Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -108,10 +108,10 @@ def run_gpt(
|
|
108 |
safe_search=safe_search,
|
109 |
) + prompt_template.format(**prompt_kwargs)
|
110 |
if VERBOSE:
|
111 |
-
logging.info(LOG_PROMPT.format(content)) # Log the prompt
|
112 |
resp = client.text_generation(content, max_new_tokens=max_new_tokens, stop_sequences=stop_tokens, temperature=0.7, top_p=0.8, repetition_penalty=1.5)
|
113 |
if VERBOSE:
|
114 |
-
logging.info(LOG_RESPONSE.format(
|
115 |
return resp
|
116 |
|
117 |
def generate(
|
@@ -121,15 +121,14 @@ def generate(
|
|
121 |
logging.info(f"Seed: {seed}") # Log the seed
|
122 |
|
123 |
# Set the agent prompt based on agent_name
|
|
|
124 |
if agent_name == "WEB_DEV":
|
125 |
-
agent
|
126 |
elif agent_name == "AI_SYSTEM_PROMPT":
|
127 |
-
agent
|
128 |
elif agent_name == "PYTHON_CODE_DEV":
|
129 |
-
agent
|
130 |
-
|
131 |
-
agent = "You are a helpful AI assistant."
|
132 |
-
|
133 |
system_prompt = f"{agent} {sys_prompt}".strip()
|
134 |
|
135 |
temperature = max(float(temperature), 1e-2)
|
@@ -142,9 +141,10 @@ def generate(
|
|
142 |
formatted_prompt = format_prompt(formatted_prompt, history, max_history_turns=5) # Truncated history
|
143 |
logging.info(f"Formatted Prompt: {formatted_prompt}")
|
144 |
|
145 |
-
|
|
|
146 |
|
147 |
-
stream =
|
148 |
formatted_prompt,
|
149 |
temperature=temperature,
|
150 |
max_new_tokens=max_new_tokens,
|
@@ -329,27 +329,31 @@ def format_prompt(message, history, max_history_turns=5):
|
|
329 |
prompt += f" {bot_response}</s> "
|
330 |
prompt += f"[INST] {message} [/INST]"
|
331 |
return prompt
|
|
|
332 |
agents =[
|
333 |
"WEB_DEV",
|
334 |
"AI_SYSTEM_PROMPT",
|
335 |
"PYTHON_CODE_DEV"
|
336 |
]
|
|
|
337 |
def generate(
|
338 |
prompt, history, agent_name=agents[0], sys_prompt="", temperature=0.9, max_new_tokens=2048, top_p=0.95, repetition_penalty=1.0, model="mistralai/Mixtral-8x7B-Instruct-v0.1"
|
339 |
):
|
340 |
seed = random.randint(1,1111111111111111)
|
|
|
341 |
|
342 |
-
#
|
|
|
343 |
if agent_name == "WEB_DEV":
|
344 |
-
agent
|
345 |
-
|
346 |
-
agent
|
347 |
-
|
348 |
-
agent
|
349 |
-
|
350 |
-
|
351 |
-
|
352 |
-
|
353 |
top_p = float(top_p)
|
354 |
|
355 |
# Add the system prompt to the beginning of the prompt
|
@@ -358,14 +362,28 @@ def generate(
|
|
358 |
# Use 'prompt' here instead of 'message'
|
359 |
formatted_prompt = format_prompt(formatted_prompt, history, max_history_turns=5) # Truncated history
|
360 |
logging.info(f"Formatted Prompt: {formatted_prompt}")
|
361 |
-
|
362 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
363 |
for response in stream:
|
364 |
resp += response.token.text
|
365 |
yield resp # This allows for streaming the response
|
366 |
|
367 |
if VERBOSE:
|
368 |
-
logging.info(
|
|
|
369 |
|
370 |
def generate_text_chunked(input_text, model, generation_parameters, max_tokens_to_generate):
|
371 |
"""Generates text in chunks to avoid token limit errors."""
|
@@ -387,17 +405,7 @@ def generate_text_chunked(input_text, model, generation_parameters, max_tokens_t
|
|
387 |
|
388 |
return ''.join(generated_text)
|
389 |
|
390 |
-
|
391 |
-
logging.info(f"Formatted Prompt: {formatted_prompt}")
|
392 |
-
stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=False)
|
393 |
-
output = ""
|
394 |
-
|
395 |
-
for response in stream:
|
396 |
-
output += response.token.text
|
397 |
-
yield output
|
398 |
-
return output
|
399 |
-
|
400 |
-
|
401 |
additional_inputs=[
|
402 |
gr.Dropdown(
|
403 |
label="Agents",
|
|
|
108 |
safe_search=safe_search,
|
109 |
) + prompt_template.format(**prompt_kwargs)
|
110 |
if VERBOSE:
|
111 |
+
logging.info(LOG_PROMPT.format(content=content)) # Log the prompt
|
112 |
resp = client.text_generation(content, max_new_tokens=max_new_tokens, stop_sequences=stop_tokens, temperature=0.7, top_p=0.8, repetition_penalty=1.5)
|
113 |
if VERBOSE:
|
114 |
+
logging.info(LOG_RESPONSE.format(resp=resp)) # Log the response
|
115 |
return resp
|
116 |
|
117 |
def generate(
|
|
|
121 |
logging.info(f"Seed: {seed}") # Log the seed
|
122 |
|
123 |
# Set the agent prompt based on agent_name
|
124 |
+
agent = "You are a helpful AI assistant."
