Max new tokens fix (#118)

* Lower max_new_tokens for OpenAI models

* updating configs

---------

Co-authored-by: Thibault LSDC <[email protected]>
Co-authored-by: ThibaultLSDC <[email protected]>

Author: gasse

src/agentlab/llm/llm_configs.py

@@ -20,41 +20,41 @@
     "openai/gpt-4o-mini-2024-07-18": OpenAIModelArgs(
         model_name="gpt-4o-mini-2024-07-18",
         max_total_tokens=128_000,
-        max_input_tokens=100_000,
-        max_new_tokens=28_000,
+        max_input_tokens=128_000,
+        max_new_tokens=16_384,
         vision_support=True,
     ),
     "openai/gpt-4-1106-preview": OpenAIModelArgs(
         model_name="gpt-4-1106-preview",
         max_total_tokens=128_000,
-        max_input_tokens=100_000,
-        max_new_tokens=28_000,
+        max_input_tokens=128_000,
+        max_new_tokens=4_096,
     ),
     "openai/gpt-4-vision-preview": OpenAIModelArgs(
         model_name="gpt-4-vision-preview",
         max_total_tokens=128_000,
-        max_input_tokens=100_000,
-        max_new_tokens=28_000,  # I think this model has very small default value if we don't set max_new_tokens
+        max_input_tokens=128_000,
+        max_new_tokens=16_384,  # I think this model has very small default value if we don't set max_new_tokens
         vision_support=True,
     ),
     "openai/gpt-4o-2024-05-13": OpenAIModelArgs(
         model_name="gpt-4o-2024-05-13",
         max_total_tokens=128_000,
-        max_input_tokens=100_000,
-        max_new_tokens=28_000,  # I think this model has very small default value if we don't set max_new_tokens
+        max_input_tokens=128_000,
+        max_new_tokens=4_096,  # I think this model has very small default value if we don't set max_new_tokens
         vision_support=True,
     ),
     "openai/gpt-3.5-turbo-0125": OpenAIModelArgs(
         model_name="gpt-3.5-turbo-0125",
         max_total_tokens=16_384,
-        max_input_tokens=15_000,
-        max_new_tokens=1_000,
+        max_input_tokens=16_384,
+        max_new_tokens=4096,
     ),
     "openai/gpt-3.5-turbo-1106": OpenAIModelArgs(
         model_name="gpt-3.5-turbo-1106",
         max_total_tokens=16_384,
-        max_input_tokens=15_000,
-        max_new_tokens=1_000,
+        max_input_tokens=16_384,
+        max_new_tokens=4096,
     ),
     "azure/gpt-35-turbo/gpt-35-turbo": AzureModelArgs(
         model_name="gpt-35-turbo",
@@ -68,23 +68,23 @@
         deployment_name="gpt-4o-2024-05-13",
         max_total_tokens=128_000,
         max_input_tokens=100_000,
-        max_new_tokens=28_000,
+        max_new_tokens=16_384,
         vision_support=True,
     ),
     "azure/gpt-4o-2024-08-06": AzureModelArgs(
         model_name="gpt-4o",
         deployment_name="gpt-4o-2024-08-06",
         max_total_tokens=128_000,
-        max_input_tokens=100_000,
-        max_new_tokens=28_000,
+        max_input_tokens=128_000,
+        max_new_tokens=16_384,
         vision_support=True,
     ),
     "azure/gpt-4o-mini-2024-07-18": AzureModelArgs(
         model_name="gpt-4o-mini",
         deployment_name="gpt-4o-mini-2024-07-18",
         max_total_tokens=128_000,
-        max_input_tokens=100_000,
-        max_new_tokens=28_000,
+        max_input_tokens=128_000,
+        max_new_tokens=16_384,
         vision_support=True,
     ),
     # ---------------- OSS LLMs ----------------#
@@ -151,8 +151,8 @@
     "openrouter/anthropic/claude-3.5-sonnet:beta": OpenRouterModelArgs(
         model_name="anthropic/claude-3.5-sonnet:beta",
         max_total_tokens=200_000,
-        max_input_tokens=160_000,
-        max_new_tokens=40_000,
+        max_input_tokens=200_000,
+        max_new_tokens=8_192,
         temperature=1e-1,
         vision_support=True,
     ),
@@ -166,8 +166,8 @@
     "openrouter/openai/o1-mini-2024-09-12": OpenRouterModelArgs(
         model_name="openai/o1-mini-2024-09-12",
         max_total_tokens=128_000,
-        max_input_tokens=100_000,
-        max_new_tokens=28_000,
+        max_input_tokens=128_000,
+        max_new_tokens=64_000,
         temperature=1e-1,
     ),
 }