Merge branch 'main' into bump-11-2

Author: carenthomas

.github/workflows/notify-letta-cloud.yml

@@ -31,4 +31,4 @@ The demo will:
 3. Create an agent named "Clippy"
 4. Start an interactive chat session
 
-Type 'quit' or 'exit' to end the conversation.
+Type 'quit' or 'exit' to end the conversation.

examples/files/README.md

@@ -63,7 +63,6 @@
 # 1. From an existing file
 # 2. From a string by encoding it into a base64 string
 #
-#
 
 # 1. From an existing file
 # "rb" means "read binary"

examples/files/main.py

@@ -5,7 +5,7 @@
     __version__ = version("letta")
 except PackageNotFoundError:
     # Fallback for development installations
-    __version__ = "0.10.0"
+    __version__ = "0.11.0"
 
 if os.environ.get("LETTA_VERSION"):
     __version__ = os.environ["LETTA_VERSION"]

letta/__init__.py

@@ -10,6 +10,7 @@
 
 ADMIN_PREFIX = "/v1/admin"
 API_PREFIX = "/v1"
+OLLAMA_API_PREFIX = "/v1"
 OPENAI_API_PREFIX = "/openai"
 
 COMPOSIO_ENTITY_ENV_VAR_KEY = "COMPOSIO_ENTITY"
@@ -50,8 +51,9 @@
 # Max steps for agent loop
 DEFAULT_MAX_STEPS = 50
 
-# minimum context window size
+# context window size
 MIN_CONTEXT_WINDOW = 4096
+DEFAULT_CONTEXT_WINDOW = 32000
 
 # number of concurrent embedding requests to sent
 EMBEDDING_BATCH_SIZE = 200
@@ -63,6 +65,7 @@
 # embeddings
 MAX_EMBEDDING_DIM = 4096  # maximum supported embeding size - do NOT change or else DBs will need to be reset
 DEFAULT_EMBEDDING_CHUNK_SIZE = 300
+DEFAULT_EMBEDDING_DIM = 1024
 
 # tokenizers
 EMBEDDING_TO_TOKENIZER_MAP = {

letta/constants.py

@@ -139,10 +139,11 @@ def get_text_embedding(self, text: str):
 
 class OllamaEmbeddings:
 
+    # Uses OpenAI API standard
     # Format:
-    # curl http://localhost:11434/api/embeddings -d '{
+    # curl http://localhost:11434/v1/embeddings -d '{
     #   "model": "mxbai-embed-large",
-    #   "prompt": "Llamas are members of the camelid family"
+    #   "input": "Llamas are members of the camelid family"
     # }'
 
     def __init__(self, model: str, base_url: str, ollama_additional_kwargs: dict):
@@ -154,18 +155,18 @@ def get_text_embedding(self, text: str):
         import httpx
 
         headers = {"Content-Type": "application/json"}
-        json_data = {"model": self.model, "prompt": text}
+        json_data = {"model": self.model, "input": text}
         json_data.update(self.ollama_additional_kwargs)
 
         with httpx.Client() as client:
             response = client.post(
-                f"{self.base_url}/api/embeddings",
+                f"{self.base_url}/embeddings",
                 headers=headers,
                 json=json_data,
             )
 
         response_json = response.json()
-        return response_json["embedding"]
+        return response_json["data"][0]["embedding"]
 
 
 class GoogleEmbeddings: