{"schemaVersion":"hugging-bay.fit-check.v1","generatedAt":"2026-07-10T02:02:51.663Z","artifactId":"hf-model-rinna-japanese-gpt-neox-3-6b-instruction-ppo","repo":"rinna/japanese-gpt-neox-3.6b-instruction-ppo","webUrl":"https://huggingbay.xyz/artifact/hf-model-rinna-japanese-gpt-neox-3-6b-instruction-ppo","name":"japanese-gpt-neox-3.6b-instruction-ppo","paramsB":3.6,"activeB":3.6,"quant":"f16","weightFile":null,"fileSizeBytes":1,"ctx":8192,"batch":1,"estimate":{"weightsGb":0,"kvGb":0.94,"overheadGb":0.8,"totalGb":1.74},"gpus":[{"id":"t4-16","label":"NVIDIA T4 16GB (free Colab)","vramGb":16,"unified":false,"verdict":"fits","headroomGb":14.26,"totalGb":1.74},{"id":"l4-24","label":"NVIDIA L4 24GB","vramGb":24,"unified":false,"verdict":"fits","headroomGb":22.26,"totalGb":1.74},{"id":"rtx3060-12","label":"RTX 3060 12GB","vramGb":12,"unified":false,"verdict":"fits","headroomGb":10.26,"totalGb":1.74},{"id":"rtx4080-16","label":"RTX 4080 16GB","vramGb":16,"unified":false,"verdict":"fits","headroomGb":14.26,"totalGb":1.74},{"id":"rtx3090-24","label":"RTX 3090 24GB","vramGb":24,"unified":false,"verdict":"fits","headroomGb":22.26,"totalGb":1.74},{"id":"rtx4090-24","label":"RTX 4090 24GB","vramGb":24,"unified":false,"verdict":"fits","headroomGb":22.26,"totalGb":1.74},{"id":"a100-40","label":"A100 40GB","vramGb":40,"unified":false,"verdict":"fits","headroomGb":38.26,"totalGb":1.74},{"id":"a100-80","label":"A100 80GB","vramGb":80,"unified":false,"verdict":"fits","headroomGb":78.26,"totalGb":1.74},{"id":"h100-80","label":"H100 80GB","vramGb":80,"unified":false,"verdict":"fits","headroomGb":78.26,"totalGb":1.74},{"id":"apple-m-16","label":"Apple M-series 16GB (unified)","vramGb":16,"unified":true,"verdict":"fits","headroomGb":14.26,"totalGb":1.74},{"id":"apple-m-32","label":"Apple M-series 32GB (unified)","vramGb":32,"unified":true,"verdict":"fits","headroomGb":30.26,"totalGb":1.74},{"id":"apple-m-64","label":"Apple M-series 64GB (unified)","vramGb":64,"unified":true,"verdict":"fits","headroomGb":62.26,"totalGb":1.74},{"id":"apple-m-128","label":"Apple M-series 128GB (unified)","vramGb":128,"unified":true,"verdict":"fits","headroomGb":126.26,"totalGb":1.74},{"id":"cpu-ram-32","label":"CPU / 32GB system RAM","vramGb":32,"unified":true,"verdict":"fits","headroomGb":30.26,"totalGb":1.74},{"id":"cpu-ram-64","label":"CPU / 64GB system RAM","vramGb":64,"unified":true,"verdict":"fits","headroomGb":62.26,"totalGb":1.74}],"target":{"id":"rtx3090-24","label":"RTX 3090 24GB","vramGb":24,"maxCtx":202240},"whatIf":[{"quant":"q2_k","bytesPerParam":0.33,"weightsGb":1.19,"totalGb":2.93,"verdict":"fits","headroomGb":21.07},{"quant":"q3_k_m","bytesPerParam":0.44,"weightsGb":1.58,"totalGb":3.32,"verdict":"fits","headroomGb":20.68},{"quant":"q4_k_m","bytesPerParam":0.57,"weightsGb":2.05,"totalGb":3.79,"verdict":"fits","headroomGb":20.21},{"quant":"q5_k_m","bytesPerParam":0.69,"weightsGb":2.48,"totalGb":4.22,"verdict":"fits","headroomGb":19.78},{"quant":"q6_k","bytesPerParam":0.82,"weightsGb":2.95,"totalGb":4.69,"verdict":"fits","headroomGb":19.31},{"quant":"q8_0","bytesPerParam":1.06,"weightsGb":3.82,"totalGb":5.56,"verdict":"fits","headroomGb":18.44},{"quant":"f16","bytesPerParam":2,"weightsGb":7.2,"totalGb":8.94,"verdict":"fits","headroomGb":15.06}],"guidance":["verdict per GPU: fits (<=90% VRAM) / tight (<=100%) / no.","If it does not fit at this precision, check /api/artifacts/{id}/quantizations or the per-quant rows for a smaller mirror.","Estimates: weights + fp16 KV-cache(ctx,batch) + overhead. Real usage varies by runtime."]}