lm-deluge 0.0.11__py3-none-any.whl → 0.0.13__py3-none-any.whl

lm_deluge/__init__.py

@@ -1,7 +1,15 @@
 from .client import LLMClient, SamplingParams, APIResponse
 from .prompt import Conversation, Message
+from .tool import Tool
 import dotenv
 
 dotenv.load_dotenv()
 
-__all__ = ["LLMClient", "SamplingParams", "APIResponse", "Conversation", "Message"]
+__all__ = [
+    "LLMClient",
+    "SamplingParams",
+    "APIResponse",
+    "Conversation",
+    "Message",
+    "Tool",
+]

lm_deluge/api_requests/anthropic.py

@@ -1,17 +1,94 @@
-import asyncio
 from aiohttp import ClientResponse
 import json
 import os
-import warnings
-from tqdm import tqdm
 from typing import Callable
 
-from lm_deluge.prompt import Conversation, Message, Text, ToolCall, Thinking
+from lm_deluge.prompt import (
+    Conversation,
+    Message,
+    Text,
+    ToolCall,
+    Thinking,
+    CachePattern,
+)
+from lm_deluge.tool import Tool
+from lm_deluge.usage import Usage
 from .base import APIRequestBase, APIResponse
 
 from ..tracker import StatusTracker
-from ..sampling_params import SamplingParams
+from ..config import SamplingParams
 from ..models import APIModel
+from ..computer_use.anthropic_tools import get_anthropic_cu_tools
+
+
+def _build_anthropic_request(
+    model: APIModel,
+    prompt: Conversation,
+    tools: list[Tool] | None,
+    sampling_params: SamplingParams,
+    cache_pattern: CachePattern | None = None,
+    computer_use: bool = False,
+    display_width: int = 1024,
+    display_height: int = 768,
+):
+    system_message, messages = prompt.to_anthropic(cache_pattern=cache_pattern)
+    request_header = {
+        "x-api-key": os.getenv(model.api_key_env_var),
+        "anthropic-version": "2023-06-01",
+        "content-type": "application/json",
+    }
+
+    # Add beta header for Computer Use
+    if computer_use:
+        request_header["anthropic-beta"] = "computer-use-2025-01-24"
+
+    request_json = {
+        "model": model.name,
+        "messages": messages,
+        "temperature": sampling_params.temperature,
+        "top_p": sampling_params.top_p,
+        "max_tokens": sampling_params.max_new_tokens,
+    }
+
+    # handle thinking
+    if model.reasoning_model and sampling_params.reasoning_effort:
+        # translate reasoning effort of low, medium, high to budget tokens
+        budget = {"low": 1024, "medium": 4096, "high": 16384}.get(
+            sampling_params.reasoning_effort
+        )
+        request_json["thinking"] = {
+            "type": "enabled",
+            "budget_tokens": budget,
+        }
+        request_json.pop("top_p")
+        request_json["temperature"] = 1.0
+        request_json["max_tokens"] += budget
+    else:
+        request_json["thinking"] = {"type": "disabled"}
+        if sampling_params.reasoning_effort:
+            print("ignoring reasoning_effort for non-reasoning model")
+    if system_message is not None:
+        request_json["system"] = system_message
+    if tools or computer_use:
+        tool_definitions = []
+        if tools:
+            tool_definitions.extend([tool.dump_for("anthropic") for tool in tools])
+        # Add Computer Use tools
+        if computer_use:
+            cu_tools = get_anthropic_cu_tools(
+                model=model.id,
+                display_width=display_width,  # todo: set from ComputerUseParams
+                display_height=display_height,
+            )
+            tool_definitions.extend(cu_tools)
+
+        # Add cache control to last tool if tools_only caching is specified
+        if cache_pattern == "tools_only" and tool_definitions:
+            tool_definitions[-1]["cache_control"] = {"type": "ephemeral"}
+
+        request_json["tools"] = tool_definitions
+
+    return request_json, request_header
 
