Launch80's picture
Revert main to v1.1.1 (v2.0 not released)
72ac3ed verified
Raw History Blame Contribute Delete
13.6 kB
Invalid JSON:Unexpected token 'N', ..."tha_err": NaN, "r"... is not valid JSON
{
"name": "Qwopus3.8-27B-Coder-Flash-Int4-baked",
"built_on": "Jackrong/Qwopus3.8-27B-Coder-Flash <- unsloth/Qwen3.8-27B",
"files": {
"model-00001-of-00005.safetensors": {
"linked": true,
"sha256": "d1d3584edf4e9eb9a9d432a8f64fb3f440b79c10f4b6c4567f59ea0597831d40"
},
"model-00002-of-00005.safetensors": {
"linked": false,
"dropped": [
"lm_head.weight"
],
"sha256": "8433039f57718eb40c8fb44bd0068315116cb39ff3f23594d781a8569594a05f"
},
"model-00003-of-00005.safetensors": {
"linked": true,
"sha256": "e89708d1af04af3489eac92c6b557e2dca396fbe0fd14e4eea77d9cfccfc63cd"
},
"model-00004-of-00005.safetensors": {
"linked": true,
"sha256": "5b1b3545e3c26f2e7ad3ee36babcf592685a8a487036338c7cb0caaa3909ed70"
},
"model-00005-of-00005.safetensors": {
"linked": false,
"dropped": [
"mtp.fc.weight",
"mtp.layers.0.mlp.down_proj.weight",
"mtp.layers.0.mlp.gate_proj.weight",
"mtp.layers.0.mlp.up_proj.weight",
"mtp.layers.0.self_attn.k_proj.weight",
"mtp.layers.0.self_attn.o_proj.weight",
"mtp.layers.0.self_attn.q_proj.weight",
"mtp.layers.0.self_attn.v_proj.weight"
],
"sha256": "06b3ffa31739af637f3ce5ceffd70a85ee1646aa9d71372b1335e513417a3a90"
},
"model-bake-int4.safetensors": {
"sha256": "45b3db19c2e6dcacf226568dbee4abc6b834def6cb34168d611ea01782a9df52"
}
},
"replaced": "lm_head + 8 MTP linears -> model-bake-int4.safetensors (GPTQ INT4 sym g128, desc_act false)",
"body": "400 GPTQ INT4 body linears byte-identical to the source; MTP norms, embedding, vision, in_proj_a/b bf16 unchanged",
"grid": {
"lm_head.weight": {
"K": 5120,
"N": 248320,
"eval_rows": 4096,
"ref_label": "none",
"settings": {
"wikitext@0.005": {
"out_err": 0.0605425126850605,
"weight_err": 0.15270298719406128
},
"wikitext@0.01": {
"out_err": 0.06027491018176079,
"weight_err": 0.1513257771730423
},
"wikitext@0.02": {
"out_err": 0.05984831973910332,
"weight_err": 0.14906899631023407
},
"wikitext+chat@0.005": {
"out_err": 0.049201108515262604,
"weight_err": 0.14964386820793152
},
"wikitext+chat@0.01": {
"out_err": 0.04915818199515343,
"weight_err": 0.14881111681461334
},
"wikitext+chat@0.02": {
"out_err": 0.04909635707736015,
"weight_err": 0.14735952019691467
},
"wikitext+chat+code@0.005": {
"out_err": 0.04684174433350563,
"weight_err": 0.14849957823753357
},
"wikitext+chat+code@0.01": {
"out_err": 0.046834904700517654,
"weight_err": 0.1477881222963333
},
"wikitext+chat+code@0.02": {
"out_err": 0.046837806701660156,
"weight_err": 0.14652249217033386
}
},
"davetha_err": NaN,
"rtn_err": 0.0817551389336586,
"chosen": {
"mix": "wikitext+chat+code",
"damp": 0.01,
"out_err": 0.046834904700517654,
"beats_davetha": false
}
},
"mtp.fc.weight": {
"K": 10240,
"N": 5120,
"eval_rows": 4096,
"ref_label": "none",
"settings": {
"wikitext@0.005": {
"out_err": 0.09442650526762009,
