Text-to-Image
Diffusers
Safetensors
MageFlowPipeline
ajh
mage-flow
mage-flow-nvfp4-ajh
nvfp4
blackwell
qwen3-vl
quantization
Instructions to use ajh-code/Mage-Flow-NVFP4-AJH with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use ajh-code/Mage-Flow-NVFP4-AJH with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("ajh-code/Mage-Flow-NVFP4-AJH", torch_dtype=torch.bfloat16, device_map="cuda") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- Draw Things
- DiffusionBee
| { | |
| "format_version": 1, | |
| "artifact_kind": "mage_flow_transformer_mlp_nvfp4_resident_v1", | |
| "created_utc": "2026-07-24T07:00:11.747181+00:00", | |
| "container": { | |
| "format": "safetensors", | |
| "header_metadata": { | |
| "mage_nvfp4_contract": "mage_flow_transformer_mlp_nvfp4_resident_v1;format_version=1" | |
| }, | |
| "header_encoding": "single deterministic contract key; legacy two-key draft headers remain readable" | |
| }, | |
| "source": { | |
| "transformer_config_path": "BASE_MODEL/transformer/config.json", | |
| "transformer_checkpoint_path": "BASE_MODEL/transformer/diffusion_pytorch_model.safetensors", | |
| "transformer_config_sha256": "8493c3b2722738c2a824ac82b1fd9c89fefb4e354fc88363207193db7fe702de", | |
| "transformer_checkpoint_sha256": "04431abbc3acd1a5b86a7f77269f8e77e05a6cac3b21f35e1aef7493f6ab4934" | |
| }, | |
| "library_hashes": { | |
| "artifact_script_sha256": "459b4e8b7fb6b15b1f2eac3267082c52f1723efd65a59fe429066f565d224b81", | |
| "resident_source_sha256": "060697d4784be3e3461d38ef6ff24128f47a4322a05e5463f8687796b03bda9f", | |
| "resident_library_sha256": "d2f7d62d26b4ae3babb7791aee8e0d95ad4794ac9c1d22146c974784381ada48", | |
| "mage_flow_py_sha256": "59b6e1bee7f95a7fd2fa7bd9e765966832951e2bb184b5ae27283884997b845f", | |
| "mage_layers_py_sha256": "4b198343b8929f48a0a14d388502c81f54be17b222f6831303d2da6a91f33a62", | |
| "pipeline_py_sha256": "b9fc57018570372dd3404a733e19b918a359188dd9f5ef7817c6b30969fc13db" | |
| }, | |
| "environment": { | |
| "python_version": "3.11.14 (main, Oct 10 2025, 08:54:04) [GCC 13.3.0]", | |
| "torch_version": "2.13.0+cu130", | |
| "safetensors_version": "0.8.0" | |
| }, | |
| "model": { | |
| "depth": 12, | |
| "hidden_size": 3072, | |
| "num_heads": 24, | |
| "context_in_dim": 2560, | |
| "in_channels": 128, | |
| "out_channels": 128, | |
| "patch_size": 1 | |
| }, | |
| "quantization": { | |
| "format": "nvfp4_two_level", | |
| "block_elements": 16, | |
| "scale_tile_outer": 128, | |
| "scale_tile_inner": 4, | |
| "bias_policy": "artifact_bfloat16" | |
| }, | |
| "targets": [ | |
| { | |
| "module_key": "transformer_blocks.0.img_mlp.net.0.proj", | |
| "weight_key": "transformer_blocks.0.img_mlp.net.0.proj.weight", | |
| "bias_key": "transformer_blocks.0.img_mlp.net.0.proj.bias", | |
| "weight_shape": [ | |
| 12288, | |
| 3072 | |
| ], | |
| "bias_shape": [ | |
| 12288 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.0.img_mlp.net.0.proj.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.0.img_mlp.net.0.proj.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.0.img_mlp.net.0.proj.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.0.img_mlp.net.0.proj.bias_bf16", | |
| "weight_tensor_scale": 0.00024559383746236563, | |
| "weight_global_amax": 0.66015625, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 192, | |
| "outer_tiles": 96, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "d05d519a8eb21260cd6c7cdcc886ebd9e32986bd925f3b5a740f75c93ca39162", | |
| "source_bias_sha256": "e032c18ee63a7667d67e97e8611440b64e31bda7e60df30fbe1784e6875ea55a" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.0.img_mlp.net.2", | |
| "weight_key": "transformer_blocks.0.img_mlp.net.2.weight", | |
| "bias_key": "transformer_blocks.0.img_mlp.net.2.bias", | |
| "weight_shape": [ | |
| 3072, | |
| 12288 | |
| ], | |
| "bias_shape": [ | |
| 3072 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.0.img_mlp.net.2.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.0.img_mlp.net.2.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.0.img_mlp.net.2.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.0.img_mlp.net.2.bias_bf16", | |
| "weight_tensor_scale": 0.0002906435984186828, | |
| "weight_global_amax": 0.78125, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 768, | |
| "outer_tiles": 24, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "8a8cf7a4a5439eb2f3154ad636457ff25f665a95e755e58b2adbab7997f0ceb1", | |
| "source_bias_sha256": "a1bece05c6b61ebba9379c39bc81d6a19fc0cb2fd0543c9a0faccd926f598a64" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.0.txt_mlp.net.0.proj", | |
| "weight_key": "transformer_blocks.0.txt_mlp.net.0.proj.weight", | |
| "bias_key": "transformer_blocks.0.txt_mlp.net.0.proj.bias", | |
| "weight_shape": [ | |
| 12288, | |
| 3072 | |
| ], | |
| "bias_shape": [ | |
| 12288 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.0.txt_mlp.net.0.proj.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.0.txt_mlp.net.0.proj.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.0.txt_mlp.net.0.proj.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.0.txt_mlp.net.0.proj.bias_bf16", | |
| "weight_tensor_scale": 0.00024704704992473125, | |
| "weight_global_amax": 0.6640625, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 192, | |
| "outer_tiles": 96, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "40a2c2bf74fa661e51bf3294462ec6901e5f57abd1b0c8d8e3f7363c475aa33a", | |
| "source_bias_sha256": "f181d4fef402d8369e38ce9f64fe91903a5002f503b6dde71120c0ecb873d21b" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.0.txt_mlp.net.2", | |
| "weight_key": "transformer_blocks.0.txt_mlp.net.2.weight", | |
| "bias_key": "transformer_blocks.0.txt_mlp.net.2.bias", | |
| "weight_shape": [ | |
| 3072, | |
| 12288 | |
| ], | |
| "bias_shape": [ | |
| 3072 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.0.txt_mlp.net.2.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.0.txt_mlp.net.2.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.0.txt_mlp.net.2.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.0.txt_mlp.net.2.bias_bf16", | |
| "weight_tensor_scale": 0.00039527530316263437, | |
| "weight_global_amax": 1.0625, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 768, | |
| "outer_tiles": 24, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "37d5796891212472bfa97ce611f85266688107d579ebe1116eec5a958125152f", | |
| "source_bias_sha256": "862c6c05b8cfc438dd7074299f2303b588af5ae52d94b1dcb3e9e58bea78c770" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.1.img_mlp.net.0.proj", | |
| "weight_key": "transformer_blocks.1.img_mlp.net.0.proj.weight", | |
| "bias_key": "transformer_blocks.1.img_mlp.net.0.proj.bias", | |
| "weight_shape": [ | |
| 12288, | |
| 3072 | |
| ], | |
| "bias_shape": [ | |
| 12288 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.1.img_mlp.net.0.proj.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.1.img_mlp.net.0.proj.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.1.img_mlp.net.0.proj.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.1.img_mlp.net.0.proj.bias_bf16", | |
| "weight_tensor_scale": 0.00022815522970631719, | |
| "weight_global_amax": 0.61328125, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 192, | |
| "outer_tiles": 96, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "1ba75e81f6d7348769dae4a585d3127592fd2f402c101e3ddc88766f92d27617", | |
| "source_bias_sha256": "76d85e6964ca67fd92a1e6907ab4b5d7b839caa55de787b6e4a2969c234fcec9" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.1.img_mlp.net.2", | |
| "weight_key": "transformer_blocks.1.img_mlp.net.2.weight", | |
| "bias_key": "transformer_blocks.1.img_mlp.net.2.bias", | |
| "weight_shape": [ | |
| 3072, | |
| 12288 | |
| ], | |
| "bias_shape": [ | |
| 3072 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.1.img_mlp.net.2.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.1.img_mlp.net.2.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.1.img_mlp.net.2.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.1.img_mlp.net.2.bias_bf16", | |
