WizardLM/WizardCoder-33B-V1.1 released!

Based off of deepseek coder, the current SOTA 33B model, allegedly has gpt 3.5 levels of performance, will be excited to test once I've made exllamav2 quants and will try to update with my findings as a copilot model