fix: pass encoding_format="float" in LiteLLM embedding calls (#434)

DeepInfra rejects requests when encoding_format is null. LiteLLM sets
it to None by default, so we explicitly pass "float" — the only format
compatible with our list[list[float]] return type.

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

Author: franchb

hindsight-api/hindsight_api/engine/embeddings.py

@@ -794,6 +794,7 @@ async def initialize(self) -> None:
                 "model": self.model,
                 "input": ["test"],
                 "api_key": self.api_key,
+                "encoding_format": "float",
             }
             if self.api_base:
                 embed_kwargs["api_base"] = self.api_base
@@ -840,6 +841,7 @@ def encode(self, texts: list[str]) -> list[list[float]]:
                     "model": self.model,
                     "input": batch,
                     "api_key": self.api_key,
+                    "encoding_format": "float",
                 }
                 if self.api_base:
                     embed_kwargs["api_base"] = self.api_base

hindsight-api/tests/test_litellm_sdk_embeddings.py

@@ -85,6 +85,7 @@ async def test_initialization_success(self, mock_litellm):
                 model="cohere/embed-english-v3.0",
                 input=["test"],
                 api_key="test_key",
+                encoding_format="float",
             )
 
     async def test_initialization_missing_package(self):
@@ -137,6 +138,7 @@ async def test_encode_single_text(self, embeddings, mock_litellm):
             model="cohere/embed-english-v3.0",
             input=["Hello world"],
             api_key="test_key",
+            encoding_format="float",
         )
 
     async def test_encode_multiple_texts(self, embeddings, mock_litellm):