Consolidate complete Molmo2 INT8 into molmo2_8b_int8_convrot.safetensors and remove all eight weight shards
8d9eae4 verified Download convrot_config.json from Alissonerdx/Molmo2-8B-Minimax-H3: direct link, hf CLI and curl.
- Browser
- Download file 50.5 kB
-
https://huggingface.co/Alissonerdx/Molmo2-8B-Minimax-H3/resolve/main/convrot_config.json
- Command line
-
hf download hf://Alissonerdx/Molmo2-8B-Minimax-H3/convrot_config.json
-
curl -L -o convrot_config.json https://huggingface.co/Alissonerdx/Molmo2-8B-Minimax-H3/resolve/main/convrot_config.json
50.5 kB
| { | |
| "format": "molmo2-convrot-int8-v1", | |
| "modules": { | |
| "model.transformer.blocks.0.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.0.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.0.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.0.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.1.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.1.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.1.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.1.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.2.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.2.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.2.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.2.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.3.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.3.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.3.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.3.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.4.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.4.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.4.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.4.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.5.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.5.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.5.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.5.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.6.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.6.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.6.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.6.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.7.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.7.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.7.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.7.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.8.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.8.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.8.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.8.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.9.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.9.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.9.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.9.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.10.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.10.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.10.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.10.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.11.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.11.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.11.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.11.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.12.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.12.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.12.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.12.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.13.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.13.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.13.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.13.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.14.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.14.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.14.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.14.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.15.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.15.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.15.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.15.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.16.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.16.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.16.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.16.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.17.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.17.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.17.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.17.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.18.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.18.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.18.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.18.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.19.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.19.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.19.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.19.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.20.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.20.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.20.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.20.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.21.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.21.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.21.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.21.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.22.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.22.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.22.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.22.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.23.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.23.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.23.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.23.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.24.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.24.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.24.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.24.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.25.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.25.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.25.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.25.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.26.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.26.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.26.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.26.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.27.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.27.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.27.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.27.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.28.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.28.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.28.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.28.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.29.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.29.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.29.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.29.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.30.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.30.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.30.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.30.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.31.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.31.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.31.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.31.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.32.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.32.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.32.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.32.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.33.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.33.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.33.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.33.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.34.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.34.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.34.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.34.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.35.self_attn.att_proj": { | |
