{ "quantization": "Q8_0", "output_tensor_type": "Q8_0", "tensors": { "Q8_0": 170, "F32": 49 }, "metadata": { "llama.context_length": 131072, "llama.attention.key_length": 128, "llama.attention.value_length": 128, "llama.rope.freq_base": 5000000.0 }, "llama_commit": "f849095365d5ac1f560886a4e6574531d559e96f", "packages": { "torch": "2.11.0+cpu", "transformers": "4.57.6", "numpy": "2.2.6", "safetensors": "0.8.0", "sentencepiece": "0.2.2" }, "file": { "bytes": 1153529088, "sha256": "0cb29dcf5b84a45b0733cd50ac4ad1b879a4aafaa01f6d190c17fc3d9a10961f" }, "base_model": "Andrew0425/AgenticASR-Refiner", "source_safetensors_sha256": "1a9635c8f288648c5ec3c9d8572caaa53295242c90f94f1655e07ef2b31d907d", "runtime": { "parallel": 8, "context_per_slot": 512, "context_total": 4096, "max_new_tokens": 256, "kv": "F16", "kv_unified": true, "flash_attention": true, "cann": "9.1.1" }, "validation": { "fp32_text_matches": 64, "cases": 64, "concurrent_128_character_requests": 8, "concurrent_matches_serial": 8, "sse_matches_nonstreaming": true, "sampled_npu_process_peak_mib": 1267, "isolated_benchmark": false } }