Instructions to use ProCreations/Image-2.1-Calibrated-FP8 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use ProCreations/Image-2.1-Calibrated-FP8 with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("ProCreations/Image-2.1-Calibrated-FP8", 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
Release calibrated Image2.1 FP8 transformer, native SM120 runtime, quality evidence and real-time demo
f642d1f verified Download transformer/quantization_config.json from ProCreations/Image-2.1-Calibrated-FP8: direct link, hf CLI and curl.
- Browser
- Download file 36.9 kB
-
https://huggingface.co/ProCreations/Image-2.1-Calibrated-FP8/resolve/main/transformer/quantization_config.json
- Command line
-
hf download hf://ProCreations/Image-2.1-Calibrated-FP8/transformer/quantization_config.json
-
curl -L -o quantization_config.json https://huggingface.co/ProCreations/Image-2.1-Calibrated-FP8/resolve/main/transformer/quantization_config.json
36.9 kB
| { | |
| "format": "procreations-calibrated-fp8-v1", | |
| "base_model": "Qwen/Qwen-Image-2.1", | |
| "base_revision": "b3179ad355be050328e483a9dfdd9e60cd62adfa", | |
| "weight_dtype": "float8_e4m3fn", | |
| "activation_dtype": "float8_e4m3fn", | |
| "weight_scaling": "per-output-channel FP32", | |
| "activation_scaling": "dynamic per-token FP32", | |
| "accumulation": "FP32, use_fast_accum=False", | |
| "output_dtype": "bfloat16", | |
| "kernel": "torch._scaled_mm native CUTLASS SM120, Triton fused smoothing and activation quantization", | |
| "quantized_modules": { | |
| "transformer_blocks.0.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.0.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.0.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.0.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.25, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.0.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.0.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.0.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.1.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.1.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.1.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.1.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.1.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.1.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.1.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.2.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.2.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.2.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.2.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.2.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.2.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.2.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.3.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.3.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.3.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.3.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.3.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.3.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.3.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.4.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.4.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.4.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.4.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.4.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.4.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.4.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.5.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.5.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.5.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.5.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.5.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.5.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.5.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.6.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.6.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.6.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.6.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.6.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.6.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.6.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.7.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.7.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.7.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.7.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.7.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.7.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.7.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.8.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.8.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.8.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.8.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.8.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.8.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.8.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.9.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.9.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.9.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.9.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.9.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.9.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.9.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.10.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.10.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.10.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.10.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.10.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.10.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.10.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.11.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.11.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.11.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.11.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.11.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.11.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.11.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.12.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.12.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.12.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.12.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.12.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.12.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.12.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.13.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.13.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.13.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.13.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.13.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.13.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.13.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.14.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.14.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.14.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.14.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.14.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.14.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.14.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.15.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.15.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.15.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.15.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.15.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.15.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.15.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.16.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.16.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.16.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.16.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.16.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.16.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.16.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.17.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.17.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.17.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.17.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.17.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.17.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.17.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.18.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.18.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.18.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.18.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.18.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.18.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.18.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.19.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.19.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.19.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.19.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.19.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.25, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.19.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.19.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.25, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.20.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.20.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.20.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.20.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.20.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.20.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.20.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.21.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.21.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.21.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.21.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.21.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.21.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.21.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.22.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.22.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.22.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.22.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.22.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.22.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.22.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.23.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.23.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.23.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.23.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.23.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.23.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.23.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.24.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.24.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.24.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.24.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.24.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.24.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.24.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.25.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.25.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.25.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.25.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.25.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.25.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.25.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.26.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.26.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.26.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.26.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.26.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.26.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.26.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.27.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.27.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.27.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.27.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.27.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.27.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.27.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.28.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.28.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.28.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.28.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.28.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.28.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.28.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.29.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.29.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.29.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.29.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.29.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.29.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.29.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.30.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.30.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.30.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.30.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.30.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.30.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.30.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.31.attn.to_q": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.31.attn.to_k": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.31.attn.to_v": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.995 | |
| }, | |
| "transformer_blocks.31.attn.to_out.0": { | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "alpha": 1.0, | |
| "weight_amax_clip_factor": 0.99 | |
| }, | |
| "transformer_blocks.31.img_mlp.proj": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.31.img_mlp.out": { | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "alpha": 0.5, | |
| "weight_amax_clip_factor": 1.0 | |
| }, | |
| "transformer_blocks.31.img_mlp.gate_layer": { | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "alpha": 0.75, | |
| "weight_amax_clip_factor": 1.0 | |
| } | |
| }, | |
| "kept_bf16_modules": [], | |
| "modified_notice": "Built with Qwen. ProCreations modified transformer weights through calibrated FP8 quantization on 2026-09-20." | |
| } |