| { |
| "schema": "trellis-webgpu-pack", |
| "version": 1, |
| "stage": "idolGaussianDecoder", |
| "upstreamRevision": "2731c2291fc65a797c5cb627a5e5526fc1490902", |
| "shards": [ |
| { |
| "file": "stage-000-e94697a18f508808d8461833475372b66f5b9275fbdee03706d3335769411595.bin", |
| "bytes": 63275264, |
| "sha256": "e94697a18f508808d8461833475372b66f5b9275fbdee03706d3335769411595" |
| }, |
| { |
| "file": "stage-001-4f2f2bb809b2a06ad6fad994acebb1ebf0b322e548a992bd6970829237fdb659.bin", |
| "bytes": 65349632, |
| "sha256": "4f2f2bb809b2a06ad6fad994acebb1ebf0b322e548a992bd6970829237fdb659" |
| }, |
| { |
| "file": "stage-002-3006c7eecf44957904ebac648eeb94c3bcac8107c0073a44a533b7c3e9f55136.bin", |
| "bytes": 66453504, |
| "sha256": "3006c7eecf44957904ebac648eeb94c3bcac8107c0073a44a533b7c3e9f55136" |
| }, |
| { |
| "file": "stage-003-9011e1272e3d48935898ad6d1b03dddcada31ee8677ea591ce65f7931b97cc88.bin", |
| "bytes": 55193984, |
| "sha256": "9011e1272e3d48935898ad6d1b03dddcada31ee8677ea591ce65f7931b97cc88" |
| } |
| ], |
| "tensors": { |
| "base_offset_scale": { |
| "shape": [ |
| 1 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 0, |
| "bytes": 4 |
| } |
| }, |
| "blocks.0.cross_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 256, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.0.cross_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 4352, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.0.cross_attn.to_kv.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 8448, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.0.cross_attn.to_kv.weight": { |
| "shape": [ |
| 2048, |
| 16 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 12544, |
| "bytes": 65536 |
| } |
| }, |
| "blocks.0.cross_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 78080, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.0.cross_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 0, |
| "offset": 80128, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 1128704, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.0.cross_attn.to_q.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 1161472, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.0.cross_attn.to_q.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 0, |
| "offset": 1163520, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 2212096, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.0.mlp.mlp.0.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 2244864, |
| "bytes": 8192 |
| } |
| }, |
| "blocks.0.mlp.mlp.0.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4096, |
| "columns": 1024, |
| "padded_rows": 4096, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 0, |
| "offset": 2253056, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 6447360, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.0.mlp.mlp.2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 6578432, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.0.mlp.mlp.2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 4096, |
| "padded_rows": 1024, |
| "padded_columns": 4096, |
| "data": { |
| "shard": 0, |
| "offset": 6580480, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 10774784, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.0.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 10905856, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.0.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 10909952, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.0.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 10914048, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.0.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 10918144, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.0.self_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 10922240, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.0.self_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 0, |
| "offset": 10924288, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 11972864, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.0.self_attn.to_qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 12005632, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.0.self_attn.to_qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1024, |
| "padded_rows": 3072, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 0, |
| "offset": 12011776, |
| "bytes": 3145728 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 15157504, |
| "bytes": 98304 |
| } |
| }, |
| "blocks.1.cross_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 15255808, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.1.cross_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 15259904, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.1.cross_attn.to_kv.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 15264000, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.1.cross_attn.to_kv.weight": { |
| "shape": [ |
| 2048, |
| 16 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 15268096, |
| "bytes": 65536 |
| } |
| }, |
| "blocks.1.cross_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 15333632, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.1.cross_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 0, |
| "offset": 15335680, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 16384256, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.1.cross_attn.to_q.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 16417024, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.1.cross_attn.to_q.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 0, |
| "offset": 16419072, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 17467648, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.1.mlp.mlp.0.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 17500416, |
| "bytes": 8192 |
| } |
| }, |
| "blocks.1.mlp.mlp.0.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4096, |
| "columns": 1024, |
| "padded_rows": 4096, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 0, |
| "offset": 17508608, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 21702912, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.1.mlp.mlp.2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 21833984, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.1.mlp.mlp.2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 4096, |
| "padded_rows": 1024, |
| "padded_columns": 4096, |
| "data": { |
| "shard": 0, |
| "offset": 21836032, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 26030336, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.1.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 26161408, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.1.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 26165504, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.1.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 26169600, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.1.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 26173696, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.1.self_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 26177792, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.1.self_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 0, |
| "offset": 26179840, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 27228416, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.1.self_attn.to_qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 27261184, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.1.self_attn.to_qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1024, |
| "padded_rows": 3072, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 0, |
| "offset": 27267328, |
| "bytes": 3145728 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 30413056, |
| "bytes": 98304 |
| } |
| }, |