| "weight_tensor_scale": 0.0004417782765813172, | |
| "weight_global_amax": 1.1875, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 768, | |
| "outer_tiles": 24, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "3f890473a9a99aa845e698713a852464055645b061b4b6ab61de59273b87be1b", | |
| "source_bias_sha256": "06b80199d80cc3916a202cfe220f59e8572dd406b0d2be8b17d59c1a3de98619" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.1.txt_mlp.net.0.proj", | |
| "weight_key": "transformer_blocks.1.txt_mlp.net.0.proj.weight", | |
| "bias_key": "transformer_blocks.1.txt_mlp.net.0.proj.bias", | |
| "weight_shape": [ | |
| 12288, | |
| 3072 | |
| ], | |
| "bias_shape": [ | |
| 12288 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.1.txt_mlp.net.0.proj.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.1.txt_mlp.net.0.proj.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.1.txt_mlp.net.0.proj.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.1.txt_mlp.net.0.proj.bias_bf16", | |
| "weight_tensor_scale": 0.00019618442456703633, | |
| "weight_global_amax": 0.52734375, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 192, | |
| "outer_tiles": 96, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "4a232bd2fdbf09232bd512cfb76c59502068529105e596c821bafff89c19813b", | |
| "source_bias_sha256": "e94fb97549fc326506f37a6824ead6a538828909466a6bf76933dfe5ce73680d" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.1.txt_mlp.net.2", | |
| "weight_key": "transformer_blocks.1.txt_mlp.net.2.weight", | |
| "bias_key": "transformer_blocks.1.txt_mlp.net.2.bias", | |
| "weight_shape": [ | |
| 3072, | |
| 12288 | |
| ], | |
| "bias_shape": [ | |
| 3072 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.1.txt_mlp.net.2.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.1.txt_mlp.net.2.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.1.txt_mlp.net.2.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.1.txt_mlp.net.2.bias_bf16", | |
| "weight_tensor_scale": 0.00039818172808736563, | |
| "weight_global_amax": 1.0703125, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 768, | |
| "outer_tiles": 24, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "042d74971fba414ffce050de7a65e1bebc33a2e49339b26f42518a9a9621701f", | |
| "source_bias_sha256": "faa59706a19669dc45d820edab49feca78c6b1d9d90aca4adbd87040c36f2fea" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.2.img_mlp.net.0.proj", | |
| "weight_key": "transformer_blocks.2.img_mlp.net.0.proj.weight", | |
| "bias_key": "transformer_blocks.2.img_mlp.net.0.proj.bias", | |
| "weight_shape": [ | |
| 12288, | |
| 3072 | |
| ], | |
| "bias_shape": [ | |
| 12288 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.2.img_mlp.net.0.proj.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.2.img_mlp.net.0.proj.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.2.img_mlp.net.0.proj.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.2.img_mlp.net.0.proj.bias_bf16", | |
| "weight_tensor_scale": 0.00030226935632526875, | |
| "weight_global_amax": 0.8125, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 192, | |
| "outer_tiles": 96, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "fceb9a0d7b4bfbddb10490b9d9e7f41c369e691bbafdccb003f8dda509fd3b1c", | |
| "source_bias_sha256": "84af366af51b7cd24dac8a9fda557a2c8a66ca80e9f8621ec4729d31d64bd241" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.2.img_mlp.net.2", | |
| "weight_key": "transformer_blocks.2.img_mlp.net.2.weight", | |
| "bias_key": "transformer_blocks.2.img_mlp.net.2.bias", | |
| "weight_shape": [ | |
| 3072, | |
| 12288 | |
| ], | |
| "bias_shape": [ | |
| 3072 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.2.img_mlp.net.2.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.2.img_mlp.net.2.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.2.img_mlp.net.2.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.2.img_mlp.net.2.bias_bf16", | |
| "weight_tensor_scale": 0.0005057198577560484, | |
| "weight_global_amax": 1.359375, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 768, | |
| "outer_tiles": 24, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "97220e0a04629f28c4ca39abfe46c07cf6183f660f391cd635be2871d325a09d", | |
| "source_bias_sha256": "6bcc96109a2496ee73d6a04bb8016641e90d568e6e46ea1a2e8b9e84925bfab9" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.2.txt_mlp.net.0.proj", | |
| "weight_key": "transformer_blocks.2.txt_mlp.net.0.proj.weight", | |
| "bias_key": "transformer_blocks.2.txt_mlp.net.0.proj.bias", | |
| "weight_shape": [ | |
| 12288, | |
| 3072 | |
| ], | |
| "bias_shape": [ | |
| 12288 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.2.txt_mlp.net.0.proj.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.2.txt_mlp.net.0.proj.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.2.txt_mlp.net.0.proj.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.2.txt_mlp.net.0.proj.bias_bf16", | |
| "weight_tensor_scale": 0.00022815522970631719, | |
| "weight_global_amax": 0.61328125, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 192, | |
| "outer_tiles": 96, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "764c79424da4eea44e62f03c1fda648bbd937ebf0b7c0bb43bede1e8b5130f29", | |
| "source_bias_sha256": "eca2f24456368d090da90c98a102af788833b3f3f9130f29b5b7a734e36f307e" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.2.txt_mlp.net.2", | |
| "weight_key": "transformer_blocks.2.txt_mlp.net.2.weight", | |
| "bias_key": "transformer_blocks.2.txt_mlp.net.2.bias", | |
| "weight_shape": [ | |
| 3072, | |
| 12288 | |
| ], | |
| "bias_shape": [ | |
| 3072 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.2.txt_mlp.net.2.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.2.txt_mlp.net.2.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.2.txt_mlp.net.2.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.2.txt_mlp.net.2.bias_bf16", | |
| "weight_tensor_scale": 0.0003589448460843414, | |
| "weight_global_amax": 0.96484375, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 768, | |
| "outer_tiles": 24, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "17e543841d93e1936514c8313a85e4b9444321db06111c33460173e47b0d1c93", | |
| "source_bias_sha256": "eed98a4db5990a369a55eb0865482f174e1418e6f4231bcf65637939a1df8354" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.3.img_mlp.net.0.proj", | |
| "weight_key": "transformer_blocks.3.img_mlp.net.0.proj.weight", | |
| "bias_key": "transformer_blocks.3.img_mlp.net.0.proj.bias", | |
| "weight_shape": [ | |
| 12288, | |
| 3072 | |
| ], | |
| "bias_shape": [ | |
| 12288 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.3.img_mlp.net.0.proj.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.3.img_mlp.net.0.proj.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.3.img_mlp.net.0.proj.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.3.img_mlp.net.0.proj.bias_bf16", | |
| "weight_tensor_scale": 0.00024850029149092734, | |
| "weight_global_amax": 0.66796875, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 192, | |
| "outer_tiles": 96, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "83c39bc7dd86d7e6a5f63eadc063b91d7bfd93a75a80c9536ddf28364d35ca6c", | |
| "source_bias_sha256": "4f40baf432bc9106a645e1182bce1c18535e1370f0851cb0f6e21d3a1646a08e" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.3.img_mlp.net.2", | |
| "weight_key": "transformer_blocks.3.img_mlp.net.2.weight", | |
| "bias_key": "transformer_blocks.3.img_mlp.net.2.bias", | |
| "weight_shape": [ | |
| 3072, | |
| 12288 | |
| ], | |
| "bias_shape": [ | |
| 3072 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.3.img_mlp.net.2.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.3.img_mlp.net.2.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.3.img_mlp.net.2.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.3.img_mlp.net.2.bias_bf16", | |
| "weight_tensor_scale": 0.0005871000466868281, | |
| "weight_global_amax": 1.578125, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 768, | |
| "outer_tiles": 24, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "e6945585cd816ccb2cd6e4a005326a105758f72270840842dea9c8617ba095ba", | |
