clef-NVFP4 / clef_quant_recipe.json
aleksblago's picture
Add clef-NVFP4
0df6418 verified
Raw History Blame Contribute Delete
1.15 kB
{
"scheme": "NVFP4",
"format": "nvfp4",
"release": "/hf/hub/models--Cloudflare--clef/snapshots/0b331204bb13fbd2ca93a64df1956f5b55478ce5",
"ignore": [
"lm_head",
"re:.*visual.*",
"re:.*linear_attn.in_proj_a$",
"re:.*linear_attn.in_proj_b$"
],
"extra_fused_layer_names": [
[
"in_proj_qkv",
"in_proj_z"
]
],
"calibration": {
"used": true,
"path": "/bench/data/calibration.jsonl",
"is_dev_fallback": false,
"sha256": "878fcb528338a6dc595ca85b050346ef203d2f7e3707fe9a1d07d8a2f36333bd",
"samples": 512,
"tokens_total": 379073,
"tokens_min": 177,
"tokens_max": 3112,
"encoder": "Cloudflare joint_schema_model.encode_record (release code)"
},
"head_sha256": {
"joint_head.safetensors": "a010ac04f078e699988e4049cbea5e62c962393f59fec366640b64e8d69a4953",
"joint_head_config.json": "890be585d75b981201eb96a35f98a9967afe37bc8d220cfdea2e72d56507534f"
},
"versions": {
"llmcompressor": "0.14.0",
"compressed_tensors": "0.19.0",
"transformers": "5.16.1",
"torch": "2.14.0a0+b2c75dd062.nv26.09"
},
"pipeline": "sequential",
"seconds": 2255.5
}