| "blocks.10.cross_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 30511360, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.10.cross_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 30515456, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.10.cross_attn.to_kv.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 30519552, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.10.cross_attn.to_kv.weight": { |
| "shape": [ |
| 2048, |
| 16 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 30523648, |
| "bytes": 65536 |
| } |
| }, |
| "blocks.10.cross_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 30589184, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.10.cross_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 0, |
| "offset": 30591232, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 31639808, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.10.cross_attn.to_q.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 31672576, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.10.cross_attn.to_q.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 0, |
| "offset": 31674624, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 32723200, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.10.mlp.mlp.0.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 32755968, |
| "bytes": 8192 |
| } |
| }, |
| "blocks.10.mlp.mlp.0.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4096, |
| "columns": 1024, |
| "padded_rows": 4096, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 0, |
| "offset": 32764160, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 36958464, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.10.mlp.mlp.2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 37089536, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.10.mlp.mlp.2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 4096, |
| "padded_rows": 1024, |
| "padded_columns": 4096, |
| "data": { |
| "shard": 0, |
| "offset": 37091584, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 41285888, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.10.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 41416960, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.10.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 41421056, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.10.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 41425152, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.10.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 41429248, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.10.self_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 41433344, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.10.self_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 0, |
| "offset": 41435392, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 42483968, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.10.self_attn.to_qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 42516736, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.10.self_attn.to_qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1024, |
| "padded_rows": 3072, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 0, |
| "offset": 42522880, |
| "bytes": 3145728 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 45668608, |
| "bytes": 98304 |
| } |
| }, |
| "blocks.11.cross_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 45766912, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.11.cross_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 45771008, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.11.cross_attn.to_kv.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 45775104, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.11.cross_attn.to_kv.weight": { |
| "shape": [ |
| 2048, |
| 16 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 45779200, |
| "bytes": 65536 |
| } |
| }, |
| "blocks.11.cross_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 45844736, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.11.cross_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 0, |
| "offset": 45846784, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 46895360, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.11.cross_attn.to_q.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 46928128, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.11.cross_attn.to_q.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 0, |
| "offset": 46930176, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 47978752, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.11.mlp.mlp.0.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 48011520, |
| "bytes": 8192 |
| } |
| }, |
| "blocks.11.mlp.mlp.0.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4096, |
| "columns": 1024, |
| "padded_rows": 4096, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 0, |
| "offset": 48019712, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 52214016, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.11.mlp.mlp.2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 52345088, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.11.mlp.mlp.2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 4096, |
| "padded_rows": 1024, |
| "padded_columns": 4096, |
| "data": { |
| "shard": 0, |
| "offset": 52347136, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 56541440, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.11.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 56672512, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.11.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 56676608, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.11.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 56680704, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.11.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 56684800, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.11.self_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 56688896, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.11.self_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 0, |
| "offset": 56690944, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 57739520, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.11.self_attn.to_qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 57772288, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.11.self_attn.to_qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1024, |
| "padded_rows": 3072, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 0, |
| "offset": 57778432, |
| "bytes": 3145728 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 60924160, |
| "bytes": 98304 |
| } |
| }, |
| "blocks.12.cross_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 61022464, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.12.cross_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 61026560, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.12.cross_attn.to_kv.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 61030656, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.12.cross_attn.to_kv.weight": { |
| "shape": [ |
| 2048, |
| 16 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 61034752, |
| "bytes": 65536 |
| } |
| }, |
| "blocks.12.cross_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 61100288, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.12.cross_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 0, |
| "offset": 61102336, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 62150912, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.12.cross_attn.to_q.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 62183680, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.12.cross_attn.to_q.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 0, |
| "offset": 62185728, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 63234304, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.12.mlp.mlp.0.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 63267072, |
| "bytes": 8192 |
| } |
| }, |
