{ "schema": "rivetcoder-fp8-quantization", "schema_version": 1, "created_at": "2026-08-27T23:01:09+09:00", "source": { "repo_id": "HCHs/RivetCoder-9B-A4B", "revision": "9a90b1917d9b5438e4d2fe1a4f6aea884db59a60" }, "target": { "repo_id": "HCHs/RivetCoder-9B-A4B-FP8", "local_directory": "RivetCoder-9B-A4B-FP8" }, "method": { "framework": "TorchAO", "config": "Float8DynamicActivationFloat8WeightConfig", "weight_dtype": "float8_e4m3fn", "activation_dtype": "float8_e4m3fn", "activation_scheme": "dynamic", "granularity": "per-tensor", "compatible_linear_modules_only": true }, "runtime": { "platform": "Windows 11", "python": "3.12.10", "torch": "2.12.0+cu130", "transformers": "5.16.1", "accelerate": "1.13.0", "safetensors": "0.8.0", "torchao": "0.15.0", "gpu": "NVIDIA GeForce RTX 5070 Ti", "compute_capability": [12, 0] }, "results": { "parameter_tensors": 1826, "fp8_tensor_subclass_parameters": 1636, "fp8_parameter_elements": 8475574272, "stored_tensor_bytes": 9000638976, "safetensors_shards": 5, "all_parameters_materialized": true, "all_parameters_on_cuda": true, "clean_reload_success": true, "clean_reload_seconds": 338.0, "clean_reload_fp8_parameters": 1636, "clean_reload_non_cuda_parameters": 0, "fast_serving_validated": true, "fast_serving_logits_bit_exact": true, "resident_cuda_allocated_bytes": 9039455232, "forward_finite": true, "forward_logits_shape": [1, 6, 128000] }, "compatibility": { "required_context": "torch.no_grad", "known_incompatible_context": "torch.inference_mode", "known_error": "Cannot set version_counter for inference tensor" } }