diff --git a/.gitattributes b/.gitattributes index 7cf5d11a9f7ab50b3c839939f6cf72e3afa6bf38..ae10b5423eac2463adc8c3269a8ddd847aed03fc 100644 --- a/.gitattributes +++ b/.gitattributes @@ -279,3 +279,6 @@ experiments/rl_finetuning/outputs/craftax_classic_ablations/figures/t_bin_grad_n experiments/rl_finetuning/outputs/craftax_classic_ablations/figures/t_bin_norms_heatmap.png filter=lfs diff=lfs merge=lfs -text experiments/rl_finetuning/outputs/craftax_classic_ablations/figures/t_distribution_analysis.png filter=lfs diff=lfs merge=lfs -text experiments/rl_finetuning/outputs/craftax_classic_ablations/figures/win_rate_and_effective_batch_size.png filter=lfs diff=lfs merge=lfs -text +checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/ocdbt.process_0/d/503349be2d11cb1bf04a17e7b3e2d1e0 filter=lfs diff=lfs merge=lfs -text +checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/ocdbt.process_0/d/781b85e3ec447139f7d23160643a9c21 filter=lfs diff=lfs merge=lfs -text +checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/ocdbt.process_0/d/92adc334c11103c7741baed0d71e66cb filter=lfs diff=lfs merge=lfs -text diff --git a/README.md b/README.md index 73aad526b83c158459e18d2e57cc1f949670ea65..4c5538f3ac273d08fc6b5198ae0bc20b44ca3046 100644 --- a/README.md +++ b/README.md @@ -26,6 +26,7 @@ Code, configs and evaluation harness: https://github.com/mathisweil/craftax-ReMD | Path | Role | Environment | Architecture | Selected at | Training | Size | |---|---|---|---|---|---|---| +| `checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M` | Diffusion planner (offline BC) | `Craftax-Classic-Symbolic-v1` | 6L, d_model 384, 8 heads, horizon 32 | 1,000,000,000 | 97,600 grad steps | 97 MB | | `checkpoints/online/Craftax-Classic-Symbolic-v1-Online-Diffusion-DAgger-100M` | Diffusion planner (online DAgger) | `Craftax-Classic-Symbolic-v1` | 6L, d_model 384, 8 heads, horizon 32 | 100,000,000 | 97,600 grad steps | 97 MB | | `checkpoints/ppo_agents/Craftax-Classic-Symbolic-v1-PPO_RNN-1000M` | PPO-RNN expert | `Craftax-Classic-Symbolic-v1` | RNN, layer size 512 | 1,000,000,000 | 1e+09 frames | 35 MB | | `checkpoints/ppo_agents/Craftax-Symbolic-v1-PPO_RNN-1000M` | PPO-RNN expert | `Craftax-Symbolic-v1` | RNN, layer size 512 | 1,000,000,000 | 1e+09 frames | 50 MB | @@ -52,14 +53,14 @@ RL fine-tuning ablation runs, as produced by `experiments/rl_finetuning/run_abla ```python from huggingface_hub import snapshot_download -# everything (~223 MB) +# everything (~320 MB) snapshot_download(repo_id="mathisweil/remdm-craftax-checkpoints", local_dir=".") # a single model snapshot_download( repo_id="mathisweil/remdm-craftax-checkpoints", local_dir=".", - allow_patterns="checkpoints/online/Craftax-Classic-Symbolic-v1-Online-Diffusion-DAgger-100M/**", + allow_patterns="checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/**", ) ``` @@ -69,7 +70,7 @@ From a clone of the code repository, after downloading into it: ```bash uv run python main.py --mode inference \ - --checkpoint checkpoints/online/Craftax-Classic-Symbolic-v1-Online-Diffusion-DAgger-100M \ + --checkpoint checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M \ --output results/inference/eval.json ``` diff --git a/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/_CHECKPOINT_METADATA b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/_CHECKPOINT_METADATA new file mode 100644 index 0000000000000000000000000000000000000000..14cc138c4dcc3a1ff2a84b66bca9e3a0f0c4f613 --- /dev/null +++ b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/_CHECKPOINT_METADATA @@ -0,0 +1 @@ +{"item_handlers": {"default": "orbax.checkpoint._src.handlers.standard_checkpoint_handler.StandardCheckpointHandler"}, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1786557601120122045, "commit_timestamp_nsecs": 1786557601501380506, "custom_metadata": {}} \ No newline at end of file diff --git a/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/commit_success.txt b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/commit_success.txt new file mode 100644 index 0000000000000000000000000000000000000000..f3533ceb62cf36d6d8484937237816f3398131cf --- /dev/null +++ b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/commit_success.txt @@ -0,0 +1 @@ +Checkpoint commit was successful to /workspace/craftax-ReMDM-planner/wandb/run-20260812_110141-jogr4aj6/files/policies/99942400 \ No newline at end of file diff --git a/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/_METADATA b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/_METADATA new file mode 100644 index 0000000000000000000000000000000000000000..8d64d23ddc877e9c3a263d5a2d60556a82bc8ff2 --- /dev/null +++ b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/_METADATA @@ -0,0 +1 @@ +{"tree_metadata": {"('step',)": {"key_metadata": [{"key": "step", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('params', 'params', 'Dense_0', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('params', 'params', 'Dense_0', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [1345, 768]}}, "('params', 'params', 'Dense_1', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('params', 'params', 'Dense_1', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 768]}}, "('params', 'params', 'Dense_2', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_2", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'Dense_2', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_2", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 384]}}, "('params', 'params', 'Dense_3', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_3", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'Dense_3', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_3", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 