Hugging Face's logo

nishimura999
/
llm-jp-3-13b-finetune-v100

Text Generation
Transformers
Safetensors
Japanese
llama
text-generation-inference
unsloth
trl
Inference Endpoints
Model card Files Community
nishimura999 commited on
Commit
e19cf17
·
verified ·
1 Parent(s): 448be84

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +8 -8
README.md CHANGED
@@ -23,8 +23,8 @@ This llama model was trained 2x faster with [Unsloth](https://github.com/unsloth
23
 
24
  [<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth)
25
 
26
-
27
- # import
28
  ```python
29
  from transformers import (
30
  AutoModelForCausalLM,
@@ -36,7 +36,7 @@ from tqdm import tqdm
36
  import json
37
  ```
38
 
39
- # setting
40
  ```python
41
  # Hugging Faceで取得したToken
42
  HF_TOKEN = "{Your hagging face token}"
@@ -45,7 +45,7 @@ HF_TOKEN = "{Your hagging face token}"
45
  model_name = "nishimura999/llm-jp-3-13b-finetune-v100"
46
  ```
47
 
48
- # confing
49
  ```python
50
  # QLoRA config
51
  bnb_config = BitsAndBytesConfig(
@@ -55,7 +55,7 @@ bnb_config = BitsAndBytesConfig(
55
  bnb_4bit_use_double_quant=False,
56
  )
57
  ```
58
- # load
59
  ```python
60
  # Load model
61
  model = AutoModelForCausalLM.from_pretrained(
@@ -69,7 +69,7 @@ model = AutoModelForCausalLM.from_pretrained(
69
  tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True, token = HF_TOKEN)
70
  ```
71
 
72
- # dataset
73
  ```python
74
  # データセットの読み込み。
75
  datasets = []
@@ -83,7 +83,7 @@ with open("./elyza-tasks-100-TV_0.jsonl", "r") as f:
83
  item = ""
84
  ```
85
 
86
- # generate
87
  ```python
88
  results = []
89
  for data in tqdm(datasets):
@@ -108,7 +108,7 @@ for data in tqdm(datasets):
108
  results.append({"task_id": data["task_id"], "input": input, "output": output})
109
  ```
110
 
111
- # output
112
  ```python
113
  import re
114
  model_name = re.sub(".*/", "", model_name)
 
23
 
24
  [<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth)
25
 
26
+ # usage
27
+ ## -import
28
  ```python
29
  from transformers import (
30
  AutoModelForCausalLM,
 
36
  import json
37
  ```
38
 
39
+ ## -setting
40
  ```python
41
  # Hugging Faceで取得したToken
42
  HF_TOKEN = "{Your hagging face token}"
 
45
  model_name = "nishimura999/llm-jp-3-13b-finetune-v100"
46
  ```
47
 
48
+ ## -confing
49
  ```python
50
  # QLoRA config
51
  bnb_config = BitsAndBytesConfig(
 
55
  bnb_4bit_use_double_quant=False,
56
  )
57
  ```
58
+ ## -load
59
  ```python
60
  # Load model
61
  model = AutoModelForCausalLM.from_pretrained(
 
69
  tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True, token = HF_TOKEN)
70
  ```
71
 
72
+ ## -dataset
73
  ```python
74
  # データセットの読み込み。
75
  datasets = []
 
83
  item = ""
84
  ```
85
 
86
+ ## -generate
87
  ```python
88
  results = []
89
  for data in tqdm(datasets):
 
108
  results.append({"task_id": data["task_id"], "input": input, "output": output})
109
  ```
110
 
111
+ ## -output
112
  ```python
113
  import re
114
  model_name = re.sub(".*/", "", model_name)