"weight_err": 0.16172412037849426
},
"wikitext@0.01": {
"out_err": 0.0936480164527893,
"weight_err": 0.15784817934036255
},
"wikitext@0.02": {
"out_err": 0.09276454895734787,
"weight_err": 0.15361575782299042
},
"wikitext+chat@0.005": {
"out_err": 0.07767980545759201,
"weight_err": 0.15656748414039612
},
"wikitext+chat@0.01": {
"out_err": 0.07755029946565628,
"weight_err": 0.15459588170051575
},
"wikitext+chat@0.02": {
"out_err": 0.07736190408468246,
"weight_err": 0.15188030898571014
},
"wikitext+chat+code@0.005": {
"out_err": 0.06983596086502075,
"weight_err": 0.15449532866477966
},
"wikitext+chat+code@0.01": {
"out_err": 0.06985525041818619,
"weight_err": 0.15293936431407928
},
"wikitext+chat+code@0.02": {
"out_err": 0.06988242268562317,
"weight_err": 0.15060630440711975
}
},
"davetha_err": NaN,
"rtn_err": 0.11469322443008423,
"chosen": {
"mix": "wikitext+chat+code",
"damp": 0.005,
"out_err": 0.06983596086502075,
"beats_davetha": false
}
},
"mtp.layers.0.self_attn.q_proj.weight": {
"K": 5120,
"N": 12288,
"eval_rows": 4096,
"ref_label": "none",
"settings": {
"wikitext@0.005": {
"out_err": 0.03477000817656517,
"weight_err": 0.157014861702919
},
"wikitext@0.01": {
"out_err": 0.034655895084142685,
"weight_err": 0.15460990369319916
},
"wikitext@0.02": {
"out_err": 0.03443285450339317,
"weight_err": 0.151133731007576
},
"wikitext+chat@0.005": {
"out_err": 0.02935590222477913,
"weight_err": 0.15426351130008698
},
"wikitext+chat@0.01": {
"out_err": 0.029292069375514984,
"weight_err": 0.15263773500919342
},
"wikitext+chat@0.02": {
"out_err": 0.02923501282930374,
"weight_err": 0.15006853640079498
},
"wikitext+chat+code@0.005": {
"out_err": 0.026783455163240433,
"weight_err": 0.15280482172966003
},
"wikitext+chat+code@0.01": {
"out_err": 0.026779530569911003,
"weight_err": 0.15140675008296967
},
"wikitext+chat+code@0.02": {
"out_err": 0.026781443506479263,
"weight_err": 0.14904899895191193
}
},
"davetha_err": NaN,
"rtn_err": 0.04437090829014778,
"chosen": {
"mix": "wikitext+chat+code",
"damp": 0.01,
"out_err": 0.026779530569911003,
"beats_davetha": false
}
},
"mtp.layers.0.self_attn.k_proj.weight": {
"K": 5120,
"N": 1024,
"eval_rows": 4096,
"ref_label": "none",
"settings": {
"wikitext@0.005": {
"out_err": 0.0788261666893959,
"weight_err": 0.1772148609161377
},
"wikitext@0.01": {
"out_err": 0.07868398725986481,
"weight_err": 0.1746716946363449
},
"wikitext@0.02": {
"out_err": 0.07811010628938675,
"weight_err": 0.17079730331897736
},
"wikitext+chat@0.005": {
"out_err": 0.06639601290225983,
"weight_err": 0.17419373989105225
},
"wikitext+chat@0.01": {
"out_err": 0.06634438037872314,
"weight_err": 0.17248499393463135
},
"wikitext+chat@0.02": {
"out_err": 0.06614624708890915,
"weight_err": 0.1696658730506897
},
"wikitext+chat+code@0.005": {
"out_err": 0.06056930869817734,
"weight_err": 0.1726793646812439
},
"wikitext+chat+code@0.01": {
"out_err": 0.06053243204951286,
"weight_err": 0.17100092768669128