384]}}, "('params', 'params', 'Dense_4', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_4", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'Dense_4', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_4", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 384]}}, "('params', 'params', 'Dense_5', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_5", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [17]}}, "('params', 'params', 'Dense_5', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_5", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 17]}}, "('params', 'params', 'Embed_0', 'embedding')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Embed_0", "key_type": 2}, {"key": "embedding", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [18, 384]}}, "('params', 'params', 'LayerNorm_0', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('params', 'params', 'LayerNorm_0', 'scale')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('params', 'params', 'LayerNorm_1', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'LayerNorm_1', 'scale')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_0', 'Dense_0', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('params', 'params', 'TransformerBlock_0', 'Dense_0', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 768]}}, "('params', 'params', 'TransformerBlock_0', 'Dense_1', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_0', 'Dense_1', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 384]}}, "('params', 'params', 'TransformerBlock_0', 'LayerNorm_0', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_0', 'LayerNorm_0', 'scale')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_0', 'LayerNorm_1', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_0', 'LayerNorm_1', 'scale')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_0', 'MultiHeadDotProductAttention_0', 'key', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('params', 'params', 'TransformerBlock_0', 'MultiHeadDotProductAttention_0', 'key', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('params', 'params', 'TransformerBlock_0', 'MultiHeadDotProductAttention_0', 'out', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_0', 'MultiHeadDotProductAttention_0', 'out', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48, 384]}}, "('params', 'params', 'TransformerBlock_0', 'MultiHeadDotProductAttention_0', 'query', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('params', 'params', 'TransformerBlock_0', 'MultiHeadDotProductAttention_0', 'query', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('params', 'params', 'TransformerBlock_0', 'MultiHeadDotProductAttention_0', 'value', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('params', 'params', 'TransformerBlock_0', 'MultiHeadDotProductAttention_0', 'value', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('params', 'params', 'TransformerBlock_1', 'Dense_0', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('params', 'params', 'TransformerBlock_1', 'Dense_0', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 768]}}, "('params', 'params', 'TransformerBlock_1', 'Dense_1', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_1', 'Dense_1', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 384]}}, "('params', 'params', 'TransformerBlock_1', 'LayerNorm_0', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_1', 'LayerNorm_0', 'scale')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_1', 'LayerNorm_1', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_1', 'LayerNorm_1', 'scale')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_1', 'MultiHeadDotProductAttention_0', 'key', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('params', 'params', 'TransformerBlock_1', 'MultiHeadDotProductAttention_0', 'key', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('params', 'params', 'TransformerBlock_1', 'MultiHeadDotProductAttention_0', 'out', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_1', 'MultiHeadDotProductAttention_0', 'out', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48, 384]}}, "('params', 'params', 'TransformerBlock_1', 'MultiHeadDotProductAttention_0', 'query', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('params', 'params', 'TransformerBlock_1', 'MultiHeadDotProductAttention_0', 'query', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('params', 'params', 'TransformerBlock_1', 'MultiHeadDotProductAttention_0', 'value', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('params', 'params', 'TransformerBlock_1', 'MultiHeadDotProductAttention_0', 'value', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('params', 'params', 'TransformerBlock_2', 'Dense_0', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('params', 'params', 'TransformerBlock_2', 'Dense_0', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 768]}}, "('params', 'params', 'TransformerBlock_2', 'Dense_1', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_2', 'Dense_1', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 384]}}, "('params', 'params', 'TransformerBlock_2', 'LayerNorm_0', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_2', 'LayerNorm_0', 'scale')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_2', 'LayerNorm_1', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_2', 'LayerNorm_1', 'scale')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_2', 'MultiHeadDotProductAttention_0', 'key', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('params', 'params', 'TransformerBlock_2', 'MultiHeadDotProductAttention_0', 'key', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('params', 