AtomicChat Releases Refusal-Reduced LFM3-1.7B Variant

AtomicChat has published a modified, quantized version of Liquid AI’s LFM3-1.7B that it says reduces refusals on harmful prompts from 61.7% to zero. The two listed GGUF files are 1.1 GB and 1.6 GB, and the publisher reports no drop on its MMLU test.