| "blocks.12.mlp.mlp.0.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4096, |
| "columns": 1024, |
| "padded_rows": 4096, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 1, |
| "offset": 0, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 4194304, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.12.mlp.mlp.2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 4325376, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.12.mlp.mlp.2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 4096, |
| "padded_rows": 1024, |
| "padded_columns": 4096, |
| "data": { |
| "shard": 1, |
| "offset": 4327424, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 8521728, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.12.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 8652800, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.12.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 8656896, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.12.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 8660992, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.12.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 8665088, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.12.self_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 8669184, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.12.self_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 1, |
| "offset": 8671232, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 9719808, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.12.self_attn.to_qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 9752576, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.12.self_attn.to_qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1024, |
| "padded_rows": 3072, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 1, |
| "offset": 9758720, |
| "bytes": 3145728 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 12904448, |
| "bytes": 98304 |
| } |
| }, |
| "blocks.13.cross_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 13002752, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.13.cross_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 13006848, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.13.cross_attn.to_kv.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 13010944, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.13.cross_attn.to_kv.weight": { |
| "shape": [ |
| 2048, |
| 16 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 13015040, |
| "bytes": 65536 |
| } |
| }, |
| "blocks.13.cross_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 13080576, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.13.cross_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 1, |
| "offset": 13082624, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 14131200, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.13.cross_attn.to_q.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 14163968, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.13.cross_attn.to_q.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 1, |
| "offset": 14166016, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 15214592, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.13.mlp.mlp.0.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 15247360, |
| "bytes": 8192 |
| } |
| }, |
| "blocks.13.mlp.mlp.0.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4096, |
| "columns": 1024, |
| "padded_rows": 4096, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 1, |
| "offset": 15255552, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 19449856, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.13.mlp.mlp.2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 19580928, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.13.mlp.mlp.2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 4096, |
| "padded_rows": 1024, |
| "padded_columns": 4096, |
| "data": { |
| "shard": 1, |
| "offset": 19582976, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 23777280, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.13.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 23908352, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.13.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 23912448, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.13.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 23916544, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.13.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 23920640, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.13.self_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 23924736, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.13.self_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 1, |
| "offset": 23926784, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 24975360, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.13.self_attn.to_qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 25008128, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.13.self_attn.to_qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1024, |
| "padded_rows": 3072, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 1, |
| "offset": 25014272, |
| "bytes": 3145728 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 28160000, |
| "bytes": 98304 |
| } |
| }, |
| "blocks.14.cross_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 28258304, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.14.cross_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 28262400, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.14.cross_attn.to_kv.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 28266496, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.14.cross_attn.to_kv.weight": { |
| "shape": [ |
| 2048, |
| 16 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 28270592, |
| "bytes": 65536 |
| } |
| }, |
| "blocks.14.cross_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 28336128, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.14.cross_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 1, |
| "offset": 28338176, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 29386752, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.14.cross_attn.to_q.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 29419520, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.14.cross_attn.to_q.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 1, |
| "offset": 29421568, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 30470144, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.14.mlp.mlp.0.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 30502912, |
| "bytes": 8192 |
| } |
| }, |
| "blocks.14.mlp.mlp.0.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4096, |
| "columns": 1024, |
| "padded_rows": 4096, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 1, |
| "offset": 30511104, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 34705408, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.14.mlp.mlp.2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 34836480, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.14.mlp.mlp.2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 4096, |
| "padded_rows": 1024, |
| "padded_columns": 4096, |
| "data": { |
| "shard": 1, |
| "offset": 34838528, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 39032832, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.14.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 39163904, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.14.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 39168000, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.14.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 39172096, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.14.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 39176192, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.14.self_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 39180288, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.14.self_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 1, |
