make 30b 3a model variant or using gpt 20b model
#1
by
Narutoouz - opened
thanks for making this rust specialized quanting model.
We are planning a new iteration of the model trained on a larger dataset (will publish that too) generated with newer models on our network.
Besides better Rust skills, we are focusing on better instruction following, as the current model is more suited for code completion than full instruction-following coding tasks. We've been evaluating several base models and recent Qwen models are strong candidates based on our Rust benchmarks, though we're still testing.