-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathvllm_client.py
More file actions
247 lines (214 loc) · 8.23 KB
/
vllm_client.py
File metadata and controls
247 lines (214 loc) · 8.23 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
"""
Cliente para la API de vLLM integrada con ModelSync
"""
import requests
import json
from typing import List, Optional, Dict, Any
from datetime import datetime
class VLLMClient:
"""Cliente para interactuar con la API de vLLM integrada con ModelSync"""
def __init__(self, base_url: str = "http://localhost:8001"):
self.base_url = base_url.rstrip('/')
self.session = requests.Session()
def health_check(self) -> Dict[str, Any]:
"""Verificar el estado del servicio"""
response = self.session.get(f"{self.base_url}/health")
response.raise_for_status()
return response.json()
def list_models(self) -> List[Dict[str, Any]]:
"""Listar modelos disponibles"""
response = self.session.get(f"{self.base_url}/models")
response.raise_for_status()
return response.json()
def generate(
self,
prompt: str,
max_tokens: int = 100,
temperature: float = 0.7,
top_p: float = 0.9,
top_k: int = 50,
stop: Optional[List[str]] = None,
model_name: Optional[str] = None,
save_to_version_control: bool = True
) -> Dict[str, Any]:
"""Generar texto con integración ModelSync"""
payload = {
"prompt": prompt,
"max_tokens": max_tokens,
"temperature": temperature,
"top_p": top_p,
"top_k": top_k,
"stream": False,
"save_to_version_control": save_to_version_control
}
if stop:
payload["stop"] = stop
if model_name:
payload["model_name"] = model_name
response = self.session.post(f"{self.base_url}/generate", json=payload)
response.raise_for_status()
return response.json()
def generate_batch(
self,
prompts: List[str],
max_tokens: int = 100,
temperature: float = 0.7,
top_p: float = 0.9,
top_k: int = 50,
stop: Optional[List[str]] = None,
save_to_version_control: bool = True
) -> List[Dict[str, Any]]:
"""Generar texto para múltiples prompts en lote"""
requests_data = []
for prompt in prompts:
request_data = {
"prompt": prompt,
"max_tokens": max_tokens,
"temperature": temperature,
"top_p": top_p,
"top_k": top_k,
"stream": False,
"save_to_version_control": save_to_version_control
}
if stop:
request_data["stop"] = stop
requests_data.append(request_data)
response = self.session.post(f"{self.base_url}/generate/batch", json=requests_data)
response.raise_for_status()
return response.json()
def get_modelsync_status(self) -> Dict[str, Any]:
"""Obtener estado de la integración ModelSync"""
response = self.session.get(f"{self.base_url}/modelsync/status")
response.raise_for_status()
return response.json()
def init_modelsync(
self,
user_name: str = "vLLM API User",
user_email: str = "vllm@modelsync.local"
) -> Dict[str, Any]:
"""Inicializar repositorio ModelSync"""
response = self.session.post(
f"{self.base_url}/modelsync/init",
params={"user_name": user_name, "user_email": user_email}
)
response.raise_for_status()
return response.json()
def get_metrics(self) -> Dict[str, Any]:
"""Obtener métricas del servicio"""
response = self.session.get(f"{self.base_url}/metrics")
response.raise_for_status()
return response.json()
class VLLMExperimentManager:
"""Gestor de experimentos para vLLM con ModelSync"""
def __init__(self, client: VLLMClient):
self.client = client
def run_experiment(
self,
experiment_name: str,
prompts: List[str],
parameters: Dict[str, Any],
description: str = ""
) -> Dict[str, Any]:
"""Ejecutar un experimento con diferentes parámetros"""
results = []
for i, prompt in enumerate(prompts):
try:
response = self.client.generate(
prompt=prompt,
**parameters,
save_to_version_control=True
)
results.append({
"prompt_index": i,
"prompt": prompt,
"response": response["text"],
"model_version_id": response.get("model_version_id"),
"usage": response["usage"],
"created": response["created"]
})
except Exception as e:
results.append({
"prompt_index": i,
"prompt": prompt,
"error": str(e),
"created": datetime.now().isoformat()
})
# Crear resumen del experimento
experiment_summary = {
"experiment_name": experiment_name,
"description": description,
"parameters": parameters,
"total_prompts": len(prompts),
"successful_generations": len([r for r in results if "error" not in r]),
"failed_generations": len([r for r in results if "error" in r]),
"total_tokens": sum(r.get("usage", {}).get("total_tokens", 0) for r in results if "usage" in r),
"results": results,
"created": datetime.now().isoformat()
}
return experiment_summary
def compare_parameters(
self,
base_prompts: List[str],
parameter_sets: List[Dict[str, Any]],
experiment_name: str = "parameter_comparison"
) -> Dict[str, Any]:
"""Comparar diferentes conjuntos de parámetros"""
comparison_results = []
for i, params in enumerate(parameter_sets):
param_name = f"param_set_{i+1}"
experiment_result = self.run_experiment(
experiment_name=f"{experiment_name}_{param_name}",
prompts=base_prompts,
parameters=params,
description=f"Parameter set {i+1}: {params}"
)
comparison_results.append({
"parameter_set": params,
"parameter_name": param_name,
"experiment_result": experiment_result
})
return {
"comparison_name": experiment_name,
"base_prompts": base_prompts,
"parameter_sets": parameter_sets,
"results": comparison_results,
"created": datetime.now().isoformat()
}
# Ejemplo de uso
if __name__ == "__main__":
client = VLLMClient()
# Verificar salud
health = client.health_check()
print(f"Estado del servicio: {health}")
# Verificar estado de ModelSync
modelsync_status = client.get_modelsync_status()
print(f"Estado ModelSync: {modelsync_status}")
# Inicializar ModelSync si es necesario
if modelsync_status["status"] == "not_initialized":
init_result = client.init_modelsync()
print(f"Inicialización ModelSync: {init_result}")
# Generar texto
response = client.generate(
prompt="Hola, ¿cómo estás?",
max_tokens=50,
temperature=0.7
)
print(f"Respuesta: {response['text']}")
print(f"Versión del modelo: {response.get('model_version_id', 'No guardado')}")
# Ejemplo de experimento
experiment_manager = VLLMExperimentManager(client)
experiment_result = experiment_manager.run_experiment(
experiment_name="test_experiment",
prompts=[
"Cuéntame un chiste",
"Explica qué es la inteligencia artificial",
"¿Cuál es la capital de Francia?"
],
parameters={
"max_tokens": 30,
"temperature": 0.7,
"top_p": 0.9
},
description="Experimento de prueba con diferentes prompts"
)
print(f"Experimento completado: {experiment_result['successful_generations']}/{experiment_result['total_prompts']} exitosos")