 
 class AnthropicRequest(APIRequestBase):
@@ -24,17 +101,19 @@ class AnthropicRequest(APIRequestBase):
         prompt: Conversation,
         attempts_left: int,
         status_tracker: StatusTracker,
-        retry_queue: asyncio.Queue,
         results_arr: list,
         request_timeout: int = 30,
         sampling_params: SamplingParams = SamplingParams(),
-        pbar: tqdm | None = None,
         callback: Callable | None = None,
-        debug: bool = False,
         # for retries
         all_model_names: list[str] | None = None,
         all_sampling_params: list[SamplingParams] | None = None,
         tools: list | None = None,
+        cache: CachePattern | None = None,
+        # Computer Use support
+        computer_use: bool = False,
+        display_width: int = 1024,
+        display_height: int = 768,
     ):
         super().__init__(
             task_id=task_id,
@@ -42,70 +121,42 @@ class AnthropicRequest(APIRequestBase):
             prompt=prompt,
             attempts_left=attempts_left,
             status_tracker=status_tracker,
-            retry_queue=retry_queue,
             results_arr=results_arr,
             request_timeout=request_timeout,
             sampling_params=sampling_params,
-            pbar=pbar,
             callback=callback,
-            debug=debug,
             all_model_names=all_model_names,
             all_sampling_params=all_sampling_params,
             tools=tools,
+            cache=cache,
         )
+        self.computer_use = computer_use
+        self.display_width = display_width
+        self.display_height = display_height
         self.model = APIModel.from_registry(model_name)
         self.url = f"{self.model.api_base}/messages"
 
-        self.system_message, messages = prompt.to_anthropic()
-        self.request_header = {
-            "x-api-key": os.getenv(self.model.api_key_env_var),
-            "anthropic-version": "2023-06-01",
-            "content-type": "application/json",
-        }
+        # Lock images as bytes if caching is enabled
+        if cache is not None:
+            prompt.lock_images_as_bytes()
 
-        self.request_json = {
-            "model": self.model.name,
-            "messages": messages,
-            "temperature": self.sampling_params.temperature,
-            "top_p": self.sampling_params.top_p,
-            "max_tokens": self.sampling_params.max_new_tokens,
-        }
-        # handle thinking
-        if self.model.reasoning_model:
-            if sampling_params.reasoning_effort:
-                # translate reasoning effort of low, medium, high to budget tokens
-                budget = {"low": 1024, "medium": 4096, "high": 16384}.get(
-                    sampling_params.reasoning_effort
-                )
-                self.request_json["thinking"] = {
-                    "type": "enabled",
-                    "budget_tokens": budget,
-                }
-                self.request_json.pop("top_p")
-                self.request_json["temperature"] = 1.0
-                self.request_json["max_tokens"] += (
-                    budget  # assume max tokens is max completion tokens
-                )
-            else:
-                # no thinking
-                self.request_json["thinking"] = {"type": "disabled"}
-        else:
-            if sampling_params.reasoning_effort:
-                warnings.warn(
-                    f"Ignoring reasoning_effort param for non-reasoning model: {model_name}"
-                )
-        if self.system_message is not None:
-            self.request_json["system"] = self.system_message
-        if tools:
-            self.request_json["tools"] = [tool.dump_for("anthropic") for tool in tools]
+        self.request_json, self.request_header = _build_anthropic_request(
+            self.model,
+            prompt,
+            tools,
+            sampling_params,
+            cache,
+            computer_use,
+            display_width,
+            display_height,
+        )
 
     async def handle_response(self, http_response: ClientResponse) -> APIResponse:
         is_error = False
         error_message = None
         thinking = None
         content = None
-        input_tokens = None
-        output_tokens = None
+        usage = None
         status_code = http_response.status
         mimetype = http_response.headers.get("Content-Type", None)
         rate_limits = {}
@@ -118,8 +169,6 @@ class AnthropicRequest(APIRequestBase):
             "anthropic-ratelimit-tokens-reset",
         ]:
             rate_limits[header] = http_response.headers.get(header, None)
-        if self.debug:
-            print(f"Rate limits: {rate_limits}")
         if status_code >= 200 and status_code < 300:
             try:
                 data = await http_response.json()
@@ -143,8 +192,7 @@ class AnthropicRequest(APIRequestBase):
                         )
 
                 content = Message("assistant", parts)
-                input_tokens = data["usage"]["input_tokens"]
-                output_tokens = data["usage"]["output_tokens"]
+                usage = Usage.from_anthropic_usage(data["usage"])
             except Exception as e:
                 is_error = True
                 error_message = (
@@ -182,6 +230,5 @@ class AnthropicRequest(APIRequestBase):
             thinking=thinking,
             model_internal=self.model_name,
             sampling_params=self.sampling_params,
-            input_tokens=input_tokens,
-            output_tokens=output_tokens,
+            usage=usage,
         )

lm_deluge/api_requests/base.py

@@ -1,19 +1,21 @@
-import aiohttp
 import asyncio
 import json
 import random
-from tqdm import tqdm
-from dataclasses import dataclass
+import traceback
 from abc import ABC, abstractmethod
+from dataclasses import dataclass
 from typing import Callable
 