'params', 'TransformerBlock_2', 'MultiHeadDotProductAttention_0', 'out', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_2', 'MultiHeadDotProductAttention_0', 'out', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48, 384]}}, "('params', 'params', 'TransformerBlock_2', 'MultiHeadDotProductAttention_0', 'query', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('params', 'params', 'TransformerBlock_2', 'MultiHeadDotProductAttention_0', 'query', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('params', 'params', 'TransformerBlock_2', 'MultiHeadDotProductAttention_0', 'value', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('params', 'params', 'TransformerBlock_2', 'MultiHeadDotProductAttention_0', 'value', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('params', 'params', 'TransformerBlock_3', 'Dense_0', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('params', 'params', 'TransformerBlock_3', 'Dense_0', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 768]}}, "('params', 'params', 'TransformerBlock_3', 'Dense_1', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_3', 'Dense_1', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 384]}}, "('params', 'params', 'TransformerBlock_3', 'LayerNorm_0', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_3', 'LayerNorm_0', 'scale')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_3', 'LayerNorm_1', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_3', 'LayerNorm_1', 'scale')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_3', 'MultiHeadDotProductAttention_0', 'key', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('params', 'params', 'TransformerBlock_3', 'MultiHeadDotProductAttention_0', 'key', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('params', 'params', 'TransformerBlock_3', 'MultiHeadDotProductAttention_0', 'out', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_3', 'MultiHeadDotProductAttention_0', 'out', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48, 384]}}, "('params', 'params', 'TransformerBlock_3', 'MultiHeadDotProductAttention_0', 'query', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('params', 'params', 'TransformerBlock_3', 'MultiHeadDotProductAttention_0', 'query', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('params', 'params', 'TransformerBlock_3', 'MultiHeadDotProductAttention_0', 'value', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('params', 'params', 'TransformerBlock_3', 'MultiHeadDotProductAttention_0', 'value', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('params', 'params', 'TransformerBlock_4', 'Dense_0', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('params', 'params', 'TransformerBlock_4', 'Dense_0', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 768]}}, "('params', 'params', 'TransformerBlock_4', 'Dense_1', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_4', 'Dense_1', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 384]}}, "('params', 'params', 'TransformerBlock_4', 'LayerNorm_0', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_4', 'LayerNorm_0', 'scale')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_4', 'LayerNorm_1', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_4', 'LayerNorm_1', 'scale')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_4', 'MultiHeadDotProductAttention_0', 'key', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('params', 'params', 'TransformerBlock_4', 'MultiHeadDotProductAttention_0', 'key', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('params', 'params', 'TransformerBlock_4', 'MultiHeadDotProductAttention_0', 'out', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_4', 'MultiHeadDotProductAttention_0', 'out', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48, 384]}}, "('params', 'params', 'TransformerBlock_4', 'MultiHeadDotProductAttention_0', 'query', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('params', 'params', 'TransformerBlock_4', 'MultiHeadDotProductAttention_0', 'query', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('params', 'params', 'TransformerBlock_4', 'MultiHeadDotProductAttention_0', 'value', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('params', 'params', 'TransformerBlock_4', 'MultiHeadDotProductAttention_0', 'value', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('params', 'params', 'TransformerBlock_5', 'Dense_0', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('params', 'params', 'TransformerBlock_5', 'Dense_0', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 768]}}, "('params', 'params', 'TransformerBlock_5', 'Dense_1', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_5', 'Dense_1', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 384]}}, "('params', 'params', 'TransformerBlock_5', 'LayerNorm_0', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_5', 'LayerNorm_0', 'scale')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_5', 'LayerNorm_1', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_5', 'LayerNorm_1', 'scale')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_5', 'MultiHeadDotProductAttention_0', 'key', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('params', 'params', 'TransformerBlock_5', 'MultiHeadDotProductAttention_0', 'key', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('params', 'params', 'TransformerBlock_5', 'MultiHeadDotProductAttention_0', 'out', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('params', 'params', 'TransformerBlock_5', 'MultiHeadDotProductAttention_0', 'out', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48, 384]}}, "('params', 'params', 'TransformerBlock_5', 