| "source_bias_sha256": "230dcffbe97056ccb1975e1dbdc4274a89a217c64b240dcb70d1a639c8d50e10" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.3.txt_mlp.net.0.proj", | |
| "weight_key": "transformer_blocks.3.txt_mlp.net.0.proj.weight", | |
| "bias_key": "transformer_blocks.3.txt_mlp.net.0.proj.bias", | |
| "weight_shape": [ | |
| 12288, | |
| 3072 | |
| ], | |
| "bias_shape": [ | |
| 12288 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.3.txt_mlp.net.0.proj.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.3.txt_mlp.net.0.proj.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.3.txt_mlp.net.0.proj.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.3.txt_mlp.net.0.proj.bias_bf16", | |
| "weight_tensor_scale": 0.00022960844216868281, | |
| "weight_global_amax": 0.6171875, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 192, | |
| "outer_tiles": 96, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "786d08fafeb3519a0ab99e194105dfae1135ebbed18ae364fd62ba323c6525a2", | |
| "source_bias_sha256": "f493d8f0a032e8a1158ffc66ca65c00a78bdf8118d3bc307c4bab75605979d67" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.3.txt_mlp.net.2", | |
| "weight_key": "transformer_blocks.3.txt_mlp.net.2.weight", | |
| "bias_key": "transformer_blocks.3.txt_mlp.net.2.bias", | |
| "weight_shape": [ | |
| 3072, | |
| 12288 | |
| ], | |
| "bias_shape": [ | |
| 3072 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.3.txt_mlp.net.2.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.3.txt_mlp.net.2.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.3.txt_mlp.net.2.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.3.txt_mlp.net.2.bias_bf16", | |
| "weight_tensor_scale": 0.0004039946070406586, | |
| "weight_global_amax": 1.0859375, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 768, | |
| "outer_tiles": 24, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "ba11326b827a9600c9d0ae9227f2d3c0ab9971a09afd7df18689f4f1436a4e76", | |
| "source_bias_sha256": "c8a38cf2ae33d7b353cd71510a74ede4727b418285ea6dd5934274dcf5bbdf4f" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.4.img_mlp.net.0.proj", | |
| "weight_key": "transformer_blocks.4.img_mlp.net.0.proj.weight", | |
| "bias_key": "transformer_blocks.4.img_mlp.net.0.proj.bias", | |
| "weight_shape": [ | |
| 12288, | |
| 3072 | |
| ], | |
| "bias_shape": [ | |
| 12288 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.4.img_mlp.net.0.proj.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.4.img_mlp.net.0.proj.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.4.img_mlp.net.0.proj.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.4.img_mlp.net.0.proj.bias_bf16", | |
| "weight_tensor_scale": 0.0003516787546686828, | |
| "weight_global_amax": 0.9453125, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 192, | |
| "outer_tiles": 96, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "2da64d233d618c639d7ac3a988c045771a20bb77634d272949dc1f9d3a26e6fa", | |
| "source_bias_sha256": "6079037543d2b43a319f02021a6f71d2874efaaafd6a5a58f2813a9e160a9ca6" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.4.img_mlp.net.2", | |
| "weight_key": "transformer_blocks.4.img_mlp.net.2.weight", | |
| "bias_key": "transformer_blocks.4.img_mlp.net.2.bias", | |
| "weight_shape": [ | |
| 3072, | |
| 12288 | |
| ], | |
| "bias_shape": [ | |
| 3072 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.4.img_mlp.net.2.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.4.img_mlp.net.2.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.4.img_mlp.net.2.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.4.img_mlp.net.2.bias_bf16", | |
| "weight_tensor_scale": 0.0009533109841868281, | |
| "weight_global_amax": 2.5625, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 768, | |
| "outer_tiles": 24, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "4db777bdb0beb299cb30ccc2043870bc4f311d5e6e30b6ad28e1cc7c29af96ed", | |
| "source_bias_sha256": "d4e4ac8e94b3fd245c412ee58d3fc91e815fd03e49684d8596ac58566145f1fe" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.4.txt_mlp.net.0.proj", | |
| "weight_key": "transformer_blocks.4.txt_mlp.net.0.proj.weight", | |
| "bias_key": "transformer_blocks.4.txt_mlp.net.0.proj.bias", | |
| "weight_shape": [ | |
| 12288, | |
| 3072 | |
| ], | |
| "bias_shape": [ | |
| 12288 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.4.txt_mlp.net.0.proj.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.4.txt_mlp.net.0.proj.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.4.txt_mlp.net.0.proj.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.4.txt_mlp.net.0.proj.bias_bf16", | |
| "weight_tensor_scale": 0.00018528530199546367, | |
| "weight_global_amax": 0.498046875, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 192, | |
| "outer_tiles": 96, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "0c956820e6938c52af5a237da6dcd4c41b509865b7631407a896cc6da6bdb985", | |
| "source_bias_sha256": "edd0dfc9060a78b09c53bd0f38a9d0a5263faadbe3c522a7022115a2e09c211e" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.4.txt_mlp.net.2", | |
| "weight_key": "transformer_blocks.4.txt_mlp.net.2.weight", | |
| "bias_key": "transformer_blocks.4.txt_mlp.net.2.bias", | |
| "weight_shape": [ | |
| 3072, | |
| 12288 | |
| ], | |
| "bias_shape": [ | |
| 3072 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.4.txt_mlp.net.2.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.4.txt_mlp.net.2.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.4.txt_mlp.net.2.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.4.txt_mlp.net.2.bias_bf16", | |
| "weight_tensor_scale": 0.0003284272679593414, | |
| "weight_global_amax": 0.8828125, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 768, | |
| "outer_tiles": 24, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "005cd523afdeb3e9d32a2b29dd0499fa994fd8a79af245604661ee8bc1af59c8", | |
| "source_bias_sha256": "f333cc3ca7d6299b5cf44859662914d0fae7cbeba0433e0ae487d64cdc755685" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.5.img_mlp.net.0.proj", | |
| "weight_key": "transformer_blocks.5.img_mlp.net.0.proj.weight", | |
| "bias_key": "transformer_blocks.5.img_mlp.net.0.proj.bias", | |
| "weight_shape": [ | |
| 12288, | |
| 3072 | |
| ], | |
| "bias_shape": [ | |
| 12288 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.5.img_mlp.net.0.proj.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.5.img_mlp.net.0.proj.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.5.img_mlp.net.0.proj.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.5.img_mlp.net.0.proj.bias_bf16", | |
| "weight_tensor_scale": 0.00040108818211592734, | |
| "weight_global_amax": 1.078125, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 192, | |
| "outer_tiles": 96, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "8419f255277f3a77facc9090e6882ab54d41e6f9f0bebb7930416e07c7276f55", | |
| "source_bias_sha256": "c738466822f96ba14226ee1f254ec83099479436e1ff34d7ae220fdc7bad423d" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.5.img_mlp.net.2", | |
| "weight_key": "transformer_blocks.5.img_mlp.net.2.weight", | |
| "bias_key": "transformer_blocks.5.img_mlp.net.2.bias", | |
| "weight_shape": [ | |
| 3072, | |
| 12288 | |
| ], | |
| "bias_shape": [ | |
| 3072 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.5.img_mlp.net.2.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.5.img_mlp.net.2.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.5.img_mlp.net.2.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.5.img_mlp.net.2.bias_bf16", | |
| "weight_tensor_scale": 0.0006684802938252687, | |
| "weight_global_amax": 1.796875, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 768, | |
| "outer_tiles": 24, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "97fbe5a8ee7151217b82e12295e02681af0ca707d3595c58b40aed31c3ae23b7", | |
| "source_bias_sha256": "af3f83a94e0368f51c65b26a880b1bf3b64543910e9e270a04faa12aa58680da" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.5.txt_mlp.net.0.proj", | |
