Hugging Face's logo
Activity Feed

AI & ML interests

Text Generation & Chat Assistants; Model Compression & Quantization (Q4/Q6/Q8, gs32); Inference & Serving (on-prem, low-latency); RAG / Retrieval; Agents & Tool Use; Distillation / LoRA / Fine-tuning
sebastavar 
updated a Space 5 months ago
sebastavar 
updated 3 models 5 months ago
sebastavar 
published 3 models 5 months ago
sebastavar 
updated 2 models 5 months ago
sebastavar 
published a model 5 months ago
sebastavar 
updated a model 5 months ago
sebastavar 
published a model 5 months ago
sebastavar 
updated a model 5 months ago
sebastavar 
published a model 5 months ago
sebastavar 
in halley-ai/gpt-oss-20b-MLX-4bit-gs32 6 months ago

UPDATE (Aug 22, 2025) — Prefer the 5-bit build -- gpt-oss-20B — MLX 4-bit (gs=32) for Apple Silicon
#1 opened 6 months ago by
sebastavar
sebastavar 
in halley-ai/README 6 months ago

GPT-OSS-20B on a 16 GB Mac (MLX): Why Q3 Quantization Didn't Work, and What I Recommend
#1 opened 6 months ago by
sebastavar
sebastavar 
in halley-ai/gpt-oss-20b-MLX-5bit-gs32 6 months ago

GPT-OSS-20B · MLX Q5 (gs=32) — practical on 24–32 GB Macs, much closer to Q8 than Q4
#1 opened 6 months ago by
sebastavar
sebastavar 
updated 2 models 6 months ago
sebastavar 
published a model 6 months ago