hud-python 0.1.0b3__py3-none-any.whl → 0.1.2a0__py3-none-any.whl

hud/__init__.py

@@ -9,7 +9,7 @@ from hud.environment import Environment, EvalSet, Observation, TaskResult
 from hud.gym import Gym
 from hud.run import Run
 
-__version__ = "0.1.0b3"
+__version__ = "0.1.2-alpha"
 
 __all__ = [
     "Environment",

hud/adapters/claude/adapter.py

@@ -2,10 +2,11 @@
 
 from __future__ import annotations
 
-from typing import Any
+from typing import Any, ClassVar
 
 from hud.adapters.common import CLA, Adapter
 from hud.adapters.common.types import (
+    CLAKey,
     ClickAction,
     DragAction,
     MoveAction,
@@ -20,11 +21,17 @@ from hud.adapters.common.types import (
 
 
 class ClaudeAdapter(Adapter):
+    KEY_MAP: ClassVar[dict[str, CLAKey]] = {"Return": "enter"}
+
     def __init__(self) -> None:
         super().__init__()
         self.agent_width = 1024  # Claude's preferred width
         self.agent_height = 768  # Claude's preferred height
 
+    def _map_key(self, key: str) -> CLAKey:
+        """Map a key to its standardized form."""
+        return self.KEY_MAP.get(key, key.lower())  # type: ignore
+
     def convert(self, data: Any) -> CLA:
         try:
             action_type = data.get("action")
@@ -32,10 +39,12 @@ class ClaudeAdapter(Adapter):
             if action_type == "key":
                 assert "text" in data
                 if "+" in data["text"]:
-                    keys = data["text"].split("+")
+                    keys: list[CLAKey] = [
+                        self._map_key(k) for k in (data["text"].split("+"))
+                    ]
                     assert len(keys) > 0
                     return PressAction(keys=keys)
-                return PressAction(keys=[data["text"]])
+                return PressAction(keys=[self._map_key(data["text"])])
 
             elif action_type == "type":
                 assert "text" in data
@@ -66,12 +75,19 @@ class ClaudeAdapter(Adapter):
                 assert len(coord) == 2
                 if (
                     len(self.memory) == 0
-                    or (self.memory[-1] is not MoveAction and self.memory[-1] is not ClickAction)
+                    or (
+                        self.memory[-1] is not MoveAction
+                        and self.memory[-1] is not ClickAction
+                    )
                     or self.memory[-1].point is None
                 ):
-                    raise ValueError("Left click drag must be preceded by a move or click action")
+                    raise ValueError(
+                        "Left click drag must be preceded by a move or click action"
+                    )
                 else:
-                    return DragAction(path=[self.memory[-1].point, Point(x=coord[0], y=coord[1])])
+                    return DragAction(
+                        path=[self.memory[-1].point, Point(x=coord[0], y=coord[1])]
+                    )
 
             elif action_type == "right_click":
                 assert "coordinate" in data
@@ -96,6 +112,17 @@ class ClaudeAdapter(Adapter):
                     point=Point(x=coord[0], y=coord[1]), button="left", pattern=[100]
                 )
 
+            elif action_type == "triple_click":
+                assert "coordinate" in data
+                coord = data["coordinate"]
+                assert isinstance(coord, list)
+                assert len(coord) == 2
+                return ClickAction(
+                    point=Point(x=coord[0], y=coord[1]),
+                    button="left",
+                    pattern=[100, 100],
+                )
+
             elif action_type == "scroll":
                 assert "scroll_direction" in data
                 direction = data["scroll_direction"]
@@ -112,7 +139,8 @@ class ClaudeAdapter(Adapter):
                     raise ValueError(f"Unsupported scroll direction: {direction}")
 
                 return ScrollAction(
-                    point=Point(x=data["coordinate"][0], y=data["coordinate"][1]), scroll=scroll
+                    point=Point(x=data["coordinate"][0], y=data["coordinate"][1]),
+                    scroll=scroll,
                 )
 
             elif action_type == "screenshot":
@@ -124,7 +152,6 @@ class ClaudeAdapter(Adapter):
             elif action_type == "wait":
                 assert "duration" in data
                 return WaitAction(time=data["duration"])
-
             else:
                 raise ValueError(f"Unsupported action type: {action_type}")
         except AssertionError:

hud/adapters/common/types.py

@@ -78,6 +78,10 @@ class PositionFetch(CLAAction):
     type: Literal["position"] = "position"
 