| "weight_key": "transformer_blocks.5.txt_mlp.net.0.proj.weight", | |
| "bias_key": "transformer_blocks.5.txt_mlp.net.0.proj.bias", | |
| "weight_shape": [ | |
| 12288, | |
| 3072 | |
| ], | |
| "bias_shape": [ | |
| 12288 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.5.txt_mlp.net.0.proj.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.5.txt_mlp.net.0.proj.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.5.txt_mlp.net.0.proj.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.5.txt_mlp.net.0.proj.bias_bf16", | |
| "weight_tensor_scale": 0.0001874651206890121, | |
| "weight_global_amax": 0.50390625, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 192, | |
| "outer_tiles": 96, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "7e3b210a8d107962865efa3f3974a9ea29a60732f8002539e71837ebf9fb3787", | |
| "source_bias_sha256": "a154c2419b8da86533eeb0e5a69c37cbfe5eff859f8196e192d5f942743c690d" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.5.txt_mlp.net.2", | |
| "weight_key": "transformer_blocks.5.txt_mlp.net.2.weight", | |
| "bias_key": "transformer_blocks.5.txt_mlp.net.2.bias", | |
| "weight_shape": [ | |
| 3072, | |
| 12288 | |
| ], | |
| "bias_shape": [ | |
| 3072 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.5.txt_mlp.net.2.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.5.txt_mlp.net.2.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.5.txt_mlp.net.2.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.5.txt_mlp.net.2.bias_bf16", | |
| "weight_tensor_scale": 0.0003807431203313172, | |
| "weight_global_amax": 1.0234375, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 768, | |
| "outer_tiles": 24, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "906a7d3e7d32c10037e731b32904f683ff3529e0605dc619132c6095e87aeec4", | |
| "source_bias_sha256": "e47066d5093c8d5df144ad237d0c0cbb8186b7c176764338b777873e623e226e" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.6.img_mlp.net.0.proj", | |
| "weight_key": "transformer_blocks.6.img_mlp.net.0.proj.weight", | |
| "bias_key": "transformer_blocks.6.img_mlp.net.0.proj.bias", | |
| "weight_shape": [ | |
| 12288, | |
| 3072 | |
| ], | |
| "bias_shape": [ | |
| 12288 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.6.img_mlp.net.0.proj.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.6.img_mlp.net.0.proj.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.6.img_mlp.net.0.proj.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.6.img_mlp.net.0.proj.bias_bf16", | |
| "weight_tensor_scale": 0.00043015251867473125, | |
| "weight_global_amax": 1.15625, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 192, | |
| "outer_tiles": 96, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "6e2eaa52c7e036559066e6d0bd3e259b8b4a9e6010cfb5201d255aa673a95e9a", | |
| "source_bias_sha256": "559dcbcbecdda86591182bdad53d03f173400d6abbdada125766b5c5e180887f" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.6.img_mlp.net.2", | |
| "weight_key": "transformer_blocks.6.img_mlp.net.2.weight", | |
| "bias_key": "transformer_blocks.6.img_mlp.net.2.bias", | |
| "weight_shape": [ | |
| 3072, | |
| 12288 | |
| ], | |
| "bias_shape": [ | |
| 3072 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.6.img_mlp.net.2.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.6.img_mlp.net.2.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.6.img_mlp.net.2.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.6.img_mlp.net.2.bias_bf16", | |
| "weight_tensor_scale": 0.0007672990905120969, | |
| "weight_global_amax": 2.0625, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 768, | |
| "outer_tiles": 24, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "5d161c4e9f626a44cd150ccd49e4b6a3b91dfe9c78628edbe2d05bcfd205ee38", | |
| "source_bias_sha256": "88e2f67cc46733f0a734a37bf6d6f6a52b12bbbfce8346a9ca81b4942a3c5c89" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.6.txt_mlp.net.0.proj", | |
| "weight_key": "transformer_blocks.6.txt_mlp.net.0.proj.weight", | |
| "bias_key": "transformer_blocks.6.txt_mlp.net.0.proj.bias", | |
| "weight_shape": [ | |
| 12288, | |
| 3072 | |
| ], | |
| "bias_shape": [ | |
| 12288 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.6.txt_mlp.net.0.proj.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.6.txt_mlp.net.0.proj.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.6.txt_mlp.net.0.proj.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.6.txt_mlp.net.0.proj.bias_bf16", | |
| "weight_tensor_scale": 0.0002601260202936828, | |
| "weight_global_amax": 0.69921875, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 192, | |
| "outer_tiles": 96, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "fdd4c5e88701022033d97501bd2781fdc526eaa9427775a790756a9bb7c1fda2", | |
| "source_bias_sha256": "37f6d759de4dfb875a8c7076fae88f75c2c5fdade293907895fead213e7a0ae7" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.6.txt_mlp.net.2", | |
| "weight_key": "transformer_blocks.6.txt_mlp.net.2.weight", | |
| "bias_key": "transformer_blocks.6.txt_mlp.net.2.bias", | |
| "weight_shape": [ | |
| 3072, | |
| 12288 | |
| ], | |
| "bias_shape": [ | |
| 3072 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.6.txt_mlp.net.2.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.6.txt_mlp.net.2.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.6.txt_mlp.net.2.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.6.txt_mlp.net.2.bias_bf16", | |
| "weight_tensor_scale": 0.00047665549209341407, | |
| "weight_global_amax": 1.28125, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 768, | |
| "outer_tiles": 24, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "b28669406ee0f0e295c1a0b19241c1de874952807a0754d31078b56b1e213f2d", | |
| "source_bias_sha256": "a5479cb1785b82d01897ca73a0b560b1382dcb7b60c8dce315e247d795e04d0f" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.7.img_mlp.net.0.proj", | |
| "weight_key": "transformer_blocks.7.img_mlp.net.0.proj.weight", | |
| "bias_key": "transformer_blocks.7.img_mlp.net.0.proj.bias", | |
| "weight_shape": [ | |
| 12288, | |
| 3072 | |
| ], | |
| "bias_shape": [ | |
| 12288 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.7.img_mlp.net.0.proj.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.7.img_mlp.net.0.proj.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.7.img_mlp.net.0.proj.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.7.img_mlp.net.0.proj.bias_bf16", | |
| "weight_tensor_scale": 0.0003138950851280242, | |
| "weight_global_amax": 0.84375, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 192, | |
| "outer_tiles": 96, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "9bcc9591965c0b74076ad9da74307faa9c7fa0ab63a81f8f25960c2286ed39bf", | |
| "source_bias_sha256": "b043d399ff8f040405104ffb775c9f902f95dca6de06a6b1ae2f6c90e5f28d7c" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.7.img_mlp.net.2", | |
| "weight_key": "transformer_blocks.7.img_mlp.net.2.weight", | |
| "bias_key": "transformer_blocks.7.img_mlp.net.2.bias", | |
| "weight_shape": [ | |
| 3072, | |
| 12288 | |
| ], | |
| "bias_shape": [ | |
| 3072 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.7.img_mlp.net.2.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.7.img_mlp.net.2.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.7.img_mlp.net.2.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.7.img_mlp.net.2.bias_bf16", | |
| "weight_tensor_scale": 0.0008603050373494625, | |
| "weight_global_amax": 2.3125, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 768, | |
| "outer_tiles": 24, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "e7efbd25a434220ac1f0f4f11d8e501b699e11a948a86148460b4139db5ba3bf", | |
| "source_bias_sha256": "b2cd1665c361e61f731ce7b474acb2e5f57087c2cacf53f92f418256c5f1ce86" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.7.txt_mlp.net.0.proj", | |
| "weight_key": "transformer_blocks.7.txt_mlp.net.0.proj.weight", | |
| "bias_key": "transformer_blocks.7.txt_mlp.net.0.proj.bias", | |
| "weight_shape": [ | |
| 12288, | |
| 3072 | |
| ], | |
| "bias_shape": [ | |
