JMAA00 commited on
Commit
44a3282
·
1 Parent(s): b5d30b6

ModificacionesV1

Browse files
Files changed (2) hide show
  1. app.py +81 -0
  2. requirements.txt +2 -1
app.py CHANGED
@@ -1,3 +1,4 @@
 
1
  import gradio as gr
2
  from huggingface_hub import InferenceClient
3
 
@@ -62,3 +63,83 @@ demo = gr.ChatInterface(
62
 
63
  if __name__ == "__main__":
64
  demo.launch()
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """
2
  import gradio as gr
3
  from huggingface_hub import InferenceClient
4
 
 
63
 
64
  if __name__ == "__main__":
65
  demo.launch()
66
+ """
67
+ import os
68
+ import gradio as gr
69
+ from huggingface_hub import InferenceClient
70
+
71
+ """
72
+ For more information on `huggingface_hub` Inference API support, please check the docs:
73
+ https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
74
+ """
75
+
76
+ # Actualizamos la referencia al modelo para usar meta-llama/Llama-3.1-8B-Instruct
77
+ # Si el modelo requiere acceso mediante token, define HF_API_TOKEN como variable de entorno
78
+ # en tu Space o en tu entorno local (huggingface-cli login).
79
+ hf_api_token = os.getenv("HF_API_TOKEN") # opcional si lo necesitas
80
+ client = InferenceClient(
81
+ model="meta-llama/Llama-3.1-8B-Instruct",
82
+ token=hf_api_token
83
+ )
84
+
85
+ def respond(
86
+ message,
87
+ history: list[tuple[str, str]],
88
+ system_message,
89
+ max_tokens,
90
+ temperature,
91
+ top_p,
92
+ ):
93
+ """
94
+ Reconstruimos la conversación como una lista de mensajes:
95
+ - El primer mensaje es de tipo "system" con el contenido de 'system_message'
96
+ - 'history' contiene pares (usuario, asistente)
97
+ - 'message' es lo que el usuario introduce en este turno
98
+ """
99
+ messages = [{"role": "system", "content": system_message}]
100
+
101
+ for val in history:
102
+ if val[0]:
103
+ messages.append({"role": "user", "content": val[0]})
104
+ if val[1]:
105
+ messages.append({"role": "assistant", "content": val[1]})
106
+
107
+ messages.append({"role": "user", "content": message})
108
+
109
+ response = ""
110
+
111
+ # Llamamos a chat_completion con streaming: vamos recibiendo tokens y construyendo la respuesta
112
+ for msg_chunk in client.chat_completion(
113
+ messages,
114
+ max_tokens=max_tokens,
115
+ stream=True,
116
+ temperature=temperature,
117
+ top_p=top_p,
118
+ ):
119
+ token = msg_chunk.choices[0].delta.get("content", "")
120
+ response += token
121
+ yield response
122
+
123
+
124
+ """
125
+ For information on how to customize the ChatInterface, peruse the gradio docs:
126
+ https://www.gradio.app/docs/chatinterface
127
+ """
128
+ demo = gr.ChatInterface(
129
+ respond,
130
+ additional_inputs=[
131
+ gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
132
+ gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
133
+ gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
134
+ gr.Slider(
135
+ minimum=0.1,
136
+ maximum=1.0,
137
+ value=0.95,
138
+ step=0.05,
139
+ label="Top-p (nucleus sampling)",
140
+ ),
141
+ ],
142
+ )
143
+
144
+ if __name__ == "__main__":
145
+ demo.launch()
requirements.txt CHANGED
@@ -1 +1,2 @@
1
- huggingface_hub==0.25.2
 
 
1
+ gradio==5.0.1
2
+ huggingface_hub==0.22.2