| "in_features": 4096, | |
| "out_features": 6144, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.35.self_attn.attn_out": { | |
| "in_features": 4096, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.35.mlp.ff_proj": { | |
| "in_features": 4096, | |
| "out_features": 24576, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.transformer.blocks.35.mlp.ff_out": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.0.attention.wq": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.0.attention.wk": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.0.attention.wv": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.0.attention.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.0.feed_forward.w1": { | |
| "in_features": 1152, | |
| "out_features": 4304, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.0.feed_forward.w2": { | |
| "in_features": 4304, | |
| "out_features": 1152, | |
| "block": 16, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.1.attention.wq": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.1.attention.wk": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.1.attention.wv": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.1.attention.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.1.feed_forward.w1": { | |
| "in_features": 1152, | |
| "out_features": 4304, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.1.feed_forward.w2": { | |
| "in_features": 4304, | |
| "out_features": 1152, | |
| "block": 16, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.2.attention.wq": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.2.attention.wk": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.2.attention.wv": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.2.attention.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.2.feed_forward.w1": { | |
| "in_features": 1152, | |
| "out_features": 4304, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.2.feed_forward.w2": { | |
| "in_features": 4304, | |
| "out_features": 1152, | |
| "block": 16, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.3.attention.wq": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.3.attention.wk": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.3.attention.wv": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.3.attention.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.3.feed_forward.w1": { | |
| "in_features": 1152, | |
| "out_features": 4304, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.3.feed_forward.w2": { | |
| "in_features": 4304, | |
| "out_features": 1152, | |
| "block": 16, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.4.attention.wq": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.4.attention.wk": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.4.attention.wv": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.4.attention.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.4.feed_forward.w1": { | |
| "in_features": 1152, | |
| "out_features": 4304, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.4.feed_forward.w2": { | |
| "in_features": 4304, | |
| "out_features": 1152, | |
| "block": 16, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.5.attention.wq": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.5.attention.wk": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.5.attention.wv": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.5.attention.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.5.feed_forward.w1": { | |
| "in_features": 1152, | |
| "out_features": 4304, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.5.feed_forward.w2": { | |
| "in_features": 4304, | |
| "out_features": 1152, | |
| "block": 16, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.6.attention.wq": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.6.attention.wk": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.6.attention.wv": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.6.attention.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.6.feed_forward.w1": { | |
| "in_features": 1152, | |
| "out_features": 4304, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.6.feed_forward.w2": { | |
| "in_features": 4304, | |
| "out_features": 1152, | |
| "block": 16, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.7.attention.wq": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.7.attention.wk": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.7.attention.wv": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.7.attention.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.7.feed_forward.w1": { | |
| "in_features": 1152, | |
| "out_features": 4304, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.7.feed_forward.w2": { | |
| "in_features": 4304, | |
| "out_features": 1152, | |
| "block": 16, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.8.attention.wq": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.8.attention.wk": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.8.attention.wv": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.8.attention.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.8.feed_forward.w1": { | |
| "in_features": 1152, | |
| "out_features": 4304, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.8.feed_forward.w2": { | |
| "in_features": 4304, | |
| "out_features": 1152, | |
| "block": 16, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.9.attention.wq": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.9.attention.wk": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.9.attention.wv": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.9.attention.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.9.feed_forward.w1": { | |
| "in_features": 1152, | |
| "out_features": 4304, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.9.feed_forward.w2": { | |
| "in_features": 4304, | |
| "out_features": 1152, | |
| "block": 16, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.10.attention.wq": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.10.attention.wk": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.10.attention.wv": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.10.attention.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.10.feed_forward.w1": { | |
| "in_features": 1152, | |
| "out_features": 4304, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.10.feed_forward.w2": { | |
| "in_features": 4304, | |
| "out_features": 1152, | |
| "block": 16, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.11.attention.wq": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.11.attention.wk": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.11.attention.wv": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.11.attention.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.11.feed_forward.w1": { | |
| "in_features": 1152, | |
| "out_features": 4304, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.11.feed_forward.w2": { | |
| "in_features": 4304, | |
| "out_features": 1152, | |
| "block": 16, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.12.attention.wq": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.12.attention.wk": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.12.attention.wv": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.12.attention.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.12.feed_forward.w1": { | |
| "in_features": 1152, | |
| "out_features": 4304, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.12.feed_forward.w2": { | |
| "in_features": 4304, | |
