Adding chat format to Models ? - What wind.surf will do ?
Browse files
app.py
CHANGED
@@ -6,11 +6,13 @@ import requests
|
|
6 |
MODELS = {
|
7 |
"Llama-3.2-3B": {
|
8 |
"repo_id": "lmstudio-community/Llama-3.2-3B-Instruct-GGUF",
|
9 |
-
"filename": "*Q4_K_M.gguf"
|
|
|
10 |
},
|
11 |
-
"Llama-3.2-
|
12 |
"repo_id": "lmstudio-community/Llama-3.2-1B-Instruct-GGUF",
|
13 |
-
"filename": "*Q4_K_M.gguf"
|
|
|
14 |
}
|
15 |
}
|
16 |
|
@@ -26,7 +28,7 @@ def load_model(model_name):
|
|
26 |
verbose=True,
|
27 |
n_ctx=32768,
|
28 |
n_threads=2,
|
29 |
-
chat_format="
|
30 |
)
|
31 |
return current_model
|
32 |
|
@@ -74,7 +76,7 @@ def respond(
|
|
74 |
yield message_repl
|
75 |
|
76 |
def get_chat_title(model_name):
|
77 |
-
return f"{model_name} -
|
78 |
|
79 |
demo = gr.ChatInterface(
|
80 |
respond,
|
|
|
6 |
MODELS = {
|
7 |
"Llama-3.2-3B": {
|
8 |
"repo_id": "lmstudio-community/Llama-3.2-3B-Instruct-GGUF",
|
9 |
+
"filename": "*Q4_K_M.gguf",
|
10 |
+
"chat_format": "chatml"
|
11 |
},
|
12 |
+
"Llama-3.2-5B": {
|
13 |
"repo_id": "lmstudio-community/Llama-3.2-1B-Instruct-GGUF",
|
14 |
+
"filename": "*Q4_K_M.gguf",
|
15 |
+
"chat_format": "chatml"
|
16 |
}
|
17 |
}
|
18 |
|
|
|
28 |
verbose=True,
|
29 |
n_ctx=32768,
|
30 |
n_threads=2,
|
31 |
+
chat_format=model_info["chat_format"]
|
32 |
)
|
33 |
return current_model
|
34 |
|
|
|
76 |
yield message_repl
|
77 |
|
78 |
def get_chat_title(model_name):
|
79 |
+
return f"{model_name} < - Load different model in Additional Inputs"
|
80 |
|
81 |
demo = gr.ChatInterface(
|
82 |
respond,
|