| 12288 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.7.txt_mlp.net.0.proj.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.7.txt_mlp.net.0.proj.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.7.txt_mlp.net.0.proj.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.7.txt_mlp.net.0.proj.bias_bf16", | |
| "weight_tensor_scale": 0.00027901786961592734, | |
| "weight_global_amax": 0.75, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 192, | |
| "outer_tiles": 96, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "cb6126a8544c10ae9f23b3d782c83258a8e700a42ac2ad2e5001099e305f271b", | |
| "source_bias_sha256": "ed4294351c27eaf4e5f85bb926daed8e479f3862e67e79142788a57490059768" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.7.txt_mlp.net.2", | |
| "weight_key": "transformer_blocks.7.txt_mlp.net.2.weight", | |
| "bias_key": "transformer_blocks.7.txt_mlp.net.2.bias", | |
| "weight_shape": [ | |
| 3072, | |
| 12288 | |
| ], | |
| "bias_shape": [ | |
| 3072 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.7.txt_mlp.net.2.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.7.txt_mlp.net.2.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.7.txt_mlp.net.2.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.7.txt_mlp.net.2.bias_bf16", | |
| "weight_tensor_scale": 0.00039527530316263437, | |
| "weight_global_amax": 1.0625, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 768, | |
| "outer_tiles": 24, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "84caab56f058269c0576a27a2b52ae9945f225ba5b6c158cd3f8b8daf2a135f7", | |
| "source_bias_sha256": "cb733e6b65392d1189b7a9d92984c51f8638a1375b6475023b26cd3c362ee7a8" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.8.img_mlp.net.0.proj", | |
| "weight_key": "transformer_blocks.8.img_mlp.net.0.proj.weight", | |
| "bias_key": "transformer_blocks.8.img_mlp.net.0.proj.bias", | |
| "weight_shape": [ | |
| 12288, | |
| 3072 | |
| ], | |
| "bias_shape": [ | |
| 12288 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.8.img_mlp.net.0.proj.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.8.img_mlp.net.0.proj.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.8.img_mlp.net.0.proj.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.8.img_mlp.net.0.proj.bias_bf16", | |
| "weight_tensor_scale": 0.00041562033584341407, | |
| "weight_global_amax": 1.1171875, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 192, | |
| "outer_tiles": 96, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "b6da93ff7c47065d4dc8d5b44f02c1f956bb8f373d89526565b2cc1cc3c731ef", | |
| "source_bias_sha256": "a155557ac69e102404fcde3489febf8d75d8df81a58f010ae722f73b808c0cf0" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.8.img_mlp.net.2", | |
| "weight_key": "transformer_blocks.8.img_mlp.net.2.weight", | |
| "bias_key": "transformer_blocks.8.img_mlp.net.2.bias", | |
| "weight_shape": [ | |
| 3072, | |
| 12288 | |
| ], | |
| "bias_shape": [ | |
| 3072 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.8.img_mlp.net.2.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.8.img_mlp.net.2.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.8.img_mlp.net.2.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.8.img_mlp.net.2.bias_bf16", | |
| "weight_tensor_scale": 0.0005260648904368281, | |
| "weight_global_amax": 1.4140625, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 768, | |
| "outer_tiles": 24, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "928218f79be4de0ca6fd6f855310ad7a91b588be9c970c9ea6a11f55b1a1e413", | |
| "source_bias_sha256": "669dfaa63f72b425feecba8fb58b4c43926e96d971f2f2e9241632855c559236" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.8.txt_mlp.net.0.proj", | |
| "weight_key": "transformer_blocks.8.txt_mlp.net.0.proj.weight", | |
| "bias_key": "transformer_blocks.8.txt_mlp.net.0.proj.bias", | |
| "weight_shape": [ | |
| 12288, | |
| 3072 | |
| ], | |
| "bias_shape": [ | |
| 12288 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.8.txt_mlp.net.0.proj.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.8.txt_mlp.net.0.proj.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.8.txt_mlp.net.0.proj.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.8.txt_mlp.net.0.proj.bias_bf16", | |
| "weight_tensor_scale": 0.0002601260202936828, | |
| "weight_global_amax": 0.69921875, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 192, | |
| "outer_tiles": 96, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "9342523cc73ba392d60e1dffdfe37824a20e5173281de80f85a4cef4893363db", | |
| "source_bias_sha256": "8c13b195dab6195c7654e07b8416da4c952df50f16fd9eccb248c1a38262df10" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.8.txt_mlp.net.2", | |
| "weight_key": "transformer_blocks.8.txt_mlp.net.2.weight", | |
| "bias_key": "transformer_blocks.8.txt_mlp.net.2.bias", | |
| "weight_shape": [ | |
| 3072, | |
| 12288 | |
| ], | |
| "bias_shape": [ | |
| 3072 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.8.txt_mlp.net.2.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.8.txt_mlp.net.2.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.8.txt_mlp.net.2.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.8.txt_mlp.net.2.bias_bf16", | |
| "weight_tensor_scale": 0.00042724609375, | |
| "weight_global_amax": 1.1484375, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 768, | |
| "outer_tiles": 24, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "b41e8d7028a6f787c256611afeae81c47990e7a91e9e682902dcedfcc9c13eaa", | |
| "source_bias_sha256": "4dcdabdb17aabba97ac13d6e3c733a3b212d7e9e7092a7e0fbde6ccd81337a76" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.9.img_mlp.net.0.proj", | |
| "weight_key": "transformer_blocks.9.img_mlp.net.0.proj.weight", | |
| "bias_key": "transformer_blocks.9.img_mlp.net.0.proj.bias", | |
| "weight_shape": [ | |
| 12288, | |
| 3072 | |
| ], | |
| "bias_shape": [ | |
| 12288 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.9.img_mlp.net.0.proj.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.9.img_mlp.net.0.proj.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.9.img_mlp.net.0.proj.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.9.img_mlp.net.0.proj.bias_bf16", | |
| "weight_tensor_scale": 0.0004911876749247313, | |
| "weight_global_amax": 1.3203125, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 192, | |
| "outer_tiles": 96, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "82c42f661aab545f0807ff6aa8a1da83f8b50c186b2fec439ab65540449f4b58", | |
| "source_bias_sha256": "790d198723850350f0430dc9262fc2019d3b1e3b7e32934212015bff7b767ff1" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.9.img_mlp.net.2", | |
| "weight_key": "transformer_blocks.9.img_mlp.net.2.weight", | |
| "bias_key": "transformer_blocks.9.img_mlp.net.2.bias", | |
| "weight_shape": [ | |
| 3072, | |
| 12288 | |
| ], | |
| "bias_shape": [ | |
| 3072 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.9.img_mlp.net.2.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.9.img_mlp.net.2.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.9.img_mlp.net.2.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.9.img_mlp.net.2.bias_bf16", | |
| "weight_tensor_scale": 0.0005435035564005375, | |
| "weight_global_amax": 1.4609375, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 768, | |
| "outer_tiles": 24, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "0c2af8a1f4e348d1150532b2fb9da63c33bee8603e1d4720b2c830632d6157c2", | |
| "source_bias_sha256": "37f0b49b8f98d5981a92cc830bdcf202e077d458dcad8bedd26982ce12aef170" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.9.txt_mlp.net.0.proj", | |
| "weight_key": "transformer_blocks.9.txt_mlp.net.0.proj.weight", | |
| "bias_key": "transformer_blocks.9.txt_mlp.net.0.proj.bias", | |
| "weight_shape": [ | |
| 12288, | |
| 3072 | |
| ], | |
| "bias_shape": [ | |
| 12288 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.9.txt_mlp.net.0.proj.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.9.txt_mlp.net.0.proj.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.9.txt_mlp.net.0.proj.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.9.txt_mlp.net.0.proj.bias_bf16", | |