'MultiHeadDotProductAttention_0', 'query', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('params', 'params', 'TransformerBlock_5', 'MultiHeadDotProductAttention_0', 'query', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('params', 'params', 'TransformerBlock_5', 'MultiHeadDotProductAttention_0', 'value', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('params', 'params', 'TransformerBlock_5', 'MultiHeadDotProductAttention_0', 'value', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '0')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "0", "key_type": 1}], "value_metadata": {"value_type": "None", "skip_deserialize": true}}, "('opt_state', '1', '0', 'count')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "count", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('opt_state', '1', '0', 'mu', 'params', 'Dense_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('opt_state', '1', '0', 'mu', 'params', 'Dense_0', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [1345, 768]}}, "('opt_state', '1', '0', 'mu', 'params', 'Dense_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('opt_state', '1', '0', 'mu', 'params', 'Dense_1', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 768]}}, "('opt_state', '1', '0', 'mu', 'params', 'Dense_2', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_2", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'Dense_2', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_2", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 384]}}, "('opt_state', '1', '0', 'mu', 'params', 'Dense_3', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_3", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'Dense_3', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_3", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 384]}}, "('opt_state', '1', '0', 'mu', 'params', 'Dense_4', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_4", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'Dense_4', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_4", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 384]}}, "('opt_state', '1', '0', 'mu', 'params', 'Dense_5', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_5", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [17]}}, "('opt_state', '1', '0', 'mu', 'params', 'Dense_5', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_5", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 17]}}, "('opt_state', '1', '0', 'mu', 'params', 'Embed_0', 'embedding')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Embed_0", "key_type": 2}, {"key": "embedding", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [18, 384]}}, "('opt_state', '1', '0', 'mu', 'params', 'LayerNorm_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('opt_state', '1', '0', 'mu', 'params', 'LayerNorm_0', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('opt_state', '1', '0', 'mu', 'params', 'LayerNorm_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'LayerNorm_1', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_0', 'Dense_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_0', 'Dense_0', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 768]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_0', 'Dense_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_0', 'Dense_1', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_0', 'LayerNorm_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_0', 'LayerNorm_0', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_0', 'LayerNorm_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_0', 'LayerNorm_1', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_0', 'MultiHeadDotProductAttention_0', 'key', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_0', 'MultiHeadDotProductAttention_0', 'key', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_0', 'MultiHeadDotProductAttention_0', 'out', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_0', 'MultiHeadDotProductAttention_0', 'out', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48, 384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_0', 'MultiHeadDotProductAttention_0', 'query', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_0', 'MultiHeadDotProductAttention_0', 'query', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_0', 'MultiHeadDotProductAttention_0', 'value', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_0', 'MultiHeadDotProductAttention_0', 'value', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_1', 'Dense_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_1', 'Dense_0', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 768]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_1', 'Dense_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_1', 'Dense_1', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_1', 'LayerNorm_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_1', 'LayerNorm_0', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_1', 'LayerNorm_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_1', 'LayerNorm_1', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_1', 'MultiHeadDotProductAttention_0', 'key', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_1', 'MultiHeadDotProductAttention_0', 'key', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_1', 'MultiHeadDotProductAttention_0', 'out', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_1', 'MultiHeadDotProductAttention_0', 'out', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48, 384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_1', 'MultiHeadDotProductAttention_0', 'query', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_1', 'MultiHeadDotProductAttention_0', 