| "offset": 39182336, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 40230912, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.14.self_attn.to_qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 40263680, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.14.self_attn.to_qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1024, |
| "padded_rows": 3072, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 1, |
| "offset": 40269824, |
| "bytes": 3145728 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 43415552, |
| "bytes": 98304 |
| } |
| }, |
| "blocks.15.cross_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 43513856, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.15.cross_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 43517952, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.15.cross_attn.to_kv.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 43522048, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.15.cross_attn.to_kv.weight": { |
| "shape": [ |
| 2048, |
| 16 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 43526144, |
| "bytes": 65536 |
| } |
| }, |
| "blocks.15.cross_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 43591680, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.15.cross_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 1, |
| "offset": 43593728, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 44642304, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.15.cross_attn.to_q.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 44675072, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.15.cross_attn.to_q.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 1, |
| "offset": 44677120, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 45725696, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.15.mlp.mlp.0.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 45758464, |
| "bytes": 8192 |
| } |
| }, |
| "blocks.15.mlp.mlp.0.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4096, |
| "columns": 1024, |
| "padded_rows": 4096, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 1, |
| "offset": 45766656, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 49960960, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.15.mlp.mlp.2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 50092032, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.15.mlp.mlp.2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 4096, |
| "padded_rows": 1024, |
| "padded_columns": 4096, |
| "data": { |
| "shard": 1, |
| "offset": 50094080, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 54288384, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.15.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 54419456, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.15.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 54423552, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.15.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 54427648, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.15.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 54431744, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.15.self_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 54435840, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.15.self_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 1, |
| "offset": 54437888, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 55486464, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.15.self_attn.to_qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 55519232, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.15.self_attn.to_qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1024, |
| "padded_rows": 3072, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 1, |
| "offset": 55525376, |
| "bytes": 3145728 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 58671104, |
| "bytes": 98304 |
| } |
| }, |
| "blocks.2.cross_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 58769408, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.2.cross_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 58773504, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.2.cross_attn.to_kv.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 58777600, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.2.cross_attn.to_kv.weight": { |
| "shape": [ |
| 2048, |
| 16 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 58781696, |
| "bytes": 65536 |
| } |
| }, |
| "blocks.2.cross_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 58847232, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.2.cross_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 1, |
| "offset": 58849280, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 59897856, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.2.cross_attn.to_q.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 59930624, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.2.cross_attn.to_q.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 1, |
| "offset": 59932672, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 60981248, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.2.mlp.mlp.0.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 61014016, |
| "bytes": 8192 |
| } |
| }, |
| "blocks.2.mlp.mlp.0.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4096, |
| "columns": 1024, |
| "padded_rows": 4096, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 1, |
| "offset": 61022208, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 65216512, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.2.mlp.mlp.2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 65347584, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.2.mlp.mlp.2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 4096, |
| "padded_rows": 1024, |
| "padded_columns": 4096, |
| "data": { |
| "shard": 2, |
| "offset": 0, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 4194304, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.2.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 4325376, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.2.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 4329472, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.2.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 4333568, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.2.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 4337664, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.2.self_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 4341760, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.2.self_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 2, |
| "offset": 4343808, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 5392384, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.2.self_attn.to_qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 5425152, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.2.self_attn.to_qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1024, |
| "padded_rows": 3072, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 2, |
| "offset": 5431296, |
| "bytes": 3145728 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 8577024, |
| "bytes": 98304 |
| } |
| }, |
| "blocks.3.cross_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 8675328, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.3.cross_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 8679424, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.3.cross_attn.to_kv.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 8683520, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.3.cross_attn.to_kv.weight": { |
| "shape": [ |
| 2048, |
| 16 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 8687616, |
| "bytes": 65536 |
| } |
| }, |
