Highlights
🛣️ Inference of longer sequences. We extended the max sequence length to 8192 tokens for Llama 2 and added chunking to avoid server out-of-memory errors (happened when processing long prefixes). This became possible thanks to multi-query attention used in Llama 2, which uses 8x less GPU memory for attention caches. Now you can process longer sequences using a Petals client and have dialogues of up to 8192 tokens at https://chat.petals.dev
🐍 Python 3.11 support. Petals clients and servers now work on Python 3.11.
🐞 Bug fixes. We fixed the server's --token argument (used to provide your 🤗 Model Hub access token for loading Llama 2), possible deadlocks in the server, issues with fine-tuning speed (servers available via relays are deprioritized) and other minor load balancing issues.
🪟 Running server on Windows. We made a better guide for running a server in WSL (Windows Subsystem for Linux).
📦 Running server on Runpod. We added a guide for using a Petals template on Runpod.
What's Changed
- Update to petals.dev by @justheuristic in #390
- Bump version to 2.0.0.post3 by @borzunov in #391
- Fix --attn_cache_tokens default by @borzunov in #392
- Fix deadlocks in MemoryCache by @borzunov in #396
- Support Python 3.11 by @borzunov in #393
- Fix routing through relay, default network RPS, --token, logging, readme by @borzunov in #399
- If speedtest fails, assume network speed of 100 Mbit/s by @borzunov in #404
- Split long sequences into chunks by @justheuristic in #403
- Add Llama 2, WSL instructions to readme by @borzunov in #406
- Update README.md by @borzunov in #407
- Update commands for hosting Llama 2 in readme by @borzunov in #409
- Update --update_period and --expiration defaults by @borzunov in #410
- Bump version to 2.0.1 by @borzunov in #411
Full Changelog: v2.0.0.post1...v2.0.1