-from lm_deluge.prompt import Conversation, Message
+import aiohttp
+from aiohttp import ClientResponse
+
+from lm_deluge.prompt import CachePattern, Conversation, Message
+from lm_deluge.usage import Usage
 
-from ..tracker import StatusTracker
-from ..sampling_params import SamplingParams
-from ..models import APIModel
+from ..config import SamplingParams
 from ..errors import raise_if_modal_exception
-from aiohttp import ClientResponse
+from ..models import APIModel
+from ..tracker import StatusTracker
 
 
 @dataclass
@@ -29,9 +31,8 @@ class APIResponse:
     is_error: bool | None
     error_message: str | None
 
-    # completion information
-    input_tokens: int | None
-    output_tokens: int | None
+    # completion information - unified usage tracking
+    usage: Usage | None = None
 
     # response content - structured format
     content: Message | None = None
@@ -48,6 +49,10 @@ class APIResponse:
     retry_with_different_model: bool | None = False
     # set to true if should NOT retry with the same model (unrecoverable error)
     give_up_if_no_other_models: bool | None = False
+    # OpenAI Responses API specific - used for computer use continuation
+    response_id: str | None = None
+    # Raw API response for debugging
+    raw_response: dict | None = None
 
     @property
     def completion(self) -> str | None:
@@ -56,6 +61,26 @@ class APIResponse:
             return self.content.completion
         return None
 
+    @property
+    def input_tokens(self) -> int | None:
+        """Get input tokens from usage object."""
+        return self.usage.input_tokens if self.usage else None
+
+    @property
+    def output_tokens(self) -> int | None:
+        """Get output tokens from usage object."""
+        return self.usage.output_tokens if self.usage else None
+
+    @property
+    def cache_read_tokens(self) -> int | None:
+        """Get cache read tokens from usage object."""
+        return self.usage.cache_read_tokens if self.usage else None
+
+    @property
+    def cache_write_tokens(self) -> int | None:
+        """Get cache write tokens from usage object."""
+        return self.usage.cache_write_tokens if self.usage else None
+
     def __post_init__(self):
         # calculate cost & get external model name
         self.id = int(self.id)
@@ -63,14 +88,13 @@ class APIResponse:
         self.model_external = api_model.name
         self.cost = None
         if (
-            self.input_tokens is not None
-            and self.output_tokens is not None
+            self.usage is not None
             and api_model.input_cost is not None
             and api_model.output_cost is not None
         ):
             self.cost = (
-                self.input_tokens * api_model.input_cost / 1e6
-                + self.output_tokens * api_model.output_cost / 1e6
+                self.usage.input_tokens * api_model.input_cost / 1e6
+                + self.usage.output_tokens * api_model.output_cost / 1e6
             )
         elif self.content is not None and self.completion is not None:
             print(
@@ -90,8 +114,7 @@ class APIResponse:
             "error_message": self.error_message,
             "completion": self.completion,  # computed property
             "content": self.content.to_log() if self.content else None,
-            "input_tokens": self.input_tokens,
-            "output_tokens": self.output_tokens,
+            "usage": self.usage.to_dict() if self.usage else None,
             "finish_reason": self.finish_reason,
             "cost": self.cost,
         }
@@ -107,6 +130,10 @@ class APIResponse:
             # Backward compatibility: create a Message with just text
             content = Message.ai(data["completion"])
 