| "weight_tensor_scale": 0.0002819242945406586, | |
| "weight_global_amax": 0.7578125, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 192, | |
| "outer_tiles": 96, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "7569c27e45b5327904b003909f88f373e99bd93b150a5508bd3c49470da94c8f", | |
| "source_bias_sha256": "5e5a68b982622caffbb11f7a7b131939ece3715bc0ffbcb32fa29dd445fd48ea" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.9.txt_mlp.net.2", | |
| "weight_key": "transformer_blocks.9.txt_mlp.net.2.weight", | |
| "bias_key": "transformer_blocks.9.txt_mlp.net.2.bias", | |
| "weight_shape": [ | |
| 3072, | |
| 12288 | |
| ], | |
| "bias_shape": [ | |
| 3072 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.9.txt_mlp.net.2.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.9.txt_mlp.net.2.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.9.txt_mlp.net.2.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.9.txt_mlp.net.2.bias_bf16", | |
| "weight_tensor_scale": 0.0006045387126505375, | |
| "weight_global_amax": 1.625, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 768, | |
| "outer_tiles": 24, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "5ce62f2c67fdf6f7caf6841ed2a2cee2e72194768207c0ab631f147dae462f80", | |
| "source_bias_sha256": "11cb1954c19d92e661518cca9971ac432a4ff17f9d8832d789f91dc9bbaeca37" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.10.img_mlp.net.0.proj", | |
| "weight_key": "transformer_blocks.10.img_mlp.net.0.proj.weight", | |
| "bias_key": "transformer_blocks.10.img_mlp.net.0.proj.bias", | |
| "weight_shape": [ | |
| 12288, | |
| 3072 | |
| ], | |
| "bias_shape": [ | |
| 12288 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.10.img_mlp.net.0.proj.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.10.img_mlp.net.0.proj.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.10.img_mlp.net.0.proj.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.10.img_mlp.net.0.proj.bias_bf16", | |
| "weight_tensor_scale": 0.00019618442456703633, | |
| "weight_global_amax": 0.52734375, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 192, | |
| "outer_tiles": 96, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "25c16255a05df5f725a32fa1183740eecd018f52fa5e52d7b117f396656307f7", | |
| "source_bias_sha256": "79ba550cd845a1d1156fa06661ea9a268331d2d3f13ed4f32225318796bdb2f1" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.10.img_mlp.net.2", | |
| "weight_key": "transformer_blocks.10.img_mlp.net.2.weight", | |
| "bias_key": "transformer_blocks.10.img_mlp.net.2.bias", | |
| "weight_shape": [ | |
| 3072, | |
| 12288 | |
| ], | |
| "bias_shape": [ | |
| 3072 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.10.img_mlp.net.2.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.10.img_mlp.net.2.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.10.img_mlp.net.2.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.10.img_mlp.net.2.bias_bf16", | |
| "weight_tensor_scale": 0.0005812871968373656, | |
| "weight_global_amax": 1.5625, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 768, | |
| "outer_tiles": 24, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "710837d28af5c4871c535b02e29431563d3d3755236b8a4acf8e0ebb8c334764", | |
| "source_bias_sha256": "021b5a6799ed0e35ee49a28af9acc77561bc1defb184a47df39d28178edf3084" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.10.txt_mlp.net.0.proj", | |
| "weight_key": "transformer_blocks.10.txt_mlp.net.0.proj.weight", | |
| "bias_key": "transformer_blocks.10.txt_mlp.net.0.proj.bias", | |
| "weight_shape": [ | |
| 12288, | |
| 3072 | |
| ], | |
| "bias_shape": [ | |
| 12288 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.10.txt_mlp.net.0.proj.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.10.txt_mlp.net.0.proj.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.10.txt_mlp.net.0.proj.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.10.txt_mlp.net.0.proj.bias_bf16", | |
| "weight_tensor_scale": 0.00015476772387046367, | |
| "weight_global_amax": 0.416015625, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 192, | |
| "outer_tiles": 96, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "dd76244de2f2319b3c617e91cc6e863fe6d05c83d0038d3b903e198993c497dc", | |
| "source_bias_sha256": "dd6336134c5a891280e5e0b22ae7690cddf6d4cf937c4198ac5865202958a162" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.10.txt_mlp.net.2", | |
| "weight_key": "transformer_blocks.10.txt_mlp.net.2.weight", | |
| "bias_key": "transformer_blocks.10.txt_mlp.net.2.bias", | |
| "weight_shape": [ | |
| 3072, | |
| 12288 | |
| ], | |
| "bias_shape": [ | |
| 3072 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.10.txt_mlp.net.2.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.10.txt_mlp.net.2.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.10.txt_mlp.net.2.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.10.txt_mlp.net.2.bias_bf16", | |
| "weight_tensor_scale": 0.00018383207498118281, | |
| "weight_global_amax": 0.494140625, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 768, | |
| "outer_tiles": 24, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "3748e2fb3c162f846342f7a4c5f8f600d614150163332c2b32ba5f3c7291f4b4", | |
| "source_bias_sha256": "333b152ef1291bf1fcd6fc796345483195b2632b70e0b6ab13ddabf8b7bb20e4" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.11.img_mlp.net.0.proj", | |
| "weight_key": "transformer_blocks.11.img_mlp.net.0.proj.weight", | |
| "bias_key": "transformer_blocks.11.img_mlp.net.0.proj.bias", | |
| "weight_shape": [ | |
| 12288, | |
| 3072 | |
| ], | |
| "bias_shape": [ | |
| 12288 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.11.img_mlp.net.0.proj.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.11.img_mlp.net.0.proj.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.11.img_mlp.net.0.proj.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.11.img_mlp.net.0.proj.bias_bf16", | |
| "weight_tensor_scale": 0.0002252487902296707, | |
| "weight_global_amax": 0.60546875, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 192, | |
| "outer_tiles": 96, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "81207ce0a6cc70bf94346c5c74e7729a9df4e2cc87e943f5461ce6ccdcddf25f", | |
| "source_bias_sha256": "3baf43b038997ba0ebe0e9326e17fe210727fe89fcc832f93082381c5cca2adc" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.11.img_mlp.net.2", | |
| "weight_key": "transformer_blocks.11.img_mlp.net.2.weight", | |
| "bias_key": "transformer_blocks.11.img_mlp.net.2.bias", | |
| "weight_shape": [ | |
| 3072, | |
| 12288 | |
| ], | |
| "bias_shape": [ | |
| 3072 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.11.img_mlp.net.2.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.11.img_mlp.net.2.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.11.img_mlp.net.2.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.11.img_mlp.net.2.bias_bf16", | |
| "weight_tensor_scale": 0.0009707496501505375, | |
| "weight_global_amax": 2.609375, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 768, | |
| "outer_tiles": 24, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "ba3889bd47bfe14a07b6b4d9f624d1187980d07fb81de6d0be765f195a22c487", | |
| "source_bias_sha256": "fd48e42ed9061e494e065910ae622964b25e6c3abc13613b38f4b4a4a0c8ec13" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.11.txt_mlp.net.0.proj", | |
| "weight_key": "transformer_blocks.11.txt_mlp.net.0.proj.weight", | |
| "bias_key": "transformer_blocks.11.txt_mlp.net.0.proj.bias", | |
| "weight_shape": [ | |
| 12288, | |
| 3072 | |
| ], | |
| "bias_shape": [ | |
| 12288 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.11.txt_mlp.net.0.proj.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.11.txt_mlp.net.0.proj.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.11.txt_mlp.net.0.proj.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.11.txt_mlp.net.0.proj.bias_bf16", | |
| "weight_tensor_scale": 1.3850983577867737e-06, | |
