vision-agent 0.2.237__py3-none-any.whl → 0.2.238__py3-none-any.whl

vision_agent/configs/anthropic_openai_config.py

@@ -96,13 +96,24 @@ class Config(BaseModel):
         }
     )
 
+    # for get_tool_for_task
+    od_judge: Type[LMM] = Field(default=AnthropicLMM)
+    od_judge_kwargs: dict = Field(
+        default_factory=lambda: {
+            "model_name": "claude-3-5-sonnet-20241022",
+            "temperature": 0.0,
+            "image_size": 512,
+        }
+    )
+
     # for suggestions module
-    suggester: Type[LMM] = Field(default=AnthropicLMM)
+    suggester: Type[LMM] = Field(default=OpenAILMM)
     suggester_kwargs: dict = Field(
         default_factory=lambda: {
-            "model_name": "claude-3-5-sonnet-20241022",
+            "model_name": "o1",
             "temperature": 1.0,
-            "image_size": 768,
+            "image_detail": "high",
+            "image_size": 1024,
         }
     )
 
@@ -143,6 +154,9 @@ class Config(BaseModel):
     def create_tool_chooser(self) -> LMM:
         return self.tool_chooser(**self.tool_chooser_kwargs)
 
+    def create_od_judge(self) -> LMM:
+        return self.od_judge(**self.od_judge_kwargs)
+
     def create_suggester(self) -> LMM:
         return self.suggester(**self.suggester_kwargs)
 

vision_agent/sim/__init__.py

@@ -2,6 +2,7 @@ from .sim import (
     AzureSim,
     OllamaSim,
     Sim,
+    StellaSim,
     get_tool_recommender,
     load_cached_sim,
     load_sim,

vision_agent/tools/planner_tools.py

@@ -368,6 +368,15 @@ def get_tool_for_task(
     tool_tester = CONFIG.create_tool_tester()
     tool_chooser = CONFIG.create_tool_chooser()
 
+    if isinstance(images, list):
+        if len(images) > 0 and isinstance(images[0], dict):
+            if all(["frame" in image for image in images]):
+                images = [image["frame"] for image in images]
+            else:
+                raise ValueError(
+                    f"Expected a list of numpy arrays or a dictionary of strings to lists of numpy arrays, got a list of dictionaries instead: {images}"
+                )
+
     if isinstance(images, list):
         images = {"image": images}
 
@@ -410,6 +419,15 @@ def get_tool_for_task_human_reviewer(
     # NOTE: this will have the same documentation as get_tool_for_task
     tool_tester = CONFIG.create_tool_tester()
 
+    if isinstance(images, list):
+        if len(images) > 0 and isinstance(images[0], dict):
+            if all(["frame" in image for image in images]):
+                images = [image["frame"] for image in images]
+            else:
+                raise ValueError(
+                    f"Expected a list of numpy arrays or a dictionary of strings to lists of numpy arrays, got a list of dictionaries instead: {images}"
+                )
+
     if isinstance(images, list):
         images = {"image": images}
 
@@ -424,6 +442,9 @@ def get_tool_for_task_human_reviewer(
                 Image.fromarray(image).save(image_path)
                 image_paths.append(image_path)
 
+        # run no more than 3 images or else it overloads the LLM
+        image_paths = image_paths[:3]
+
         tools = [
             t.__name__
             for t in get_tools()

{vision_agent-0.2.237.dist-info → vision_agent-0.2.238.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: vision-agent
-Version: 0.2.237
+Version: 0.2.238
 Summary: Toolset for Vision Agent
 Author: Landing AI
 Author-email: dev@landing.ai
@@ -15,7 +15,6 @@ Requires-Dist: e2b (>=0.17.2a50,<0.18.0)
 Requires-Dist: e2b-code-interpreter (==0.0.11a37)
 Requires-Dist: flake8 (>=7.0.0,<8.0.0)
 Requires-Dist: ipykernel (>=6.29.4,<7.0.0)
-Requires-Dist: langsmith (>=0.1.58,<0.2.0)
 Requires-Dist: libcst (>=1.5.0,<2.0.0)
 Requires-Dist: matplotlib (>=3.9.2,<4.0.0)
 Requires-Dist: nbclient (>=0.10.0,<0.11.0)

