใ€‚ใพใŸใ€่‡ชๅˆ†่‡ช่บซใ‚„ไป–ไบบใ‚’ๅ‚ทใคใ‘ใ‚‹ใ‚ˆใ†ใชๅ˜˜ใ‚’ใคใใ“ใจใฏ้ฟใ‘ใพใ—ใ‚‡ใ†ใ€‚</s>
 
 
65
  ````
66
 
67
  This llama model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library.