| "weight_global_amax": 0.00372314453125, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 192, | |
| "outer_tiles": 96, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "3e43f754f9d63e3dd6d6a71c4cc0c15344bd56da4d5a19a7adeb54445b820d72", | |
| "source_bias_sha256": "062da4757b9f03b3273c98cf91a517e2b4a667cbfe1f0ba190b02b626b0ac75c" | |
| }, | |
| { | |
| "module_key": "transformer_blocks.11.txt_mlp.net.2", | |
| "weight_key": "transformer_blocks.11.txt_mlp.net.2.weight", | |
| "bias_key": "transformer_blocks.11.txt_mlp.net.2.bias", | |
| "weight_shape": [ | |
| 3072, | |
| 12288 | |
| ], | |
| "bias_shape": [ | |
| 3072 | |
| ], | |
| "weight_tensor_scale_key": "targets.transformer_blocks.11.txt_mlp.net.2.weight_tensor_scale", | |
| "artifact_weight_key": "targets.transformer_blocks.11.txt_mlp.net.2.packed_weight_e2m1", | |
| "artifact_scale_key": "targets.transformer_blocks.11.txt_mlp.net.2.packed_scales_ue4m3", | |
| "artifact_bias_key": "targets.transformer_blocks.11.txt_mlp.net.2.bias_bf16", | |
| "weight_tensor_scale": 6.925491788933869e-07, | |
| "weight_global_amax": 0.001861572265625, | |
| "packed_weight_bytes": 18874368, | |
| "packed_scale_bytes": 2359296, | |
| "scale_layout": { | |
| "inner_dim": 768, | |
| "outer_tiles": 24, | |
| "bytes": 2359296 | |
| }, | |
| "source_weight_sha256": "4a74e0fc231d978ddbca106304430d9dcb511c7284e3c18c245005626e88224d", | |
| "source_bias_sha256": "6ef0d2f05867e5017824fc95cfd0a5803ce3d736cd622d82d92b603938d66e41" | |
| } | |
| ], | |
| "non_target_keys": [ | |
| "img_in.bias", | |
| "img_in.weight", | |
| "norm_out.linear.bias", | |
| "norm_out.linear.weight", | |
| "proj_out.bias", | |
| "proj_out.weight", | |
| "time_text_embed.timestep_embedder.linear_1.bias", | |
| "time_text_embed.timestep_embedder.linear_1.weight", | |
| "time_text_embed.timestep_embedder.linear_2.bias", | |
| "time_text_embed.timestep_embedder.linear_2.weight", | |
| "transformer_blocks.0.attn.add_k_proj.bias", | |
| "transformer_blocks.0.attn.add_k_proj.weight", | |
| "transformer_blocks.0.attn.add_q_proj.bias", | |
| "transformer_blocks.0.attn.add_q_proj.weight", | |
| "transformer_blocks.0.attn.add_v_proj.bias", | |
| "transformer_blocks.0.attn.add_v_proj.weight", | |
| "transformer_blocks.0.attn.norm_added_k.weight", | |
| "transformer_blocks.0.attn.norm_added_q.weight", | |
| "transformer_blocks.0.attn.norm_k.weight", | |
| "transformer_blocks.0.attn.norm_q.weight", | |
| "transformer_blocks.0.attn.to_add_out.bias", | |
| "transformer_blocks.0.attn.to_add_out.weight", | |
| "transformer_blocks.0.attn.to_k.bias", | |
| "transformer_blocks.0.attn.to_k.weight", | |
| "transformer_blocks.0.attn.to_out.0.bias", | |
| "transformer_blocks.0.attn.to_out.0.weight", | |
| "transformer_blocks.0.attn.to_q.bias", | |
| "transformer_blocks.0.attn.to_q.weight", | |
| "transformer_blocks.0.attn.to_v.bias", | |
| "transformer_blocks.0.attn.to_v.weight", | |
| "transformer_blocks.0.img_mod.1.bias", | |
| "transformer_blocks.0.img_mod.1.weight", | |
| "transformer_blocks.0.txt_mod.1.bias", | |
| "transformer_blocks.0.txt_mod.1.weight", | |
| "transformer_blocks.1.attn.add_k_proj.bias", | |
| "transformer_blocks.1.attn.add_k_proj.weight", | |
| "transformer_blocks.1.attn.add_q_proj.bias", | |
| "transformer_blocks.1.attn.add_q_proj.weight", | |
| "transformer_blocks.1.attn.add_v_proj.bias", | |
| "transformer_blocks.1.attn.add_v_proj.weight", | |
| "transformer_blocks.1.attn.norm_added_k.weight", | |
| "transformer_blocks.1.attn.norm_added_q.weight", | |
| "transformer_blocks.1.attn.norm_k.weight", | |
| "transformer_blocks.1.attn.norm_q.weight", | |
| "transformer_blocks.1.attn.to_add_out.bias", | |
| "transformer_blocks.1.attn.to_add_out.weight", | |
| "transformer_blocks.1.attn.to_k.bias", | |
| "transformer_blocks.1.attn.to_k.weight", | |
| "transformer_blocks.1.attn.to_out.0.bias", | |
| "transformer_blocks.1.attn.to_out.0.weight", | |
| "transformer_blocks.1.attn.to_q.bias", | |
| "transformer_blocks.1.attn.to_q.weight", | |
| "transformer_blocks.1.attn.to_v.bias", | |
| "transformer_blocks.1.attn.to_v.weight", | |
| "transformer_blocks.1.img_mod.1.bias", | |
| "transformer_blocks.1.img_mod.1.weight", | |
| "transformer_blocks.1.txt_mod.1.bias", | |
| "transformer_blocks.1.txt_mod.1.weight", | |
| "transformer_blocks.10.attn.add_k_proj.bias", | |
| "transformer_blocks.10.attn.add_k_proj.weight", | |
| "transformer_blocks.10.attn.add_q_proj.bias", | |
| "transformer_blocks.10.attn.add_q_proj.weight", | |
| "transformer_blocks.10.attn.add_v_proj.bias", | |
| "transformer_blocks.10.attn.add_v_proj.weight", | |
| "transformer_blocks.10.attn.norm_added_k.weight", | |
| "transformer_blocks.10.attn.norm_added_q.weight", | |
| "transformer_blocks.10.attn.norm_k.weight", | |
| "transformer_blocks.10.attn.norm_q.weight", | |
| "transformer_blocks.10.attn.to_add_out.bias", | |
| "transformer_blocks.10.attn.to_add_out.weight", | |
| "transformer_blocks.10.attn.to_k.bias", | |
| "transformer_blocks.10.attn.to_k.weight", | |
| "transformer_blocks.10.attn.to_out.0.bias", | |
| "transformer_blocks.10.attn.to_out.0.weight", | |
| "transformer_blocks.10.attn.to_q.bias", | |
| "transformer_blocks.10.attn.to_q.weight", | |
| "transformer_blocks.10.attn.to_v.bias", | |
| "transformer_blocks.10.attn.to_v.weight", | |
| "transformer_blocks.10.img_mod.1.bias", | |
| "transformer_blocks.10.img_mod.1.weight", | |
| "transformer_blocks.10.txt_mod.1.bias", | |
| "transformer_blocks.10.txt_mod.1.weight", | |
| "transformer_blocks.11.attn.add_k_proj.bias", | |
| "transformer_blocks.11.attn.add_k_proj.weight", | |
| "transformer_blocks.11.attn.add_q_proj.bias", | |
| "transformer_blocks.11.attn.add_q_proj.weight", | |
| "transformer_blocks.11.attn.add_v_proj.bias", | |
| "transformer_blocks.11.attn.add_v_proj.weight", | |
| "transformer_blocks.11.attn.norm_added_k.weight", | |
| "transformer_blocks.11.attn.norm_added_q.weight", | |
| "transformer_blocks.11.attn.norm_k.weight", | |
| "transformer_blocks.11.attn.norm_q.weight", | |
| "transformer_blocks.11.attn.to_add_out.bias", | |
| "transformer_blocks.11.attn.to_add_out.weight", | |
| "transformer_blocks.11.attn.to_k.bias", | |
| "transformer_blocks.11.attn.to_k.weight", | |
| "transformer_blocks.11.attn.to_out.0.bias", | |
| "transformer_blocks.11.attn.to_out.0.weight", | |
| "transformer_blocks.11.attn.to_q.bias", | |
| "transformer_blocks.11.attn.to_q.weight", | |
| "transformer_blocks.11.attn.to_v.bias", | |
| "transformer_blocks.11.attn.to_v.weight", | |
| "transformer_blocks.11.img_mod.1.bias", | |
| "transformer_blocks.11.img_mod.1.weight", | |
| "transformer_blocks.11.txt_mod.1.bias", | |
| "transformer_blocks.11.txt_mod.1.weight", | |
| "transformer_blocks.2.attn.add_k_proj.bias", | |
| "transformer_blocks.2.attn.add_k_proj.weight", | |
| "transformer_blocks.2.attn.add_q_proj.bias", | |
| "transformer_blocks.2.attn.add_q_proj.weight", | |
| "transformer_blocks.2.attn.add_v_proj.bias", | |
| "transformer_blocks.2.attn.add_v_proj.weight", | |
| "transformer_blocks.2.attn.norm_added_k.weight", | |
| "transformer_blocks.2.attn.norm_added_q.weight", | |
| "transformer_blocks.2.attn.norm_k.weight", | |
| "transformer_blocks.2.attn.norm_q.weight", | |
| "transformer_blocks.2.attn.to_add_out.bias", | |
| "transformer_blocks.2.attn.to_add_out.weight", | |
| "transformer_blocks.2.attn.to_k.bias", | |
| "transformer_blocks.2.attn.to_k.weight", | |
| "transformer_blocks.2.attn.to_out.0.bias", | |
| "transformer_blocks.2.attn.to_out.0.weight", | |
| "transformer_blocks.2.attn.to_q.bias", | |
| "transformer_blocks.2.attn.to_q.weight", | |
| "transformer_blocks.2.attn.to_v.bias", | |
| "transformer_blocks.2.attn.to_v.weight", | |
| "transformer_blocks.2.img_mod.1.bias", | |
| "transformer_blocks.2.img_mod.1.weight", | |
| "transformer_blocks.2.txt_mod.1.bias", | |
| "transformer_blocks.2.txt_mod.1.weight", | |
| "transformer_blocks.3.attn.add_k_proj.bias", | |
| "transformer_blocks.3.attn.add_k_proj.weight", | |
| "transformer_blocks.3.attn.add_q_proj.bias", | |
| "transformer_blocks.3.attn.add_q_proj.weight", | |
| "transformer_blocks.3.attn.add_v_proj.bias", | |
