Hjgugugjhuhjggg commited on
Commit
017c336
verified
1 Parent(s): e739fbf

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +13 -4
app.py CHANGED
@@ -1,22 +1,30 @@
1
  from flask import Flask, request, jsonify, render_template_string
2
  from vllm import LLM, SamplingParams
3
  from langchain_community.cache import GPTCache
 
4
 
5
  app = Flask(__name__)
6
 
 
 
 
 
7
  modelos = {
8
- "facebook/opt-125m": LLM(model="facebook/opt-125m"),
9
- "llama-3.2-1B": LLM(model="Hjgugugjhuhjggg/llama-3.2-1B-spinquant-hf"),
10
- "gpt2": LLM(model="gpt2")
11
  }
12
 
 
13
  caches = {
14
- nombre: GPTCache(modelo, max_size=1000)
15
  for nombre, modelo in modelos.items()
16
  }
17
 
 
18
  sampling_params = SamplingParams(temperature=0.8, top_p=0.95)
19
 
 
20
  html_code_docs = """
21
  <!DOCTYPE html>
22
  <html>
@@ -63,6 +71,7 @@ html_code_docs = """
63
  </html>
64
  """
65
 
 
66
  html_code_chatbot = """
67
  <!DOCTYPE html>
68
  <html>
 
1
  from flask import Flask, request, jsonify, render_template_string
2
  from vllm import LLM, SamplingParams
3
  from langchain_community.cache import GPTCache
4
+ import torch
5
 
6
  app = Flask(__name__)
7
 
8
+ # Verificar si hay una GPU disponible, si no usar la CPU
9
+ device = "cuda" if torch.cuda.is_available() else "cpu"
10
+
11
+ # Inicializar los modelos con el dispositivo adecuado (GPU o CPU)
12
  modelos = {
13
+ "facebook/opt-125m": LLM(model="facebook/opt-125m", device=device),
14
+ "llama-3.2-1B": LLM(model="Hjgugugjhuhjggg/llama-3.2-1B-spinquant-hf", device=device),
15
+ "gpt2": LLM(model="gpt2", device=device)
16
  }
17
 
18
+ # Configuraci贸n de cach茅 para los modelos
19
  caches = {
20
+ nombre: GPTCache(modelo, max_size=1000)
21
  for nombre, modelo in modelos.items()
22
  }
23
 
24
+ # Par谩metros de muestreo para la generaci贸n de texto
25
  sampling_params = SamplingParams(temperature=0.8, top_p=0.95)
26
 
27
+ # C贸digo HTML para la documentaci贸n de la API
28
  html_code_docs = """
29
  <!DOCTYPE html>
30
  <html>
 
71
  </html>
72
  """
73
 
74
+ # C贸digo HTML para la interfaz del chatbot
75
  html_code_chatbot = """
76
  <!DOCTYPE html>
77
  <html>