 
+class CustomAction(CLAAction):
+    type: Literal["custom"] = "custom"
+    action: str
+
 # Union of all possible actions
 CLA = Annotated[
     Union[
@@ -90,6 +94,7 @@ CLA = Annotated[
         DragAction,
         ScreenshotFetch,
         PositionFetch,
+        CustomAction,
     ],
     Field(discriminator="type"),
 ]

hud/environment.py

@@ -13,6 +13,17 @@ from hud.settings import settings
 if TYPE_CHECKING:
     from .adapters.common import Adapter
 
+
+class BaseResponseWithLogs(BaseModel):
+    """Base model for API responses that include logs."""
+    logs: str | None = None
+    error: str | None = None
+
+
+class RewardResponse(BaseResponseWithLogs):
+    reward: float
+
+
 logger = logging.getLogger("hud.environment")
 
 
@@ -199,19 +210,19 @@ class Environment:
             return self.adapter.adapt_list(action)
         return [self.adapter.adapt(action)]
 
-    async def evaluate(self) -> float:
+    async def evaluate(self) -> RewardResponse:
         """
         Get final evaluation score.
 
         Returns:
-            float: The evaluation score
+            RewardResponse: The evaluation response containing reward, logs, and possible error
         """
         data = await make_request(
             method="POST",
             url=f"{settings.base_url}/evaluation/{self.id}",
             api_key=settings.api_key,
         )
-        return data["reward"]
+        return RewardResponse(**data)
 
     async def close(self) -> None:
         """
@@ -273,6 +284,7 @@ class EvalSet:
         id: str,
         name: str,
         tasks: list[str] | None = None,
+        configs: dict[str, Any] | None = None,
     ) -> None:
         """
         Initialize an evaluation set.
@@ -285,6 +297,7 @@ class EvalSet:
         self.id = id
         self.name = name
         self.tasks = tasks or []
+        self.configs = configs or {}
 
     async def fetch_tasks(self) -> list[str]:
         """
@@ -298,5 +311,7 @@ class EvalSet:
             url=f"{settings.base_url}/evalsets/{self.id}/tasks",
             api_key=settings.api_key,
         )
+        # Extracts a list of task ids and list of config objects for the evalset
         self.tasks = data["tasks"]
+        self.configs = data["evalset"]
         return self.tasks

{hud_python-0.1.0b3.dist-info → hud_python-0.1.2a0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: hud-python
-Version: 0.1.0b3
+Version: 0.1.2a0
 Summary: SDK for the HUD evaluation platform.
 Project-URL: Homepage, https://github.com/Human-Data/hud-sdk
 Project-URL: Bug Tracker, https://github.com/Human-Data/hud-sdk/issues
@@ -57,9 +57,9 @@ Description-Content-Type: text/markdown
 
 # HUD
 
-A Python SDK for interacting with HUD environments and evaluation benchmarks for browser use and computer use models. Visit [hud.so](https://hud.so).
+A Python SDK for interacting with HUD environments and evaluation benchmarks for browser use and computer use models.
 
-> **Alpha Release Notice**: This SDK is currently in alpha status (v0.1.0-alpha). The API is evolving and may change in future releases as we gather feedback and improve functionality.
+> **Alpha Release Notice**: This SDK is currently in early release status. The API is evolving and may change in future releases as we gather feedback and improve functionality.
 
 [![PyPI version](https://img.shields.io/pypi/v/hud-python)](https://pypi.org/project/hud-python/)
 
@@ -70,13 +70,12 @@ A Python SDK for interacting with HUD environments and evaluation benchmarks for
 
 [RECOMMENDED] To set get started with an agent, see the [Claude Computer use example](https://github.com/Human-Data/hud-sdk/tree/main/examples).
 
-
-Otherwise, install the package with Python>=3.9:
+Install the package with Python>=3.9:
 ```bash
 pip install hud-python
 ```
 
