joheras commited on
Commit
2f43cc5
1 Parent(s): d1f03d4

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +43 -3
README.md CHANGED
@@ -6,6 +6,8 @@ tags:
6
  model-index:
7
  - name: bertin-gpt-clara-med
8
  results: []
 
 
9
  ---
10
 
11
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -17,9 +19,47 @@ This model is a fine-tuned version of [bertin-project/bertin-gpt-j-6B-alpaca](ht
17
  It achieves the following results on the evaluation set:
18
  - Loss: 0.6110
19
 
20
- ## Model description
21
 
22
- More information needed
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
23
 
24
  ## Intended uses & limitations
25
 
@@ -62,4 +102,4 @@ The following hyperparameters were used during training:
62
  - Transformers 4.32.1
63
  - Pytorch 2.0.0+cu117
64
  - Datasets 2.14.4
65
- - Tokenizers 0.13.3
 
6
  model-index:
7
  - name: bertin-gpt-clara-med
8
  results: []
9
+ datasets:
10
+ - CLARA-MeD/CLARA-MeD
11
  ---
12
 
13
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 
19
  It achieves the following results on the evaluation set:
20
  - Loss: 0.6110
21
 
22
+ ## Usage
23
 
24
+ ```python
25
+ from transformers import AutoModelForCausalLM, AutoTokenizer, GenerationConfig, pipeline
26
+
27
+ base_model = "CLARA-MeD/bertin-gpt"
28
+ tokenizer = AutoTokenizer.from_pretrained(base_model)
29
+ model = AutoModelForCausalLM.from_pretrained(base_model).cuda()
30
+ ```
31
+
32
+ For generation, we can use the model's `.generate()` method. Remember that the prompt needs a **Spanish** template:
33
+
34
+ ```python
35
+ # Generate responses
36
+ def generate(input):
37
+ prompt = f"""A continuación hay una instrucción que describe una tarea, junto con una entrada que proporciona más contexto. Escribe una respuesta que complete adecuadamente lo que se pide.
38
+
39
+ ### Instrucción:
40
+ Simplifica la siguiente frase
41
+
42
+ ### Entrada:
43
+ {input}
44
+
45
+ ### Respuesta:"""
46
+
47
+ inputs = tokenizer(prompt, return_tensors="pt")
48
+ input_ids = inputs["input_ids"].cuda()
49
+ generation_output = model.generate(
50
+ input_ids=input_ids,
51
+ generation_config=GenerationConfig(temperature=0.2, top_p=0.75, num_beams=4),
52
+ return_dict_in_generate=True,
53
+ output_scores=True,
54
+ max_new_tokens=256
55
+ )
56
+ for seq in generation_output.sequences:
57
+ output = tokenizer.decode(seq, skip_special_tokens=True)
58
+ print(output.split("### Respuesta:")[-1].strip())
59
+
60
+ generate("Al sujeto se le ha tratado previamente con antagonistas del factor de necrosis tumoral alfa (TNF-α) sin respuesta clínica documentada al tratamiento. También puede ocurrir que al sujeto no se le tratara anteriormente con antagonistas de TNF-α, pero está fallando el tratamiento convencional actual.")
61
+
62
+ ```
63
 
64
  ## Intended uses & limitations
65
 
 
102
  - Transformers 4.32.1
103
  - Pytorch 2.0.0+cu117
104
  - Datasets 2.14.4
105
+ - Tokenizers 0.13.3