|
125 |
if agent_name == "WEB_DEV":
|
126 |
+
agent += " You are a web developer."
|
127 |
elif agent_name == "AI_SYSTEM_PROMPT":
|
128 |
+
agent += " You are an AI system."
|
129 |
elif agent_name == "PYTHON_CODE_DEV":
|
130 |
+
agent += " You are a Python code developer."
|
131 |
+
|
|
|
|
|
132 |
system_prompt = f"{agent} {sys_prompt}".strip()
|
133 |
|
134 |
temperature = max(float(temperature), 1e-2)
|
|
|
141 |
formatted_prompt = format_prompt(formatted_prompt, history, max_history_turns=5) # Truncated history
|
142 |
logging.info(f"Formatted Prompt: {formatted_prompt}")
|
143 |
|
144 |
+
# Conditionally create client
|
145 |
+
this_client = InferenceClient(model) if model != "mistralai/Mixtral-8x7B-Instruct-v0.1" else client
|
146 |
|
147 |
+
stream = this_client.text_generation(
|
148 |
formatted_prompt,
|
149 |
temperature=temperature,
|
150 |
max_new_tokens=max_new_tokens,
|
|
|
329 |
prompt += f" {bot_response}</s> "
|
330 |
prompt += f"[INST] {message} [/INST]"
|
331 |
return prompt
|
332 |
+
|
333 |
agents =[
|
334 |
"WEB_DEV",
|
335 |
"AI_SYSTEM_PROMPT",
|
336 |
"PYTHON_CODE_DEV"
|
337 |
]
|
338 |
+
|
339 |
def generate(
|
340 |
prompt, history, agent_name=agents[0], sys_prompt="", temperature=0.9, max_new_tokens=2048, top_p=0.95, repetition_penalty=1.0, model="mistralai/Mixtral-8x7B-Instruct-v0.1"
|
341 |
):
|
342 |
seed = random.randint(1,1111111111111111)
|
343 |
+
logging.info(f"Seed: {seed}") # Log the seed
|
344 |
|
345 |
+
# Set the agent prompt based on agent_name
|
346 |
+
agent = "You are a helpful AI assistant."
|
347 |
if agent_name == "WEB_DEV":
|
348 |
+
agent += " You are a web developer."
|
349 |
+
elif agent_name == "AI_SYSTEM_PROMPT":
|
350 |
+
agent += " You are an AI system."
|
351 |
+
elif agent_name == "PYTHON_CODE_DEV":
|
352 |
+
agent += " You are a Python code developer."
|
353 |
+
|
354 |
+
system_prompt = f"{agent} {sys_prompt}".strip()
|
355 |
+
|
356 |
+
temperature = max(float(temperature), 1e-2)
|
357 |
top_p = float(top_p)
|
358 |
|
359 |
# Add the system prompt to the beginning of the prompt
|
|
|
362 |
# Use 'prompt' here instead of 'message'
|
363 |
formatted_prompt = format_prompt(formatted_prompt, history, max_history_turns=5) # Truncated history
|
364 |
logging.info(f"Formatted Prompt: {formatted_prompt}")
|
365 |
+
|
366 |
+
# Conditionally create client
|
367 |
+
this_client = InferenceClient(model) if model != "mistralai/Mixtral-8x7B-Instruct-v0.1" else client
|
368 |
+
|
369 |
+
stream = this_client.text_generation(
|
370 |
+
formatted_prompt,
|
371 |
+
temperature=temperature,
|
372 |
+
max_new_tokens=max_new_tokens,
|
373 |
+
top_p=top_p,
|
374 |
+
repetition_penalty=repetition_penalty,
|
375 |
+
stream=True,
|
376 |
+
details=True,
|
377 |
+
return_full_text=False
|
378 |
+
)
|
379 |
+
resp = ""
|
380 |
for response in stream:
|
381 |
resp += response.token.text
|
382 |
yield resp # This allows for streaming the response
|
383 |
|
384 |
if VERBOSE:
|
385 |
+
logging.info(f"RESPONSE: {resp}") # Log the response directly
|
386 |
+
|
387 |
|
388 |
def generate_text_chunked(input_text, model, generation_parameters, max_tokens_to_generate):
|
389 |
"""Generates text in chunks to avoid token limit errors."""
|
|
|
405 |
|
406 |
return ''.join(generated_text)
|
407 |
|
408 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
409 |
additional_inputs=[
|
410 |
gr.Dropdown(
|
411 |
label="Agents",
|