| "blocks.3.cross_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 8753152, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.3.cross_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 2, |
| "offset": 8755200, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 9803776, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.3.cross_attn.to_q.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 9836544, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.3.cross_attn.to_q.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 2, |
| "offset": 9838592, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 10887168, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.3.mlp.mlp.0.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 10919936, |
| "bytes": 8192 |
| } |
| }, |
| "blocks.3.mlp.mlp.0.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4096, |
| "columns": 1024, |
| "padded_rows": 4096, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 2, |
| "offset": 10928128, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 15122432, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.3.mlp.mlp.2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 15253504, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.3.mlp.mlp.2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 4096, |
| "padded_rows": 1024, |
| "padded_columns": 4096, |
| "data": { |
| "shard": 2, |
| "offset": 15255552, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 19449856, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.3.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 19580928, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.3.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 19585024, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.3.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 19589120, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.3.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 19593216, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.3.self_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 19597312, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.3.self_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 2, |
| "offset": 19599360, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 20647936, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.3.self_attn.to_qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 20680704, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.3.self_attn.to_qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1024, |
| "padded_rows": 3072, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 2, |
| "offset": 20686848, |
| "bytes": 3145728 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 23832576, |
| "bytes": 98304 |
| } |
| }, |
| "blocks.4.cross_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 23930880, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.4.cross_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 23934976, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.4.cross_attn.to_kv.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 23939072, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.4.cross_attn.to_kv.weight": { |
| "shape": [ |
| 2048, |
| 16 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 23943168, |
| "bytes": 65536 |
| } |
| }, |
| "blocks.4.cross_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 24008704, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.4.cross_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 2, |
| "offset": 24010752, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 25059328, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.4.cross_attn.to_q.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 25092096, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.4.cross_attn.to_q.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 2, |
| "offset": 25094144, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 26142720, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.4.mlp.mlp.0.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 26175488, |
| "bytes": 8192 |
| } |
| }, |
| "blocks.4.mlp.mlp.0.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4096, |
| "columns": 1024, |
| "padded_rows": 4096, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 2, |
| "offset": 26183680, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 30377984, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.4.mlp.mlp.2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 30509056, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.4.mlp.mlp.2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 4096, |
| "padded_rows": 1024, |
| "padded_columns": 4096, |
| "data": { |
| "shard": 2, |
| "offset": 30511104, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 34705408, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.4.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 34836480, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.4.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 34840576, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.4.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 34844672, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.4.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 34848768, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.4.self_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 34852864, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.4.self_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 2, |
| "offset": 34854912, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 35903488, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.4.self_attn.to_qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 35936256, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.4.self_attn.to_qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1024, |
| "padded_rows": 3072, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 2, |
| "offset": 35942400, |
| "bytes": 3145728 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 39088128, |
| "bytes": 98304 |
| } |
| }, |
| "blocks.5.cross_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 39186432, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.5.cross_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 39190528, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.5.cross_attn.to_kv.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 39194624, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.5.cross_attn.to_kv.weight": { |
| "shape": [ |
| 2048, |
| 16 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 39198720, |
| "bytes": 65536 |
| } |
| }, |
| "blocks.5.cross_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 39264256, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.5.cross_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 2, |
| "offset": 39266304, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 40314880, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.5.cross_attn.to_q.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 40347648, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.5.cross_attn.to_q.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 2, |
| "offset": 40349696, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 41398272, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.5.mlp.mlp.0.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 41431040, |
| "bytes": 8192 |
| } |
| }, |
| "blocks.5.mlp.mlp.0.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4096, |
| "columns": 1024, |
| "padded_rows": 4096, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 2, |
| "offset": 41439232, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 45633536, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.5.mlp.mlp.2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 45764608, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.5.mlp.mlp.2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 4096, |
