Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
|
@@ -358,41 +358,47 @@ async def generate_text(
|
|
| 358 |
if kw in prompt_text:
|
| 359 |
score += 1
|
| 360 |
|
| 361 |
-
chosen_model = "
|
| 362 |
provider = "groq"
|
| 363 |
-
|
| 364 |
-
score = 10
|
| 365 |
-
if uses_tools:
|
| 366 |
-
if score >= 4:
|
| 367 |
-
chosen_model = "openai/gpt-oss-120b"
|
| 368 |
-
else:
|
| 369 |
-
chosen_model = "openai/gpt-oss-20b"
|
| 370 |
-
provider = "groq"
|
| 371 |
-
|
| 372 |
-
elif code_present:
|
| 373 |
|
| 374 |
-
|
| 375 |
-
chosen_model = "gpt-oss-120b"
|
| 376 |
-
provider = "cerebras"
|
| 377 |
-
|
| 378 |
-
elif score >= 4:
|
| 379 |
-
chosen_model = "llama-3.3-70b-versatile"
|
| 380 |
-
provider = "groq"
|
| 381 |
-
|
| 382 |
-
elif score >= 4:
|
| 383 |
chosen_model = "meta-llama/llama-4-scout-17b-16e-instruct"
|
| 384 |
provider = "groq"
|
| 385 |
-
|
| 386 |
-
|
| 387 |
-
|
| 388 |
-
|
| 389 |
-
|
| 390 |
-
|
| 391 |
-
|
| 392 |
-
|
| 393 |
-
|
| 394 |
-
|
| 395 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 396 |
|
| 397 |
await check_chat_rate_limit(request, authorization, x_client_id)
|
| 398 |
|
|
|
|
| 358 |
if kw in prompt_text:
|
| 359 |
score += 1
|
| 360 |
|
| 361 |
+
chosen_model = "llama-3.1-8b-instant"
|
| 362 |
provider = "groq"
|
| 363 |
+
has_images = contains_images(messages)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 364 |
|
| 365 |
+
if has_images:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 366 |
chosen_model = "meta-llama/llama-4-scout-17b-16e-instruct"
|
| 367 |
provider = "groq"
|
| 368 |
+
else:
|
| 369 |
+
if score > 10:
|
| 370 |
+
score = 10
|
| 371 |
+
if uses_tools:
|
| 372 |
+
if score >= 4:
|
| 373 |
+
chosen_model = "openai/gpt-oss-120b"
|
| 374 |
+
else:
|
| 375 |
+
chosen_model = "openai/gpt-oss-20b"
|
| 376 |
+
provider = "groq"
|
| 377 |
+
|
| 378 |
+
elif code_present:
|
| 379 |
+
|
| 380 |
+
if code_heavy and score >= 6:
|
| 381 |
+
chosen_model = "gpt-oss-120b"
|
| 382 |
+
provider = "cerebras"
|
| 383 |
+
|
| 384 |
+
elif score >= 4:
|
| 385 |
+
chosen_model = "llama-3.3-70b-versatile"
|
| 386 |
+
provider = "groq"
|
| 387 |
+
|
| 388 |
+
elif score >= 4:
|
| 389 |
+
chosen_model = "meta-llama/llama-4-scout-17b-16e-instruct"
|
| 390 |
+
provider = "groq"
|
| 391 |
+
|
| 392 |
+
if provider == "groq" and (
|
| 393 |
+
total_chars > MAX_GROQ_PROMPT_CHARS or total_bytes > MAX_GROQ_PROMPT_BYTES
|
| 394 |
+
):
|
| 395 |
+
raise HTTPException(
|
| 396 |
+
status_code=413,
|
| 397 |
+
detail=(
|
| 398 |
+
f"Prompt exceeds Groq-safe size ({total_chars} chars, {total_bytes} bytes). "
|
| 399 |
+
f"Max Groq-safe size is {MAX_GROQ_PROMPT_CHARS} chars or {MAX_GROQ_PROMPT_BYTES} bytes."
|
| 400 |
+
),
|
| 401 |
+
)
|
| 402 |
|
| 403 |
await check_chat_rate_limit(request, authorization, x_client_id)
|
| 404 |
|