{vision_agent-0.2.237.dist-info → vision_agent-0.2.238.dist-info}/RECORD

@@ -21,7 +21,7 @@ vision_agent/clients/http.py,sha256=k883i6M_4nl7zwwHSI-yP5sAgQZIDPM1nrKD6YFJ3Xs,
 vision_agent/clients/landing_public_api.py,sha256=Vz9lldtNbaJRWzT7T8-uQrC-dMnt47LIsDrxHgoVdEw,1492
 vision_agent/configs/__init__.py,sha256=Iu75-w9_nlPmnB_qKA7nYaaaHf7xtTrDmK8N4v2WV34,27
 vision_agent/configs/anthropic_config.py,sha256=T1UuESgiY8913A6wA42P7-cg8FTk9-LkJpyywo7OnIQ,4298
-vision_agent/configs/anthropic_openai_config.py,sha256=YQjFxmlxppn5L55dJjK_v1myBJQ_V5J4q25pmUtwTOU,4310
+vision_agent/configs/anthropic_openai_config.py,sha256=rUz5zca4Pn5dTUwJXiJzRDYua5PWizApCKI3y0zOvhc,4699
 vision_agent/configs/config.py,sha256=rUz5zca4Pn5dTUwJXiJzRDYua5PWizApCKI3y0zOvhc,4699
 vision_agent/configs/openai_config.py,sha256=v2_AIY89d7LKWn4uqA2G047U2IdmnqZrGH2Iww9gRIw,4498
 vision_agent/fonts/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -32,11 +32,11 @@ vision_agent/models/__init__.py,sha256=qAdygB-0EsmxMHNzYTPNM6tAF8Fym95gm9bsHJafd
 vision_agent/models/agent_types.py,sha256=dIdxATH_PP76pD5Wfo0oofWt6iPQh0vpf48QbEQSzhs,2472
 vision_agent/models/lmm_types.py,sha256=v04h-NjbczHOIN8UWa1vvO5-1BDuZ4JQhD2mge1cXmw,305
 vision_agent/models/tools_types.py,sha256=8hYf2OZhI58gvf65KGaeGkt4EQ56nwLFqIQDPHioOBc,2339
-vision_agent/sim/__init__.py,sha256=XYL4BKCB-pamJzCR1y2d5lC8FL64WGu0oEzWzLfguAQ,120
+vision_agent/sim/__init__.py,sha256=Aouz6HEPPTYcLxR5_0fTYCL1OvPKAH1RMWAF90QXAlA,135
 vision_agent/sim/sim.py,sha256=VSU_1rYd4ifvF45xKWBEYugxdeeEQVpj0QL6rjx49i4,9801
 vision_agent/tools/__init__.py,sha256=T-MPNBVbvWtfo71hobaZsdYzQ52oyymolk_OAb2Pq_g,2463
 vision_agent/tools/meta_tools.py,sha256=-heMwGkx0hX_9zUp1dgBqsJpVnl6Y6tErMsjFy0dwLM,28652
-vision_agent/tools/planner_tools.py,sha256=iXyHjTBIWeQOCfcdQNufoQXfipHu_H38DIoK375FdnA,18492
+vision_agent/tools/planner_tools.py,sha256=orBTdJQz2NKoLuX9WE6XixaYuG305xz0UBYvZOiuquQ,19474
 vision_agent/tools/prompts.py,sha256=V1z4YJLXZuUl_iZ5rY0M5hHc_2tmMEUKr0WocXKGt4E,1430
 vision_agent/tools/tools.py,sha256=-xg5Msq5ZtHgaISpHnbq5rJ5MIERwfH6wPHg6KpaYjg,111457
 vision_agent/utils/__init__.py,sha256=mANUs_84VL-3gpZbXryvV2mWU623eWnRlJCSUHtMjuw,122
@@ -49,7 +49,7 @@ vision_agent/utils/tools_doc.py,sha256=yFue6KSXoa_Z1ngCdBEc4SdPZOWF1rVLeaHu02I8W
 vision_agent/utils/type_defs.py,sha256=BE12s3JNQy36QvauXHjwyeffVh5enfcvd4vTzSwvEZI,1384
 vision_agent/utils/video.py,sha256=Dt9_pqGgr63gmpurzisnpF6d9tr65-zxS1CccXdVuxk,6458
 vision_agent/utils/video_tracking.py,sha256=GM9qfeawqhmZVWoKrzw5-NETd4gEo7ImMfWtBnhC3bw,12086
-vision_agent-0.2.237.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-vision_agent-0.2.237.dist-info/METADATA,sha256=MkwC7kWf1f5E1ArMWdjNx_GGNgFwfWQtHbfyDzHN8EM,5755
-vision_agent-0.2.237.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
-vision_agent-0.2.237.dist-info/RECORD,,
+vision_agent-0.2.238.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+vision_agent-0.2.238.dist-info/METADATA,sha256=VnupHm4Iav889sO4JPGeWYM7902KwPKaJYem81_EDCk,5712
+vision_agent-0.2.238.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
+vision_agent-0.2.238.dist-info/RECORD,,