| { |
| "activation_backends": { |
| "triton_cuda_packed_w4": 300 |
| }, |
| "audio_tokens": 207, |
| "capability": [ |
| 12, |
| 0 |
| ], |
| "cuda": "12.8", |
| "cuda_after_load_bytes": 18328776192, |
| "cuda_forward_ms": 2865.6015625, |
| "cuda_peak_bytes": 46381307904, |
| "effective_runtime_modes": { |
| "native_packed_matmul": 300 |
| }, |
| "gpu": "NVIDIA RTX PRO 6000 Blackwell Server Edition", |
| "h3_dtype_views": 50, |
| "load_dtype": "default", |
| "load_seconds": 0.24018437741324306, |
| "model_dtype": "torch.bfloat16", |
| "move_to_cuda_seconds": 1.1501922300085425, |
| "num_blocks": 50, |
| "orbitquant_linear_count": 300, |
| "orbitquant_linear_cuda_ms": 1077.5865263938904, |
| "orbitquant_linear_inputs": { |
| "transformer_blocks.0.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.0.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.0.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.0.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.0.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.0.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.1.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.1.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.1.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.1.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.1.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.1.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.10.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.10.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.10.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.10.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.10.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.10.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.11.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.11.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.11.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.11.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.11.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.11.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.12.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.12.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.12.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.12.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.12.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.12.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.13.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.13.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.13.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.13.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.13.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.13.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.14.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.14.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.14.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.14.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.14.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.14.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.15.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.15.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.15.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.15.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.15.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.15.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.16.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.16.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.16.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.16.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.16.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.16.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.17.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.17.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.17.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.17.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.17.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.17.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.18.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.18.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.18.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.18.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.18.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.18.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.19.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.19.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.19.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.19.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.19.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.19.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.2.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.2.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.2.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.2.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.2.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.2.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.20.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.20.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.20.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.20.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.20.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.20.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.21.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.21.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.21.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.21.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.21.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.21.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.22.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.22.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.22.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.22.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.22.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.22.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.23.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.23.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.23.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.23.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.23.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.23.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.24.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.24.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.24.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.24.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.24.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.24.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.25.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.25.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.25.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.25.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.25.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.25.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.26.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.26.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.26.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.26.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.26.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.26.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.27.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.27.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.27.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.27.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.27.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.27.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.28.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.28.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.28.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.28.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.28.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.28.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.29.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.29.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.29.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.29.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.29.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.29.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.3.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.3.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.3.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.3.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.3.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.3.