},
"wikitext+chat+code@0.02": {
"out_err": 0.06060367450118065,
"weight_err": 0.1684817671775818
}
},
"davetha_err": NaN,
"rtn_err": 0.10084359347820282,
"chosen": {
"mix": "wikitext+chat+code",
"damp": 0.01,
"out_err": 0.06053243204951286,
"beats_davetha": false
}
},
"mtp.layers.0.self_attn.v_proj.weight": {
"K": 5120,
"N": 1024,
"eval_rows": 4096,
"ref_label": "none",
"settings": {
"wikitext@0.005": {
"out_err": 0.06692653149366379,
"weight_err": 0.16015149652957916
},
"wikitext@0.01": {
"out_err": 0.06651435792446136,
"weight_err": 0.1578548103570938
},
"wikitext@0.02": {
"out_err": 0.06576719880104065,
"weight_err": 0.15426428616046906
},
"wikitext+chat@0.005": {
"out_err": 0.0561215840280056,
"weight_err": 0.1574513167142868
},
"wikitext+chat@0.01": {
"out_err": 0.056023649871349335,
"weight_err": 0.15586549043655396
},
"wikitext+chat@0.02": {
"out_err": 0.05587271228432655,
"weight_err": 0.15320546925067902
},
"wikitext+chat+code@0.005": {
"out_err": 0.05113956704735756,
"weight_err": 0.1558873951435089
},
"wikitext+chat+code@0.01": {
"out_err": 0.05115906894207001,
"weight_err": 0.15450233221054077
},
"wikitext+chat+code@0.02": {
"out_err": 0.051164932548999786,
"weight_err": 0.15216301381587982
}
},
"davetha_err": NaN,
"rtn_err": 0.08277183771133423,
"chosen": {
"mix": "wikitext+chat+code",
"damp": 0.005,
"out_err": 0.05113956704735756,
"beats_davetha": false
}
},
"mtp.layers.0.self_attn.o_proj.weight": {
"K": 6144,
"N": 5120,
"eval_rows": 4096,
"ref_label": "none",
"settings": {
"wikitext@0.005": {
"out_err": 0.09032029658555984,
"weight_err": 0.17671002447605133
},
"wikitext@0.01": {
"out_err": 0.08870720863342285,
"weight_err": 0.1694953888654709
},
"wikitext@0.02": {
"out_err": 0.08771047741174698,
"weight_err": 0.16327469050884247
},
"wikitext+chat@0.005": {
"out_err": 0.07257946580648422,
"weight_err": 0.17309002578258514
},
"wikitext+chat@0.01": {
"out_err": 0.07263592630624771,
"weight_err": 0.16735611855983734
},
"wikitext+chat@0.02": {
"out_err": 0.07243864983320236,
"weight_err": 0.16225409507751465
},
"wikitext+chat+code@0.005": {
"out_err": 0.06669317185878754,
"weight_err": 0.17187811434268951
},
"wikitext+chat+code@0.01": {
"out_err": 0.06670280545949936,
"weight_err": 0.16667558252811432
},
"wikitext+chat+code@0.02": {
"out_err": 0.06682953983545303,
"weight_err": 0.16194599866867065
}
},
"davetha_err": NaN,
"rtn_err": 0.13036604225635529,
"chosen": {
"mix": "wikitext+chat+code",
"damp": 0.005,
"out_err": 0.06669317185878754,
"beats_davetha": false
}
},
"mtp.layers.0.mlp.gate_proj.weight": {
"K": 5120,
"N": 17408,
"eval_rows": 4096,
"ref_label": "none",
"settings": {
"wikitext@0.005": {
"out_err": 0.04800385981798172,
"weight_err": 0.14573533833026886
},
"wikitext@0.01": {
"out_err": 0.047810908406972885,
"weight_err": 0.14436183869838715
},
"wikitext@0.02": {
"out_err": 0.047500718384981155,
"weight_err": 0.1421309858560562
},
"wikitext+chat@0.005": {