| "out_features": 1152, | |
| "block": 16, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.13.attention.wq": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.13.attention.wk": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.13.attention.wv": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.13.attention.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.13.feed_forward.w1": { | |
| "in_features": 1152, | |
| "out_features": 4304, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.13.feed_forward.w2": { | |
| "in_features": 4304, | |
| "out_features": 1152, | |
| "block": 16, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.14.attention.wq": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.14.attention.wk": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.14.attention.wv": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.14.attention.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.14.feed_forward.w1": { | |
| "in_features": 1152, | |
| "out_features": 4304, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.14.feed_forward.w2": { | |
| "in_features": 4304, | |
| "out_features": 1152, | |
| "block": 16, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.15.attention.wq": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.15.attention.wk": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.15.attention.wv": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.15.attention.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.15.feed_forward.w1": { | |
| "in_features": 1152, | |
| "out_features": 4304, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.15.feed_forward.w2": { | |
| "in_features": 4304, | |
| "out_features": 1152, | |
| "block": 16, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.16.attention.wq": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.16.attention.wk": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.16.attention.wv": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.16.attention.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.16.feed_forward.w1": { | |
| "in_features": 1152, | |
| "out_features": 4304, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.16.feed_forward.w2": { | |
| "in_features": 4304, | |
| "out_features": 1152, | |
| "block": 16, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.17.attention.wq": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.17.attention.wk": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.17.attention.wv": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.17.attention.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.17.feed_forward.w1": { | |
| "in_features": 1152, | |
| "out_features": 4304, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.17.feed_forward.w2": { | |
| "in_features": 4304, | |
| "out_features": 1152, | |
| "block": 16, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.18.attention.wq": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.18.attention.wk": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.18.attention.wv": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.18.attention.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.18.feed_forward.w1": { | |
| "in_features": 1152, | |
| "out_features": 4304, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.18.feed_forward.w2": { | |
| "in_features": 4304, | |
| "out_features": 1152, | |
| "block": 16, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.19.attention.wq": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.19.attention.wk": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.19.attention.wv": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.19.attention.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.19.feed_forward.w1": { | |
| "in_features": 1152, | |
| "out_features": 4304, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.19.feed_forward.w2": { | |
| "in_features": 4304, | |
| "out_features": 1152, | |
| "block": 16, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.20.attention.wq": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.20.attention.wk": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.20.attention.wv": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.20.attention.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.20.feed_forward.w1": { | |
| "in_features": 1152, | |
| "out_features": 4304, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.20.feed_forward.w2": { | |
| "in_features": 4304, | |
| "out_features": 1152, | |
| "block": 16, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.21.attention.wq": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.21.attention.wk": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.21.attention.wv": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.21.attention.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.21.feed_forward.w1": { | |
| "in_features": 1152, | |
| "out_features": 4304, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.21.feed_forward.w2": { | |
| "in_features": 4304, | |
| "out_features": 1152, | |
| "block": 16, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.22.attention.wq": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.22.attention.wk": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.22.attention.wv": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.22.attention.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.22.feed_forward.w1": { | |
| "in_features": 1152, | |
| "out_features": 4304, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.22.feed_forward.w2": { | |
| "in_features": 4304, | |
| "out_features": 1152, | |
| "block": 16, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.23.attention.wq": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.23.attention.wk": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.23.attention.wv": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.23.attention.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.23.feed_forward.w1": { | |
| "in_features": 1152, | |
| "out_features": 4304, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.23.feed_forward.w2": { | |
| "in_features": 4304, | |
| "out_features": 1152, | |
| "block": 16, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.24.attention.wq": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.24.attention.wk": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.24.attention.wv": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.24.attention.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.24.feed_forward.w1": { | |
| "in_features": 1152, | |
| "out_features": 4304, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_vit.transformer.resblocks.24.feed_forward.w2": { | |
| "in_features": 4304, | |
| "out_features": 1152, | |
| "block": 16, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_pooling_2d.wq": { | |
| "in_features": 2304, | |
| "out_features": 1152, | |
| "block": 256, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_pooling_2d.wk": { | |
| "in_features": 2304, | |
| "out_features": 1152, | |
| "block": 256, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_pooling_2d.wv": { | |
| "in_features": 2304, | |
| "out_features": 1152, | |
| "block": 256, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_pooling_2d.wo": { | |
| "in_features": 1152, | |
| "out_features": 1152, | |
| "block": 64, | |
| "bias": true | |
| }, | |
| "model.vision_backbone.image_projector.w1": { | |
| "in_features": 1152, | |
| "out_features": 12288, | |
| "block": 64, | |
| "bias": false | |
| }, | |
| "model.vision_backbone.image_projector.w2": { | |
| "in_features": 12288, | |
| "out_features": 4096, | |
| "block": 256, | |
| "bias": false | |
| }, | |
| "model.vision_backbone.image_projector.w3": { | |
| "in_features": 1152, | |
| "out_features": 12288, | |
| "block": 64, | |
| "bias": false | |
| }, | |
| "lm_head": { | |
| "in_features": 4096, | |
| "out_features": 151936, | |
| "block": 256, | |
| "bias": false | |
| } | |
| }, | |
| "shards": [ | |
| "molmo2_8b_int8_convrot.safetensors" | |
| ] | |
| } |