| "transformer_blocks.3.attn.add_v_proj.weight", | |
| "transformer_blocks.3.attn.norm_added_k.weight", | |
| "transformer_blocks.3.attn.norm_added_q.weight", | |
| "transformer_blocks.3.attn.norm_k.weight", | |
| "transformer_blocks.3.attn.norm_q.weight", | |
| "transformer_blocks.3.attn.to_add_out.bias", | |
| "transformer_blocks.3.attn.to_add_out.weight", | |
| "transformer_blocks.3.attn.to_k.bias", | |
| "transformer_blocks.3.attn.to_k.weight", | |
| "transformer_blocks.3.attn.to_out.0.bias", | |
| "transformer_blocks.3.attn.to_out.0.weight", | |
| "transformer_blocks.3.attn.to_q.bias", | |
| "transformer_blocks.3.attn.to_q.weight", | |
| "transformer_blocks.3.attn.to_v.bias", | |
| "transformer_blocks.3.attn.to_v.weight", | |
| "transformer_blocks.3.img_mod.1.bias", | |
| "transformer_blocks.3.img_mod.1.weight", | |
| "transformer_blocks.3.txt_mod.1.bias", | |
| "transformer_blocks.3.txt_mod.1.weight", | |
| "transformer_blocks.4.attn.add_k_proj.bias", | |
| "transformer_blocks.4.attn.add_k_proj.weight", | |
| "transformer_blocks.4.attn.add_q_proj.bias", | |
| "transformer_blocks.4.attn.add_q_proj.weight", | |
| "transformer_blocks.4.attn.add_v_proj.bias", | |
| "transformer_blocks.4.attn.add_v_proj.weight", | |
| "transformer_blocks.4.attn.norm_added_k.weight", | |
| "transformer_blocks.4.attn.norm_added_q.weight", | |
| "transformer_blocks.4.attn.norm_k.weight", | |
| "transformer_blocks.4.attn.norm_q.weight", | |
| "transformer_blocks.4.attn.to_add_out.bias", | |
| "transformer_blocks.4.attn.to_add_out.weight", | |
| "transformer_blocks.4.attn.to_k.bias", | |
| "transformer_blocks.4.attn.to_k.weight", | |
| "transformer_blocks.4.attn.to_out.0.bias", | |
| "transformer_blocks.4.attn.to_out.0.weight", | |
| "transformer_blocks.4.attn.to_q.bias", | |
| "transformer_blocks.4.attn.to_q.weight", | |
| "transformer_blocks.4.attn.to_v.bias", | |
| "transformer_blocks.4.attn.to_v.weight", | |
| "transformer_blocks.4.img_mod.1.bias", | |
| "transformer_blocks.4.img_mod.1.weight", | |
| "transformer_blocks.4.txt_mod.1.bias", | |
| "transformer_blocks.4.txt_mod.1.weight", | |
| "transformer_blocks.5.attn.add_k_proj.bias", | |
| "transformer_blocks.5.attn.add_k_proj.weight", | |
| "transformer_blocks.5.attn.add_q_proj.bias", | |
| "transformer_blocks.5.attn.add_q_proj.weight", | |
| "transformer_blocks.5.attn.add_v_proj.bias", | |
| "transformer_blocks.5.attn.add_v_proj.weight", | |
| "transformer_blocks.5.attn.norm_added_k.weight", | |
| "transformer_blocks.5.attn.norm_added_q.weight", | |
| "transformer_blocks.5.attn.norm_k.weight", | |
| "transformer_blocks.5.attn.norm_q.weight", | |
| "transformer_blocks.5.attn.to_add_out.bias", | |
| "transformer_blocks.5.attn.to_add_out.weight", | |
| "transformer_blocks.5.attn.to_k.bias", | |
| "transformer_blocks.5.attn.to_k.weight", | |
| "transformer_blocks.5.attn.to_out.0.bias", | |
| "transformer_blocks.5.attn.to_out.0.weight", | |
| "transformer_blocks.5.attn.to_q.bias", | |
| "transformer_blocks.5.attn.to_q.weight", | |
| "transformer_blocks.5.attn.to_v.bias", | |
| "transformer_blocks.5.attn.to_v.weight", | |
| "transformer_blocks.5.img_mod.1.bias", | |
| "transformer_blocks.5.img_mod.1.weight", | |
| "transformer_blocks.5.txt_mod.1.bias", | |
| "transformer_blocks.5.txt_mod.1.weight", | |
| "transformer_blocks.6.attn.add_k_proj.bias", | |
| "transformer_blocks.6.attn.add_k_proj.weight", | |
| "transformer_blocks.6.attn.add_q_proj.bias", | |
| "transformer_blocks.6.attn.add_q_proj.weight", | |
| "transformer_blocks.6.attn.add_v_proj.bias", | |
| "transformer_blocks.6.attn.add_v_proj.weight", | |
| "transformer_blocks.6.attn.norm_added_k.weight", | |
| "transformer_blocks.6.attn.norm_added_q.weight", | |
| "transformer_blocks.6.attn.norm_k.weight", | |
| "transformer_blocks.6.attn.norm_q.weight", | |
| "transformer_blocks.6.attn.to_add_out.bias", | |
| "transformer_blocks.6.attn.to_add_out.weight", | |
| "transformer_blocks.6.attn.to_k.bias", | |
| "transformer_blocks.6.attn.to_k.weight", | |
| "transformer_blocks.6.attn.to_out.0.bias", | |
| "transformer_blocks.6.attn.to_out.0.weight", | |
| "transformer_blocks.6.attn.to_q.bias", | |
| "transformer_blocks.6.attn.to_q.weight", | |
| "transformer_blocks.6.attn.to_v.bias", | |
| "transformer_blocks.6.attn.to_v.weight", | |
| "transformer_blocks.6.img_mod.1.bias", | |
| "transformer_blocks.6.img_mod.1.weight", | |
| "transformer_blocks.6.txt_mod.1.bias", | |
| "transformer_blocks.6.txt_mod.1.weight", | |
| "transformer_blocks.7.attn.add_k_proj.bias", | |
| "transformer_blocks.7.attn.add_k_proj.weight", | |
| "transformer_blocks.7.attn.add_q_proj.bias", | |
| "transformer_blocks.7.attn.add_q_proj.weight", | |
| "transformer_blocks.7.attn.add_v_proj.bias", | |
| "transformer_blocks.7.attn.add_v_proj.weight", | |
| "transformer_blocks.7.attn.norm_added_k.weight", | |
| "transformer_blocks.7.attn.norm_added_q.weight", | |
| "transformer_blocks.7.attn.norm_k.weight", | |
| "transformer_blocks.7.attn.norm_q.weight", | |
| "transformer_blocks.7.attn.to_add_out.bias", | |
| "transformer_blocks.7.attn.to_add_out.weight", | |
| "transformer_blocks.7.attn.to_k.bias", | |
| "transformer_blocks.7.attn.to_k.weight", | |
| "transformer_blocks.7.attn.to_out.0.bias", | |
| "transformer_blocks.7.attn.to_out.0.weight", | |
| "transformer_blocks.7.attn.to_q.bias", | |
| "transformer_blocks.7.attn.to_q.weight", | |
| "transformer_blocks.7.attn.to_v.bias", | |
| "transformer_blocks.7.attn.to_v.weight", | |
| "transformer_blocks.7.img_mod.1.bias", | |
| "transformer_blocks.7.img_mod.1.weight", | |
| "transformer_blocks.7.txt_mod.1.bias", | |
| "transformer_blocks.7.txt_mod.1.weight", | |
| "transformer_blocks.8.attn.add_k_proj.bias", | |
| "transformer_blocks.8.attn.add_k_proj.weight", | |
| "transformer_blocks.8.attn.add_q_proj.bias", | |
| "transformer_blocks.8.attn.add_q_proj.weight", | |
| "transformer_blocks.8.attn.add_v_proj.bias", | |
| "transformer_blocks.8.attn.add_v_proj.weight", | |
| "transformer_blocks.8.attn.norm_added_k.weight", | |
| "transformer_blocks.8.attn.norm_added_q.weight", | |
| "transformer_blocks.8.attn.norm_k.weight", | |
| "transformer_blocks.8.attn.norm_q.weight", | |
| "transformer_blocks.8.attn.to_add_out.bias", | |
| "transformer_blocks.8.attn.to_add_out.weight", | |
| "transformer_blocks.8.attn.to_k.bias", | |
| "transformer_blocks.8.attn.to_k.weight", | |
| "transformer_blocks.8.attn.to_out.0.bias", | |
| "transformer_blocks.8.attn.to_out.0.weight", | |
| "transformer_blocks.8.attn.to_q.bias", | |
| "transformer_blocks.8.attn.to_q.weight", | |
| "transformer_blocks.8.attn.to_v.bias", | |
| "transformer_blocks.8.attn.to_v.weight", | |
| "transformer_blocks.8.img_mod.1.bias", | |
| "transformer_blocks.8.img_mod.1.weight", | |
| "transformer_blocks.8.txt_mod.1.bias", | |
| "transformer_blocks.8.txt_mod.1.weight", | |
| "transformer_blocks.9.attn.add_k_proj.bias", | |
| "transformer_blocks.9.attn.add_k_proj.weight", | |
| "transformer_blocks.9.attn.add_q_proj.bias", | |
| "transformer_blocks.9.attn.add_q_proj.weight", | |
| "transformer_blocks.9.attn.add_v_proj.bias", | |
| "transformer_blocks.9.attn.add_v_proj.weight", | |
| "transformer_blocks.9.attn.norm_added_k.weight", | |
| "transformer_blocks.9.attn.norm_added_q.weight", | |
| "transformer_blocks.9.attn.norm_k.weight", | |
| "transformer_blocks.9.attn.norm_q.weight", | |
| "transformer_blocks.9.attn.to_add_out.bias", | |
| "transformer_blocks.9.attn.to_add_out.weight", | |
| "transformer_blocks.9.attn.to_k.bias", | |
| "transformer_blocks.9.attn.to_k.weight", | |
| "transformer_blocks.9.attn.to_out.0.bias", | |
| "transformer_blocks.9.attn.to_out.0.weight", | |
| "transformer_blocks.9.attn.to_q.bias", | |
| "transformer_blocks.9.attn.to_q.weight", | |
| "transformer_blocks.9.attn.to_v.bias", | |
| "transformer_blocks.9.attn.to_v.weight", | |
| "transformer_blocks.9.img_mod.1.bias", | |
| "transformer_blocks.9.img_mod.1.weight", | |
| "transformer_blocks.9.txt_mod.1.bias", | |
| "transformer_blocks.9.txt_mod.1.weight", | |
| "txt_in.bias", | |
| "txt_in.weight", | |
| "txt_norm.weight" | |
| ] | |
| } | |