| "padded_rows": 1024, |
| "padded_columns": 4096, |
| "data": { |
| "shard": 2, |
| "offset": 45766656, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 49960960, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.5.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 50092032, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.5.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 50096128, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.5.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 50100224, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.5.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 50104320, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.5.self_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 50108416, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.5.self_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 2, |
| "offset": 50110464, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 51159040, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.5.self_attn.to_qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 51191808, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.5.self_attn.to_qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1024, |
| "padded_rows": 3072, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 2, |
| "offset": 51197952, |
| "bytes": 3145728 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 54343680, |
| "bytes": 98304 |
| } |
| }, |
| "blocks.6.cross_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 54441984, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.6.cross_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 54446080, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.6.cross_attn.to_kv.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 54450176, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.6.cross_attn.to_kv.weight": { |
| "shape": [ |
| 2048, |
| 16 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 54454272, |
| "bytes": 65536 |
| } |
| }, |
| "blocks.6.cross_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 54519808, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.6.cross_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 2, |
| "offset": 54521856, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 55570432, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.6.cross_attn.to_q.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 55603200, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.6.cross_attn.to_q.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 2, |
| "offset": 55605248, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 56653824, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.6.mlp.mlp.0.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 56686592, |
| "bytes": 8192 |
| } |
| }, |
| "blocks.6.mlp.mlp.0.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4096, |
| "columns": 1024, |
| "padded_rows": 4096, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 2, |
| "offset": 56694784, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 60889088, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.6.mlp.mlp.2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 61020160, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.6.mlp.mlp.2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 4096, |
| "padded_rows": 1024, |
| "padded_columns": 4096, |
| "data": { |
| "shard": 2, |
| "offset": 61022208, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 65216512, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.6.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 65347584, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.6.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 65351680, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.6.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 65355776, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.6.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 65359872, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.6.self_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 65363968, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.6.self_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 2, |
| "offset": 65366016, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 66414592, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.6.self_attn.to_qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 66447360, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.6.self_attn.to_qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1024, |
| "padded_rows": 3072, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 3, |
| "offset": 0, |
| "bytes": 3145728 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 3145728, |
| "bytes": 98304 |
| } |
| }, |
| "blocks.7.cross_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 3244032, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.7.cross_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 3248128, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.7.cross_attn.to_kv.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 3252224, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.7.cross_attn.to_kv.weight": { |
| "shape": [ |
| 2048, |
| 16 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 3256320, |
| "bytes": 65536 |
| } |
| }, |
| "blocks.7.cross_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 3321856, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.7.cross_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 3, |
| "offset": 3323904, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 4372480, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.7.cross_attn.to_q.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 4405248, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.7.cross_attn.to_q.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 3, |
| "offset": 4407296, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 5455872, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.7.mlp.mlp.0.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 5488640, |
| "bytes": 8192 |
| } |
| }, |
| "blocks.7.mlp.mlp.0.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4096, |
| "columns": 1024, |
| "padded_rows": 4096, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 3, |
| "offset": 5496832, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 9691136, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.7.mlp.mlp.2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 9822208, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.7.mlp.mlp.2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 4096, |
| "padded_rows": 1024, |
| "padded_columns": 4096, |
| "data": { |
| "shard": 3, |
| "offset": 9824256, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 14018560, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.7.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 14149632, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.7.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 14153728, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.7.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 14157824, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.7.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 14161920, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.7.self_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 14166016, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.7.self_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 3, |
| "offset": 14168064, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 15216640, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.7.self_attn.to_qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 15249408, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.7.self_attn.to_qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1024, |
| "padded_rows": 3072, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 3, |