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.30.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.30.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.30.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.30.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.30.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.30.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.31.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.31.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.31.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.31.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.31.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.31.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.32.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.32.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.32.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.32.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.32.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.32.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.33.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.33.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.33.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.33.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.33.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.33.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.34.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.34.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.34.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.34.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.34.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.34.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.35.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.35.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.35.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.35.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.35.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.35.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.36.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.36.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.36.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.36.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.36.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.36.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.37.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.37.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.37.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.37.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.37.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.37.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.38.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.38.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.38.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.38.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.38.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.38.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.39.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.39.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.39.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.39.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.39.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.39.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.4.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.4.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.4.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.4.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.4.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.4.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.40.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.40.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.40.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.40.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.40.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.40.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.41.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.41.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.41.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.41.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.41.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.41.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.42.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.42.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.42.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.42.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.42.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.42.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.43.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.43.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.43.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.43.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.43.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.43.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.44.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.44.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.44.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.44.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.44.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.44.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.45.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.45.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.45.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.45.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.45.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.45.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.46.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.46.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.46.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.46.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.46.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.46.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.47.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.47.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.47.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.47.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.47.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.47.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.48.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.48.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.48.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.48.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.48.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.48.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.49.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.49.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.49.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.49.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.49.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.49.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.5.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.5.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.5.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.5.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.5.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.5.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.6.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.6.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.6.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.6.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.6.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.6.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.7.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.7.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.7.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.7.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.7.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.7.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.8.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.8.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.8.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.8.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.8.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.8.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| }, |
| "transformer_blocks.9.attn.to_k": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.9.attn.to_out.0": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 7168 |
| ], |
| "stride": [ |
| 80740352, |
| 7168, |
| 1 |
| ] |
| }, |