-Make sure to setup your account [here](https://hud.so/settings) and add your API key to the environment variables:
+Make sure to setup your account with us (email founders@hud.so) and add your API key to the environment variables:
 ```bash
 HUD_API_KEY=<your-api-key>
 ```
@@ -117,20 +116,9 @@ if __name__ == "__main__":
     asyncio.run(main())
 ```
 
-## Features
-
-- Connect to HUD evaluation environments
-- Run benchmarks across various tasks
-- Support for different agent adapters
-- Asynchronous API
-
 ## Documentation
 
-For comprehensive guides, examples, and API reference, visit:
-- [Getting Started](https://docs.hud.so/introduction)
-- [Installation](https://docs.hud.so/installation)
-- [API Reference](https://docs.hud.so/api-reference)
-- [Examples](https://docs.hud.so/examples)
+For comprehensive guides, examples, and API reference, visit [our docs](https://docs.hud.so/introduction)
 
 ## License
 

{hud_python-0.1.0b3.dist-info → hud_python-0.1.2a0.dist-info}/RECORD

@@ -1,21 +1,21 @@
-hud/__init__.py,sha256=Xam6plJLHFqKPKcnVhwLQf4bsApDuxZ8BJF0FEAjkos,416
+hud/__init__.py,sha256=IEEme8kZA7zs9URZV-C35gqBOC5sxMEU7e6NjSAgXR4,420
 hud/client.py,sha256=7WHXTQhVK-T9Rj4ZooADE_c1pah5Bc1DJ9ZRqUyUnuQ,5724
-hud/environment.py,sha256=39tna-Cpzg9T6HqKebPARP2DXaF2n0xPr1W0qx8y160,8401
+hud/environment.py,sha256=9r8eK3OVqr-wpPGlhnrpuDt-z6FIp3S3oukTZ7swN3o,8899
 hud/gym.py,sha256=aanBHtlsXrJwrFax9SbXWwk_By-X8wE3M9deS-E_s4c,463
 hud/run.py,sha256=5ukjuRNLjj5fczaWxpR_5NebFbQpoy8w81eRYy309Vg,6401
 hud/settings.py,sha256=1ScSac0ta03LkckkH2gi6SyKY2M7nr15vRGugo2C_xs,1015
 hud/adapters/__init__.py,sha256=y3H7yMl7rC-rrXG2WvePdSojoNFSui02eYTH17Xd7OY,87
 hud/adapters/claude/__init__.py,sha256=i7QEF-29FLb9qxp1eYtXs-adIk_tG54tL-9g6d3xodk,100
-hud/adapters/claude/adapter.py,sha256=oi2lvO42g7i-L151tIWIGQGA80skcYRwzQ52-0f2OpA,4840
+hud/adapters/claude/adapter.py,sha256=ekYZixANKfx-4lENlXGaomh6Ecw4SRKtLWD5quGNWdM,5782
 hud/adapters/common/__init__.py,sha256=BjdZWJVs_AKtpFrt-tNsdQRjnz7D97DFEQirJ-r0mp8,118
 hud/adapters/common/adapter.py,sha256=SCtOuRjW5Szzd45LXCaqDEaKr2lhA-nIqSEMJ9KLsKI,5799
-hud/adapters/common/types.py,sha256=d9tIF06tjK7VCb-yBJ9epwHlXRHlObo9YWetrv33s8c,4511
+hud/adapters/common/types.py,sha256=Kgj0ZhiWOU6V95qxrvf-mMCvodLV_6rGBHwP1FQdMBk,4620
 hud/server/__init__.py,sha256=VPrhyyqg3inge9J7BjcmDBNJRuvkCA9ZDXS_R5Q8ZtY,129
 hud/server/requests.py,sha256=pPPaMpwqmA1RyWwzQN1ybgAnbSHJLeeIaW6MJwhJYks,6052
 hud/utils/__init__.py,sha256=0m8klSLnMLeIJT23ipBXfFACk4hNWPsA6ZNqZDpv6oY,99
 hud/utils/config.py,sha256=dze0BGE4q14omjj9822kL9BeiIgWQvJyuU29A2wa1SE,193
 hud/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-hud_python-0.1.0b3.dist-info/METADATA,sha256=mWp4cHyIzYuzxk3alNFspztWL2S8_6ZlvGP0UqzIF48,5146
-hud_python-0.1.0b3.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-hud_python-0.1.0b3.dist-info/licenses/LICENSE,sha256=IVdfcZ8xq5apYGJS5GzRLLbm9r03Aecxd03isi-3P9k,1075
-hud_python-0.1.0b3.dist-info/RECORD,,
+hud_python-0.1.2a0.dist-info/METADATA,sha256=oGcQYuTzPo9FSMeeARPiPws3ERRoWNPNu8tM8W_1Uow,4787
+hud_python-0.1.2a0.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+hud_python-0.1.2a0.dist-info/licenses/LICENSE,sha256=IVdfcZ8xq5apYGJS5GzRLLbm9r03Aecxd03isi-3P9k,1075
+hud_python-0.1.2a0.dist-info/RECORD,,