+        usage = None
+        if "usage" in data and data["usage"] is not None:
+            usage = Usage.from_dict(data["usage"])
+
         return cls(
             id=data.get("id", random.randint(0, 1_000_000_000)),
             model_internal=data["model_internal"],
@@ -115,8 +142,7 @@ class APIResponse:
             status_code=data["status_code"],
             is_error=data["is_error"],
             error_message=data["error_message"],
-            input_tokens=data["input_tokens"],
-            output_tokens=data["output_tokens"],
+            usage=usage,
             content=content,
             thinking=data.get("thinking"),
             model_external=data.get("model_external"),
@@ -155,19 +181,15 @@ class APIRequestBase(ABC):
         prompt: Conversation,
         attempts_left: int,
         status_tracker: StatusTracker,
-        retry_queue: asyncio.Queue,
         # needed in order to retry with a different model and not throw the output away
         results_arr: list["APIRequestBase"],
         request_timeout: int = 30,
         sampling_params: SamplingParams = SamplingParams(),
-        logprobs: bool = False,
-        top_logprobs: int | None = None,
-        pbar: tqdm | None = None,
         callback: Callable | None = None,
-        debug: bool = False,
         all_model_names: list[str] | None = None,
         all_sampling_params: list[SamplingParams] | None = None,
         tools: list | None = None,
+        cache: CachePattern | None = None,
     ):
         if all_model_names is None:
             raise ValueError("all_model_names must be provided.")
@@ -177,19 +199,15 @@ class APIRequestBase(ABC):
         self.prompt = prompt
         self.attempts_left = attempts_left
         self.status_tracker = status_tracker
-        self.retry_queue = retry_queue
         self.request_timeout = request_timeout
         self.sampling_params = sampling_params
-        self.logprobs = logprobs  # len(completion) logprobs
-        self.top_logprobs = top_logprobs
-        self.pbar = pbar
         self.callback = callback
         self.num_tokens = prompt.count_tokens(sampling_params.max_new_tokens)
         self.results_arr = results_arr
-        self.debug = debug
         self.all_model_names = all_model_names
         self.all_sampling_params = all_sampling_params
         self.tools = tools
+        self.cache: CachePattern | None = cache
         self.result = []  # list of APIResponse objects from each attempt
 
         # these should be set in the __init__ of the subclass
@@ -199,8 +217,7 @@ class APIRequestBase(ABC):
         self.region = None
 
     def increment_pbar(self):
-        if self.pbar is not None:
-            self.pbar.update(1)
+        self.status_tracker.increment_pbar()
 
     def call_callback(self):
         if self.callback is not None:
@@ -209,7 +226,6 @@ class APIRequestBase(ABC):
 
     def handle_success(self, data):
         self.call_callback()
-        self.increment_pbar()
         self.status_tracker.task_succeeded(self.task_id)
 
     def handle_error(self, create_new_request=False, give_up_if_no_other_models=False):
@@ -230,7 +246,8 @@ class APIRequestBase(ABC):
         if self.attempts_left > 0:
             self.attempts_left -= 1
             if not create_new_request:
-                self.retry_queue.put_nowait(self)
+                assert self.status_tracker.retry_queue
+                self.status_tracker.retry_queue.put_nowait(self)
                 return
             else:
                 # make sure we have another model to send it to besides the current one
@@ -244,7 +261,8 @@ class APIRequestBase(ABC):
                         print(
                             f"No other models to try for task {self.task_id}. Retrying with same model."
                         )
-                        self.retry_queue.put_nowait(self)
+                        assert self.status_tracker.retry_queue
+                        self.status_tracker.retry_queue.put_nowait(self)
                 else:
                     # two things to change: model_name and sampling_params
                     new_model_name = self.model_name
@@ -269,20 +287,21 @@ class APIRequestBase(ABC):
                         prompt=self.prompt,
                         attempts_left=self.attempts_left,
                         status_tracker=self.status_tracker,
-                        retry_queue=self.retry_queue,
                         results_arr=self.results_arr,
                         request_timeout=self.request_timeout,
                         sampling_params=new_sampling_params,
-                        logprobs=self.logprobs,
-                        top_logprobs=self.top_logprobs,
-                        pbar=self.pbar,
                         callback=self.callback,
                         all_model_names=self.all_model_names,
                         all_sampling_params=self.all_sampling_params,
                         tools=self.tools,
+                        cache=self.cache,
+                        computer_use=getattr(self, "computer_use", False),
+                        display_width=getattr(self, "display_width", 1024),
+                        display_height=getattr(self, "display_height", 768),
                     )
                     # PROBLEM: new request is never put into results array, so we can't get the result.
-                    self.retry_queue.put_nowait(new_request)
+                    assert self.status_tracker.retry_queue
+                    self.status_tracker.retry_queue.put_nowait(self)
                     # SOLUTION: just need to make sure it's deduplicated by task_id later.
                     self.results_arr.append(new_request)
         else:
@@ -323,14 +342,15 @@ class APIRequestBase(ABC):
                     is_error=True,
                     error_message="Request timed out (terminated by client).",
                     content=None,
-                    input_tokens=None,
-                    output_tokens=None,
+                    usage=None,
                 )
             )
             self.handle_error(create_new_request=False)
 
