Merge pull request #324 from sudoleg/renovate/chromadb-1.x

chore(deps): update dependency chromadb to v1

Author: sudoleg

docker-compose.yml

@@ -4,24 +4,12 @@ networks:
 
 services:
   chromadb:
-    image: chromadb/chroma:0.6.3
+    image: chromadb/chroma:1.0.12
     container_name: chroma-db
     volumes:
-      # Be aware that indexed data are located in "/chroma/chroma/"
-      # Default configuration for persist_directory in chromadb/config.py
-      # Read more about deployments: https://docs.trychroma.com/deployment
-      - chroma:/chroma/chroma
-    environment:
-      - IS_PERSISTENT=TRUE
-      - ALLOW_RESET=TRUE
+      - chroma:/data
     ports:
       - "8000:8000"
-    healthcheck:
-      # Adjust below to match your container port
-      test: [ "CMD", "curl", "-f", "http://localhost:8000/api/v2/heartbeat" ]
-      interval: 30s
-      timeout: 10s
-      retries: 3
     networks:
       - net
 

modules/helpers.py

@@ -218,7 +218,14 @@ def num_tokens_from_string(string: str, model: str = "gpt-4o-mini") -> int:
 
     See https://cookbook.openai.com/examples/how_to_count_tokens_with_tiktoken
     """
-    encoding_name = tiktoken.encoding_name_for_model(model_name=model)
+
+    try:
+        encoding_name = tiktoken.encoding_name_for_model(model_name=model)
+    except KeyError as e:
+        logging.error("Couldn't map %s to tokenizer: %s", model, str(e))
+        # workaround until https://github.com/openai/tiktoken/issues/395 is fixed
+        encoding_name = "o200k_base"
+
     encoding = tiktoken.get_encoding(encoding_name)
     return len(encoding.encode(string))
 

modules/ui.py

@@ -95,8 +95,7 @@ def display_model_settings_sidebar():
             )
         if model != get_default_config_value("default_model.gpt"):
             st.warning(
-                """:warning: More advanced models (like gpt-4 and gpt-4o) have better reasoning capabilities and larger context windows. However, they likely won't make
-                a big difference for short videos and simple tasks, like plain summarization. Also, beware of the higher costs of other [flagship models](https://platform.openai.com/docs/models/flagship-models)."""
+                """:warning: Be aware of the higher costs and potentially higher latencies when using more advanced models (like gpt-4 and gpt-4o). You can see details (incl. costs) about the models and compare them [here](https://platform.openai.com/docs/models/compare)."""
             )
 
 

requirements.txt

@@ -8,8 +8,8 @@ watchdog==6.0.0
 pytest==8.3.5
 peewee==3.18.1
 python-dotenv==1.1.0
-chromadb==0.6.3
-langchain-chroma==0.2.3
+chromadb==1.0.12
+langchain-chroma==0.2.4
 randomname==0.2.1
 tiktoken==0.9.0
 openai-whisper==20240930