| "transformer_blocks.9.attn.to_q": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.9.attn.to_v": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.9.ff.net.0.proj": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 5376 |
| ], |
| "stride": [ |
| 60555264, |
| 5376, |
| 1 |
| ] |
| }, |
| "transformer_blocks.9.ff.net.2": { |
| "contiguous": true, |
| "dtype": "torch.bfloat16", |
| "shape": [ |
| 1, |
| 11264, |
| 14336 |
| ], |
| "stride": [ |
| 161480704, |
| 14336, |
| 1 |
| ] |
| } |
| }, |
| "orbitquant_shape_cuda_ms": { |
| "14336x5376": { |
| "calls": 50, |
| "total_ms": 230.40703773498535 |
| }, |
| "5376x28672": { |
| "calls": 50, |
| "total_ms": 354.5018243789673 |
| }, |
| "5376x7168": { |
| "calls": 150, |
| "total_ms": 368.4010241031647 |
| }, |
| "7168x5376": { |
| "calls": 50, |
| "total_ms": 124.27664017677307 |
| } |
| }, |
| "orbitquant_shapes": { |
| "14336x5376": 50, |
| "5376x28672": 50, |
| "5376x7168": 150, |
| "7168x5376": 50 |
| }, |
| "output_finite": [ |
| true, |
| true |
| ], |
| "output_shapes": [ |
| [ |
| 1, |
| 10545, |
| 96 |
| ], |
| [ |
| 1, |
| 207, |
| 32 |
| ] |
| ], |
| "runtime_mode": "native_packed_matmul", |
| "slowest_orbitquant_linears": [ |
| { |
| "cuda_ms": 18.62041664123535, |
| "name": "transformer_blocks.0.attn.to_q", |
| "shape": "5376x7168" |
| }, |
| { |
| "cuda_ms": 9.111200332641602, |
| "name": "transformer_blocks.0.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 8.695232391357422, |
| "name": "transformer_blocks.0.ff.net.2", |
| "shape": "14336x5376" |
| }, |
| { |
| "cuda_ms": 7.266304016113281, |
| "name": "transformer_blocks.0.attn.to_out.0", |
| "shape": "7168x5376" |
| }, |
| { |
| "cuda_ms": 7.081600189208984, |
| "name": "transformer_blocks.6.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 7.080512046813965, |
| "name": "transformer_blocks.27.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 7.074463844299316, |
| "name": "transformer_blocks.25.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 7.072256088256836, |
| "name": "transformer_blocks.10.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 7.0702080726623535, |
| "name": "transformer_blocks.9.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 7.0693440437316895, |
| "name": "transformer_blocks.4.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 7.068672180175781, |
| "name": "transformer_blocks.2.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 7.068160057067871, |
| "name": "transformer_blocks.11.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 7.067647933959961, |
| "name": "transformer_blocks.37.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 7.063136100769043, |
| "name": "transformer_blocks.24.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 7.063039779663086, |
| "name": "transformer_blocks.29.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 7.062047958374023, |
| "name": "transformer_blocks.32.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 7.062016010284424, |
| "name": "transformer_blocks.20.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 7.061503887176514, |
| "name": "transformer_blocks.15.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 7.061503887176514, |
| "name": "transformer_blocks.16.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 7.061024188995361, |
| "name": "transformer_blocks.22.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 7.059103965759277, |
| "name": "transformer_blocks.23.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 7.058047771453857, |
| "name": "transformer_blocks.47.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 7.056896209716797, |
| "name": "transformer_blocks.33.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 7.055935859680176, |
| "name": "transformer_blocks.40.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 7.054336071014404, |
| "name": "transformer_blocks.17.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 7.053952217102051, |
| "name": "transformer_blocks.39.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 7.053855895996094, |
| "name": "transformer_blocks.28.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 7.053311824798584, |
| "name": "transformer_blocks.3.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 7.0502400398254395, |
| "name": "transformer_blocks.1.ff.net.0.proj", |
| "shape": "5376x28672" |
| }, |
| { |
| "cuda_ms": 7.049215793609619, |
| "name": "transformer_blocks.5.ff.net.0.proj", |
| "shape": "5376x28672" |
| } |
| ], |
| "status": "pass", |
| "text_tokens": 512, |
| "torch": "2.8.0+cu128", |
| "transformer": "/workspace/orbitquant-h3-perf/model/transformer", |
| "transformer_block_cuda_ms": [ |
| { |
| "cuda_ms": 310.5533752441406, |
| "name": "transformer_blocks.0" |
| }, |
| { |
| "cuda_ms": 48.46879959106445, |
| "name": "transformer_blocks.1" |
| }, |
| { |
| "cuda_ms": 48.629886627197266, |
| "name": "transformer_blocks.2" |
| }, |
| { |
| "cuda_ms": 48.53862380981445, |
| "name": "transformer_blocks.3" |
| }, |
| { |
| "cuda_ms": 48.653438568115234, |
| "name": "transformer_blocks.4" |
| }, |
| { |
| "cuda_ms": 48.58992004394531, |
| "name": "transformer_blocks.5" |
| }, |
| { |
| "cuda_ms": 48.66368103027344, |
| "name": "transformer_blocks.6" |
| }, |
| { |
| "cuda_ms": 48.59689712524414, |
| "name": "transformer_blocks.7" |
| }, |
| { |
| "cuda_ms": 48.5167350769043, |
| "name": "transformer_blocks.8" |
| }, |
| { |
| "cuda_ms": 48.62108612060547, |
| "name": "transformer_blocks.9" |
| }, |
| { |
| "cuda_ms": 48.51871871948242, |
| "name": "transformer_blocks.10" |
| }, |
| { |
| "cuda_ms": 48.5648307800293, |
| "name": "transformer_blocks.11" |
| }, |
| { |
| "cuda_ms": 48.545310974121094, |
| "name": "transformer_blocks.12" |
| }, |
| { |
| "cuda_ms": 48.61705780029297, |
| "name": "transformer_blocks.13" |
| }, |
| { |
| "cuda_ms": 48.762943267822266, |
| "name": "transformer_blocks.14" |
| }, |
| { |
| "cuda_ms": 48.692222595214844, |
| "name": "transformer_blocks.15" |
| }, |
| { |
| "cuda_ms": 48.60326385498047, |
| "name": "transformer_blocks.16" |
| }, |
| { |
| "cuda_ms": 48.44297790527344, |
| "name": "transformer_blocks.17" |
| }, |
| { |
| "cuda_ms": 48.3702392578125, |
| "name": "transformer_blocks.18" |
| }, |
| { |
| "cuda_ms": 48.5167350769043, |
| "name": "transformer_blocks.19" |
| }, |
| { |
| "cuda_ms": 48.67536163330078, |
| "name": "transformer_blocks.20" |
| }, |
| { |
| "cuda_ms": 48.69500732421875, |
| "name": "transformer_blocks.21" |
| }, |
| { |
| "cuda_ms": 48.48076629638672, |
| "name": "transformer_blocks.22" |
| }, |
| { |
| "cuda_ms": 48.34979248046875, |
| "name": "transformer_blocks.23" |
| }, |
| { |
| "cuda_ms": 48.50944137573242, |
| "name": "transformer_blocks.24" |
| }, |
| { |
| "cuda_ms": 48.56480026245117, |
| "name": "transformer_blocks.25" |
| }, |
| { |
| "cuda_ms": 48.68355178833008, |
| "name": "transformer_blocks.26" |
| }, |
| { |
| "cuda_ms": 48.3375358581543, |
| "name": "transformer_blocks.27" |
| }, |
| { |
| "cuda_ms": 48.75628662109375, |
| "name": "transformer_blocks.28" |
| }, |
| { |
| "cuda_ms": 48.42652893066406, |
| "name": "transformer_blocks.29" |
| }, |
| { |
| "cuda_ms": 48.335487365722656, |
| "name": "transformer_blocks.30" |
| }, |
| { |
| "cuda_ms": 48.421470642089844, |
| "name": "transformer_blocks.31" |
| }, |
| { |
| "cuda_ms": 48.22582244873047, |
| "name": "transformer_blocks.32" |
| }, |
| { |
| "cuda_ms": 48.533119201660156, |
| "name": "transformer_blocks.33" |
| }, |
| { |
| "cuda_ms": 48.40399932861328, |
| "name": "transformer_blocks.34" |
| }, |
| { |
| "cuda_ms": 48.340030670166016, |
| "name": "transformer_blocks.35" |
| }, |
| { |
| "cuda_ms": 48.41379165649414, |
| "name": "transformer_blocks.36" |
| }, |
| { |
| "cuda_ms": 48.17622375488281, |
| "name": "transformer_blocks.37" |
| }, |
| { |
| "cuda_ms": 48.51875305175781, |
| "name": "transformer_blocks.38" |
| }, |
| { |
| "cuda_ms": 48.15116882324219, |
| "name": "transformer_blocks.39" |
| }, |
| { |
| "cuda_ms": 48.514591217041016, |
| "name": "transformer_blocks.40" |
| }, |
| { |
| "cuda_ms": 48.00166320800781, |
| "name": "transformer_blocks.41" |
| }, |
| { |
| "cuda_ms": 48.32204818725586, |
| "name": "transformer_blocks.42" |
| }, |
| { |
| "cuda_ms": 48.416385650634766, |
| "name": "transformer_blocks.43" |
| }, |
| { |
| "cuda_ms": 47.402591705322266, |
| "name": "transformer_blocks.44" |
| }, |
| { |
| "cuda_ms": 48.48284912109375, |
| "name": "transformer_blocks.45" |
| }, |
| { |
| "cuda_ms": 48.11532974243164, |
| "name": "transformer_blocks.46" |
| }, |
| { |
| "cuda_ms": 48.25551986694336, |
| "name": "transformer_blocks.47" |
| }, |
| { |
| "cuda_ms": 48.40300750732422, |
| "name": "transformer_blocks.48" |
| }, |
| { |
| "cuda_ms": 47.34415817260742, |
| "name": "transformer_blocks.49" |
| } |
| ], |
| "video_tokens": 10545, |
| "wall_seconds": 2.8667881772853434 |
| } |
|
|
