Commit
•
fcfbb80
1
Parent(s):
d7ce43f
Update app.py
Browse files
app.py
CHANGED
@@ -90,7 +90,7 @@ HF_TOKEN = os.environ.get("HF_TOKEN")
|
|
90 |
# will use api to restart space on a unrecoverable error
|
91 |
api = HfApi(token=HF_TOKEN)
|
92 |
|
93 |
-
repo_id = "coqui/voice-chat-with-
|
94 |
|
95 |
|
96 |
default_system_message = f"""
|
@@ -208,41 +208,13 @@ def format_prompt_zephyr(message, history, system_message=system_message):
|
|
208 |
print(prompt)
|
209 |
return prompt
|
210 |
|
211 |
-
def generate(
|
212 |
-
prompt, history, temperature=0.9, max_new_tokens=256, top_p=0.95, repetition_penalty=1.0,
|
213 |
-
):
|
214 |
-
temperature = float(temperature)
|
215 |
-
if temperature < 1e-2:
|
216 |
-
temperature = 1e-2
|
217 |
-
top_p = float(top_p)
|
218 |
-
|
219 |
-
generate_kwargs = dict(
|
220 |
-
temperature=temperature,
|
221 |
-
max_tokens=max_new_tokens,
|
222 |
-
max_new_tokens=max_new_tokens,
|
223 |
-
top_p=top_p,
|
224 |
-
stop=LLM_STOP_WORDS
|
225 |
-
)
|
226 |
-
formatted_prompt = format_prompt(prompt, history)
|
227 |
-
|
228 |
-
del generate_kwargs["max_tokens"]
|
229 |
-
del generate_kwargs["stop"]
|
230 |
-
|
231 |
-
stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=False)
|
232 |
-
output = ""
|
233 |
-
|
234 |
-
for response in stream:
|
235 |
-
output += response.token.text
|
236 |
-
yield output
|
237 |
-
return output
|
238 |
-
|
239 |
def generate_local(
|
240 |
prompt,
|
241 |
history,
|
242 |
llm_model="zephyr",
|
243 |
system_message=None,
|
244 |
-
temperature=0.
|
245 |
-
max_tokens=
|
246 |
top_p=0.95,
|
247 |
stop = LLM_STOP_WORDS
|
248 |
):
|
@@ -296,7 +268,7 @@ def generate_local(
|
|
296 |
return
|
297 |
|
298 |
|
299 |
-
output +=
|
300 |
yield output
|
301 |
|
302 |
except Exception as e:
|
|
|
90 |
# will use api to restart space on a unrecoverable error
|
91 |
api = HfApi(token=HF_TOKEN)
|
92 |
|
93 |
+
repo_id = "coqui/voice-chat-with-mistral"
|
94 |
|
95 |
|
96 |
default_system_message = f"""
|
|
|
208 |
print(prompt)
|
209 |
return prompt
|
210 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
211 |
def generate_local(
|
212 |
prompt,
|
213 |
history,
|
214 |
llm_model="zephyr",
|
215 |
system_message=None,
|
216 |
+
temperature=0.85,
|
217 |
+
max_tokens=128,
|
218 |
top_p=0.95,
|
219 |
stop = LLM_STOP_WORDS
|
220 |
):
|
|
|
268 |
return
|
269 |
|
270 |
|
271 |
+
output += character.replace("<|assistant|>","").replace("<|user|>","")
|
272 |
yield output
|
273 |
|
274 |
except Exception as e:
|