         except Exception as e:
             raise_if_modal_exception(e)
+            tb = traceback.format_exc()
+            print(tb)
             self.result.append(
                 APIResponse(
                     id=self.task_id,
@@ -341,8 +361,7 @@ class APIRequestBase(ABC):
                     is_error=True,
                     error_message=f"Unexpected {type(e).__name__}: {str(e) or 'No message.'}",
                     content=None,
-                    input_tokens=None,
-                    output_tokens=None,
+                    usage=None,
                 )
             )
             # maybe consider making True?
@@ -359,41 +378,75 @@ def create_api_request(
     prompt: Conversation,
     attempts_left: int,
     status_tracker: StatusTracker,
-    retry_queue: asyncio.Queue,
     results_arr: list["APIRequestBase"],
     request_timeout: int = 30,
     sampling_params: SamplingParams = SamplingParams(),
-    logprobs: bool = False,
-    top_logprobs: int | None = None,
-    pbar: tqdm | None = None,
     callback: Callable | None = None,
     all_model_names: list[str] | None = None,
     all_sampling_params: list[SamplingParams] | None = None,
     tools: list | None = None,
+    cache: CachePattern | None = None,
+    computer_use: bool = False,
+    display_width: int = 1024,
+    display_height: int = 768,
+    use_responses_api: bool = False,
 ) -> APIRequestBase:
     from .common import CLASSES  # circular import so made it lazy, does this work?
 
     model_obj = APIModel.from_registry(model_name)
-    request_class = CLASSES.get(model_obj.api_spec, None)
+
+    # Choose API spec based on use_responses_api flag and model support
+    api_spec = model_obj.api_spec
+    if use_responses_api and model_obj.supports_responses and api_spec == "openai":
+        api_spec = "openai-responses"
+
+    request_class = CLASSES.get(api_spec, None)
     if request_class is None:
-        raise ValueError(f"Unsupported API spec: {model_obj.api_spec}")
-    kwargs = (
-        {} if not logprobs else {"logprobs": logprobs, "top_logprobs": top_logprobs}
-    )
+        raise ValueError(f"Unsupported API spec: {api_spec}")
+    kwargs = {}
+    # Add computer_use to kwargs if the request class supports it
+    model_obj = APIModel.from_registry(model_name)
+    if computer_use and api_spec in ["anthropic", "bedrock", "openai-responses"]:
+        kwargs.update(
+            {
+                "computer_use": computer_use,
+                "display_width": display_width,
+                "display_height": display_height,
+            }
+        )
+
     return request_class(
         task_id=task_id,
         model_name=model_name,
         prompt=prompt,
         attempts_left=attempts_left,
         status_tracker=status_tracker,
-        retry_queue=retry_queue,
         results_arr=results_arr,
         request_timeout=request_timeout,
         sampling_params=sampling_params,
-        pbar=pbar,
         callback=callback,
         all_model_names=all_model_names,
         all_sampling_params=all_sampling_params,
         tools=tools,
+        cache=cache,
         **kwargs,
     )
+
+
+def deduplicate_responses(results: list[APIRequestBase]) -> list[APIResponse]:
+    deduplicated = {}
+    for request in results:
+        if request.task_id not in deduplicated:
+            deduplicated[request.task_id] = request.result[-1]
+        else:
+            current_response: APIResponse = deduplicated[request.task_id]
+            # only replace if the current request has no completion and the new one does
+            if (
+                request.result[-1].completion is not None
+                and current_response.completion is None
+            ):
+                deduplicated[request.task_id] = request.result[-1]
+
+    output = [deduplicated[request.task_id] for request in results]
+
+    return output