"out_err": 0.04013431817293167,
"weight_err": 0.14326751232147217
},
"wikitext+chat@0.01": {
"out_err": 0.04008609801530838,
"weight_err": 0.14236515760421753
},
"wikitext+chat@0.02": {
"out_err": 0.04001238942146301,
"weight_err": 0.1408526450395584
},
"wikitext+chat+code@0.005": {
"out_err": 0.03672541305422783,
"weight_err": 0.14236244559288025
},
"wikitext+chat+code@0.01": {
"out_err": 0.03672228753566742,
"weight_err": 0.14158716797828674
},
"wikitext+chat+code@0.02": {
"out_err": 0.036723554134368896,
"weight_err": 0.1401895135641098
}
},
"davetha_err": NaN,
"rtn_err": 0.06349170207977295,
"chosen": {
"mix": "wikitext+chat+code",
"damp": 0.01,
"out_err": 0.03672228753566742,
"beats_davetha": false
}
},
"mtp.layers.0.mlp.up_proj.weight": {
"K": 5120,
"N": 17408,
"eval_rows": 4096,
"ref_label": "none",
"settings": {
"wikitext@0.005": {
"out_err": 0.07442658394575119,
"weight_err": 0.14497707784175873
},
"wikitext@0.01": {
"out_err": 0.0740872323513031,
"weight_err": 0.14358636736869812
},
"wikitext@0.02": {
"out_err": 0.07367879152297974,
"weight_err": 0.14138270914554596
},
"wikitext+chat@0.005": {
"out_err": 0.06228446960449219,
"weight_err": 0.1424804925918579
},
"wikitext+chat@0.01": {
"out_err": 0.062201157212257385,
"weight_err": 0.1416160613298416
},
"wikitext+chat@0.02": {
"out_err": 0.06211536005139351,
"weight_err": 0.14011147618293762
},
"wikitext+chat+code@0.005": {
"out_err": 0.05698351934552193,
"weight_err": 0.1416272670030594
},
"wikitext+chat+code@0.01": {
"out_err": 0.05698757246136665,
"weight_err": 0.1408454179763794
},
"wikitext+chat+code@0.02": {
"out_err": 0.056989941745996475,
"weight_err": 0.1394772231578827
}
},
"davetha_err": NaN,
"rtn_err": 0.09559829533100128,
"chosen": {
"mix": "wikitext+chat+code",
"damp": 0.005,
"out_err": 0.05698351934552193,
"beats_davetha": false
}
},
"mtp.layers.0.mlp.down_proj.weight": {
"K": 17408,
"N": 5120,
"eval_rows": 4096,
"ref_label": "none",
"settings": {
"wikitext@0.005": {
"out_err": 0.10945171862840652,
"weight_err": 0.1984400600194931
},
"wikitext@0.01": {
"out_err": 0.10685558617115021,
"weight_err": 0.18855606019496918
},
"wikitext@0.02": {
"out_err": 0.10374335199594498,
"weight_err": 0.17687879502773285
},
"wikitext+chat@0.005": {
"out_err": 0.08473674207925797,
"weight_err": 0.17660214006900787
},
"wikitext+chat@0.01": {
"out_err": 0.0844150260090828,
"weight_err": 0.17259404063224792
},
"wikitext+chat@0.02": {
"out_err": 0.08395801484584808,
"weight_err": 0.16705484688282013
},
"wikitext+chat+code@0.005": {
"out_err": 0.07583369314670563,
"weight_err": 0.1744048148393631
},
"wikitext+chat+code@0.01": {
"out_err": 0.0757918581366539,
"weight_err": 0.17089152336120605
},
"wikitext+chat+code@0.02": {
"out_err": 0.07572303712368011,
"weight_err": 0.16594621539115906
}
},
"davetha_err": NaN,
"rtn_err": 0.11964161694049835,
"chosen": {
"mix": "wikitext+chat+code",
"damp": 0.02,
"out_err": 0.07572303712368011,
"beats_davetha": false
}
}
}
}