Fix: use 'document' prompt for index embeddings, not 'query'

Fix: model filtering should use OR not AND for likes/downloads

Remove dead get_embedding_function and unused import

Fix: embed queries manually to avoid ChromaDB embedding function conflict

Fix: debug mount path, use load_dataset as Hub fallback

Load pre-computed embeddings from Hub dataset on startup

Fix NameError: rename DATA_DIR references to MOUNT_DIR

Copy ChromaDB from bucket to /tmp on startup

Debug: log mount permissions before opening ChromaDB

Update dependencies: chromadb 1.0.12, add transformers

Migrate off persistent storage to bucket-mounted ChromaDB

refactor: optimize database setup logging and limit sample sizes for clarity

refactor: enhance database setup logging and retrieve all metadata for accurate timestamp

refactor: adjust database setup call

refactor: update model and embedding configurations, enhance logging for database setup

refactor: remove commented-out flashinfer-python URL and clean up imports

add flashinfer-python dependency and related comments

refactor: import vllm module for improved functionality

refactor: add logging for CUDA and PyTorch versions

refactor: simplify LLM initialization by removing gpu_memory_utilization parameter

refactor: update LLM initialization parameters for improved memory utilization and chunked prefill

refactor: remove incorrect batch size parameters from LLM generate function

refactor: remove FLASHINFER environment variable and update LLM initialization for batch processing

don't set flashinfer

auto set torch

fix package name

try extra urls

remove flashinfer

fix: Remove custom index URLs to resolve dependency conflicts

remove pinning of flash-infer

fix toml formatting

docs: Update system dependencies and add tool index URLs in generate_summaries_uv.py

add setuptools

docs: Add system dependencies to generate_summaries_uv.py

load weights manually

env variables

add hf_transfer

summaries script

update requirements to use latest versions of chromadb and aiohttp, and adjust protobuf version

update requirements.txt to reflect changes in hf-xet dependency and clarify source

update requirements.in to replace huggingface_hub with hf-xet and restore httpx[http2]

update requirements.in to modify huggingface_hub dependencies

update dependencies in requirements.txt to latest versions

refactor schema retrieval in database setup to use collect_schema method

refactor model data loading to handle optional 'param_count' column and improve logging

add dotenv support for environment variables and Hugging Face login

refactor trending models fetching to improve parameter filtering and logging

add parameter count filters to model search and trending endpoints

add param_count to model metadata and update requirements for compatibility

add 'trending' option to sorting criteria in find_similar_models function