Debug
Browse files
app.py
CHANGED
@@ -98,41 +98,26 @@ def chat_with_model(messages):
|
|
98 |
return
|
99 |
|
100 |
current_id = patient_id.value
|
101 |
-
if current_id
|
102 |
yield messages
|
103 |
return
|
104 |
|
105 |
-
# 🛠 Missing variable initializations
|
106 |
max_new_tokens = 1024
|
107 |
output_text = ""
|
108 |
in_think = False
|
109 |
generated_tokens = 0
|
110 |
|
111 |
-
pad_id = current_tokenizer.pad_token_id
|
112 |
eos_id = current_tokenizer.eos_token_id
|
113 |
-
if pad_id is None:
|
114 |
-
pad_id = current_tokenizer.unk_token_id or 0
|
115 |
-
|
116 |
-
# Remove the initial welcome if present
|
117 |
-
filtered_messages = [msg for msg in messages if not (msg["role"] == "assistant" and "Welcome to the Radiologist's Companion" in msg["content"])]
|
118 |
|
119 |
-
# Build system context
|
120 |
system_messages = [
|
121 |
{
|
122 |
"role": "system",
|
123 |
"content": (
|
124 |
"You are a radiologist's companion, here to answer questions about the patient and assist in the diagnosis if asked to do so. "
|
125 |
"You are able to call specialized tools. "
|
126 |
-
"At the moment, you have one tool available: an organ segmentation algorithm for abdominal CTs
|
127 |
-
"If the user requests an organ segmentation, output a JSON object in this structure:\n"
|
128 |
-
"{\n"
|
129 |
-
" \"function\": \"segment_organ\",\n"
|
130 |
-
" \"arguments\": {\n"
|
131 |
-
" \"scan_path\": \"<path_to_ct_scan>\",\n"
|
132 |
-
" \"organ\": \"<organ_name>\"\n"
|
133 |
-
" }\n"
|
134 |
-
"}\n\n"
|
135 |
-
"Once you call the function, the app will execute it and return the result."
|
136 |
)
|
137 |
},
|
138 |
{
|
@@ -141,8 +126,13 @@ def chat_with_model(messages):
|
|
141 |
}
|
142 |
]
|
143 |
|
|
|
|
|
|
|
|
|
144 |
full_messages = system_messages + filtered_messages
|
145 |
|
|
|
146 |
prompt = format_prompt(full_messages)
|
147 |
|
148 |
device = torch.device("cuda")
|
@@ -169,15 +159,13 @@ def chat_with_model(messages):
|
|
169 |
thread = threading.Thread(target=current_model.generate, kwargs=generation_kwargs)
|
170 |
thread.start()
|
171 |
|
172 |
-
|
173 |
-
messages.
|
174 |
-
|
175 |
-
print(messages)
|
176 |
|
177 |
for token_info in streamer:
|
178 |
token_str = token_info["token"]
|
179 |
token_id = token_info["token_id"]
|
180 |
-
is_special = token_info["is_special"]
|
181 |
|
182 |
if token_id == eos_id:
|
183 |
break
|
@@ -196,27 +184,26 @@ def chat_with_model(messages):
|
|
196 |
|
197 |
if "\nUser" in output_text:
|
198 |
output_text = output_text.split("\nUser")[0].rstrip()
|
199 |
-
|
200 |
break
|
201 |
|
202 |
generated_tokens += 1
|
203 |
if generated_tokens >= max_new_tokens:
|
204 |
break
|
205 |
|
206 |
-
|
207 |
|
208 |
-
|
209 |
-
|
210 |
-
|
211 |
-
yield messages
|
212 |
|
213 |
if in_think:
|
214 |
output_text += "*"
|
215 |
-
|
216 |
|
217 |
torch.cuda.empty_cache()
|
218 |
-
|
219 |
-
return
|
|
|
220 |
|
221 |
|
222 |
def load_model_on_selection(model_name, progress=gr.Progress(track_tqdm=False)):
|
|
|
98 |
return
|
99 |
|
100 |
current_id = patient_id.value
|
101 |
+
if not current_id:
|
102 |
yield messages
|
103 |
return
|
104 |
|
|
|
105 |
max_new_tokens = 1024
|
106 |
output_text = ""
|
107 |
in_think = False
|
108 |
generated_tokens = 0
|
109 |
|
110 |
+
pad_id = current_tokenizer.pad_token_id or current_tokenizer.unk_token_id or 0
|
111 |
eos_id = current_tokenizer.eos_token_id
|
|
|
|
|
|
|
|
|
|
|
112 |
|
113 |
+
# --- Build system context
|
114 |
system_messages = [
|
115 |
{
|
116 |
"role": "system",
|
117 |
"content": (
|
118 |
"You are a radiologist's companion, here to answer questions about the patient and assist in the diagnosis if asked to do so. "
|
119 |
"You are able to call specialized tools. "
|
120 |
+
"At the moment, you have one tool available: an organ segmentation algorithm for abdominal CTs."
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
121 |
)
|
122 |
},
|
123 |
{
|
|
|
126 |
}
|
127 |
]
|
128 |
|
129 |
+
# Remove welcome message (only once shown)
|
130 |
+
# filtered_messages = [msg for msg in messages if not (msg["role"] == "assistant" and "Welcome to the Radiologist's Companion" in msg["content"])]
|
131 |
+
|
132 |
+
# FULL conversation
|
133 |
full_messages = system_messages + filtered_messages
|
134 |
|
135 |
+
# --- Generate from full context
|
136 |
prompt = format_prompt(full_messages)
|
137 |
|
138 |
device = torch.device("cuda")
|
|
|
159 |
thread = threading.Thread(target=current_model.generate, kwargs=generation_kwargs)
|
160 |
thread.start()
|
161 |
|
162 |
+
# Now extend previous messages
|
163 |
+
updated_messages = messages.copy()
|
164 |
+
updated_messages.append({"role": "assistant", "content": ""})
|
|
|
165 |
|
166 |
for token_info in streamer:
|
167 |
token_str = token_info["token"]
|
168 |
token_id = token_info["token_id"]
|
|
|
169 |
|
170 |
if token_id == eos_id:
|
171 |
break
|
|
|
184 |
|
185 |
if "\nUser" in output_text:
|
186 |
output_text = output_text.split("\nUser")[0].rstrip()
|
187 |
+
updated_messages[-1]["content"] = output_text
|
188 |
break
|
189 |
|
190 |
generated_tokens += 1
|
191 |
if generated_tokens >= max_new_tokens:
|
192 |
break
|
193 |
|
194 |
+
updated_messages[-1]["content"] = output_text
|
195 |
|
196 |
+
patient_conversations[current_id] = updated_messages
|
197 |
+
yield updated_messages
|
|
|
|
|
198 |
|
199 |
if in_think:
|
200 |
output_text += "*"
|
201 |
+
updated_messages[-1]["content"] = output_text
|
202 |
|
203 |
torch.cuda.empty_cache()
|
204 |
+
updated_messages[-1]["content"] = output_text
|
205 |
+
return updated_messages
|
206 |
+
|
207 |
|
208 |
|
209 |
def load_model_on_selection(model_name, progress=gr.Progress(track_tqdm=False)):
|