Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -4,18 +4,24 @@ from ctransformers import AutoModelForCausalLM
|
|
| 4 |
# Cargar modelo GGUF
|
| 5 |
model = AutoModelForCausalLM.from_pretrained(
|
| 6 |
"tinyllama-1.1b-chat-v1.0.Q4_0.gguf",
|
| 7 |
-
model_type="llama"
|
|
|
|
|
|
|
| 8 |
)
|
| 9 |
|
| 10 |
def chat_fn(message, history):
|
| 11 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
output = model(
|
| 13 |
-
|
| 14 |
max_new_tokens=200,
|
| 15 |
temperature=0.7
|
| 16 |
)
|
| 17 |
|
| 18 |
-
# Convertir a string por seguridad
|
| 19 |
if isinstance(output, bytes):
|
| 20 |
output = output.decode("utf-8", errors="ignore")
|
| 21 |
|
|
@@ -29,4 +35,3 @@ demo = gr.ChatInterface(
|
|
| 29 |
|
| 30 |
demo.launch()
|
| 31 |
|
| 32 |
-
|
|
|
|
| 4 |
# Cargar modelo GGUF
|
| 5 |
model = AutoModelForCausalLM.from_pretrained(
|
| 6 |
"tinyllama-1.1b-chat-v1.0.Q4_0.gguf",
|
| 7 |
+
model_type="llama",
|
| 8 |
+
gpu_layers=0,
|
| 9 |
+
threads=4
|
| 10 |
)
|
| 11 |
|
| 12 |
def chat_fn(message, history):
|
| 13 |
+
prompt = (
|
| 14 |
+
"<|system|>Eres un asistente útil especializado en trámites y administración en España.<|end|>\n"
|
| 15 |
+
f"<|user|>{message}<|end|>\n"
|
| 16 |
+
"<|assistant|>"
|
| 17 |
+
)
|
| 18 |
+
|
| 19 |
output = model(
|
| 20 |
+
prompt,
|
| 21 |
max_new_tokens=200,
|
| 22 |
temperature=0.7
|
| 23 |
)
|
| 24 |
|
|
|
|
| 25 |
if isinstance(output, bytes):
|
| 26 |
output = output.decode("utf-8", errors="ignore")
|
| 27 |
|
|
|
|
| 35 |
|
| 36 |
demo.launch()
|
| 37 |
|
|
|