| "offset": 15255552, |
| "bytes": 3145728 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 18401280, |
| "bytes": 98304 |
| } |
| }, |
| "blocks.8.cross_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 18499584, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.8.cross_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 18503680, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.8.cross_attn.to_kv.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 18507776, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.8.cross_attn.to_kv.weight": { |
| "shape": [ |
| 2048, |
| 16 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 18511872, |
| "bytes": 65536 |
| } |
| }, |
| "blocks.8.cross_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 18577408, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.8.cross_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 3, |
| "offset": 18579456, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 19628032, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.8.cross_attn.to_q.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 19660800, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.8.cross_attn.to_q.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 3, |
| "offset": 19662848, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 20711424, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.8.mlp.mlp.0.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 20744192, |
| "bytes": 8192 |
| } |
| }, |
| "blocks.8.mlp.mlp.0.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4096, |
| "columns": 1024, |
| "padded_rows": 4096, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 3, |
| "offset": 20752384, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 24946688, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.8.mlp.mlp.2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 25077760, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.8.mlp.mlp.2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 4096, |
| "padded_rows": 1024, |
| "padded_columns": 4096, |
| "data": { |
| "shard": 3, |
| "offset": 25079808, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 29274112, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.8.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 29405184, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.8.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 29409280, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.8.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 29413376, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.8.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 29417472, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.8.self_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 29421568, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.8.self_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 3, |
| "offset": 29423616, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 30472192, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.8.self_attn.to_qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 30504960, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.8.self_attn.to_qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1024, |
| "padded_rows": 3072, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 3, |
| "offset": 30511104, |
| "bytes": 3145728 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 33656832, |
| "bytes": 98304 |
| } |
| }, |
| "blocks.9.cross_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 33755136, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.9.cross_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 33759232, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.9.cross_attn.to_kv.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 33763328, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.9.cross_attn.to_kv.weight": { |
| "shape": [ |
| 2048, |
| 16 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 33767424, |
| "bytes": 65536 |
| } |
| }, |
| "blocks.9.cross_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 33832960, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.9.cross_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 3, |
| "offset": 33835008, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 34883584, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.9.cross_attn.to_q.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 34916352, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.9.cross_attn.to_q.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 3, |
| "offset": 34918400, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 35966976, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.9.mlp.mlp.0.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 35999744, |
| "bytes": 8192 |
| } |
| }, |
| "blocks.9.mlp.mlp.0.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4096, |
| "columns": 1024, |
| "padded_rows": 4096, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 3, |
| "offset": 36007936, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 40202240, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.9.mlp.mlp.2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 40333312, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.9.mlp.mlp.2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 4096, |
| "padded_rows": 1024, |
| "padded_columns": 4096, |
| "data": { |
| "shard": 3, |
| "offset": 40335360, |
| "bytes": 4194304 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 44529664, |
| "bytes": 131072 |
| } |
| }, |
| "blocks.9.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 44660736, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.9.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 44664832, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.9.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 44668928, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.9.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 16, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 44673024, |
| "bytes": 4096 |
| } |
| }, |
| "blocks.9.self_attn.to_out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 44677120, |
| "bytes": 2048 |
| } |
| }, |
| "blocks.9.self_attn.to_out.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1024, |
| "columns": 1024, |
| "padded_rows": 1024, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 3, |
| "offset": 44679168, |
| "bytes": 1048576 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 45727744, |
| "bytes": 32768 |
| } |
| }, |
| "blocks.9.self_attn.to_qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 45760512, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.9.self_attn.to_qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1024, |
| "padded_rows": 3072, |
| "padded_columns": 1024, |
| "data": { |
| "shard": 3, |
| "offset": 45766656, |
| "bytes": 3145728 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 48912384, |
| "bytes": 98304 |
| } |
| }, |
| "in_proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 49010688, |
| "bytes": 4096 |
| } |
| }, |
| "in_proj.weight": { |
| "shape": [ |
| 1024, |
| 3 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 49014784, |
| "bytes": 12288 |
| } |
| }, |
| "input_layer.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 49027072, |
| "bytes": 4096 |
| } |
| }, |
| "input_layer.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 49031168, |
| "bytes": 4194304 |
| } |
| }, |
| "out_proj.bias": { |
| "shape": [ |
| 480 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 53225472, |
| "bytes": 1920 |
| } |
| }, |
| "out_proj.weight": { |
| "shape": [ |
| 480, |
| 1024 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 53227520, |
| "bytes": 1966080 |
| } |
| }, |
| "points_offset_perturbation": { |
| "shape": [ |
| 32, |
| 3 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 55193600, |
| "bytes": 384 |
| } |
| } |
| } |
| } |
|
|
