Version 1.0.1
This release contains may small and some large improvements:
- sliced Llama, Phi and Phi3 evaluation, unlocking virtually unlimited model size
- 30% speedup of model evaluation
- period re-evaluation of all models
- sync model/metagraph rewrite
- subprocess logging improvement
- improved error handling and reporting
- reduction/prevention of excessive errors/logging
- logging of more statistics
- even more runtime configurable parameters
- improved competitions.json format
- fixes for tokenizer hacks
- allow for a limited number of tokenization issues (true fix under investigation)
- various clarifications and simplifications