'query', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_1', 'MultiHeadDotProductAttention_0', 'value', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_1', 'MultiHeadDotProductAttention_0', 'value', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_2', 'Dense_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_2', 'Dense_0', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 768]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_2', 'Dense_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_2', 'Dense_1', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_2', 'LayerNorm_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_2', 'LayerNorm_0', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_2', 'LayerNorm_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_2', 'LayerNorm_1', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_2', 'MultiHeadDotProductAttention_0', 'key', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_2', 'MultiHeadDotProductAttention_0', 'key', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_2', 'MultiHeadDotProductAttention_0', 'out', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_2', 'MultiHeadDotProductAttention_0', 'out', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48, 384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_2', 'MultiHeadDotProductAttention_0', 'query', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_2', 'MultiHeadDotProductAttention_0', 'query', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_2', 'MultiHeadDotProductAttention_0', 'value', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_2', 'MultiHeadDotProductAttention_0', 'value', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_3', 'Dense_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_3', 'Dense_0', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 768]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_3', 'Dense_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_3', 'Dense_1', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_3', 'LayerNorm_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_3', 'LayerNorm_0', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_3', 'LayerNorm_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_3', 'LayerNorm_1', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_3', 'MultiHeadDotProductAttention_0', 'key', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_3', 'MultiHeadDotProductAttention_0', 'key', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_3', 'MultiHeadDotProductAttention_0', 'out', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_3', 'MultiHeadDotProductAttention_0', 'out', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48, 384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_3', 'MultiHeadDotProductAttention_0', 'query', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_3', 'MultiHeadDotProductAttention_0', 'query', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_3', 'MultiHeadDotProductAttention_0', 'value', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_3', 'MultiHeadDotProductAttention_0', 'value', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_4', 'Dense_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_4', 'Dense_0', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 768]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_4', 'Dense_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_4', 'Dense_1', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_4', 'LayerNorm_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_4', 'LayerNorm_0', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_4', 'LayerNorm_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_4', 'LayerNorm_1', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_4', 'MultiHeadDotProductAttention_0', 'key', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_4', 'MultiHeadDotProductAttention_0', 'key', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_4', 'MultiHeadDotProductAttention_0', 'out', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_4', 'MultiHeadDotProductAttention_0', 'out', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48, 384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_4', 'MultiHeadDotProductAttention_0', 'query', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_4', 'MultiHeadDotProductAttention_0', 'query', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_4', 'MultiHeadDotProductAttention_0', 'value', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_4', 'MultiHeadDotProductAttention_0', 'value', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_5', 'Dense_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_5', 'Dense_0', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 768]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_5', 'Dense_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_5', 'Dense_1', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_5', 'LayerNorm_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_5', 'LayerNorm_0', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_5', 'LayerNorm_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_5', 'LayerNorm_1', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_5', 'MultiHeadDotProductAttention_0', 'key', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_5', 'MultiHeadDotProductAttention_0', 'key', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_5', 'MultiHeadDotProductAttention_0', 'out', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_5', 'MultiHeadDotProductAttention_0', 'out', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48, 384]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_5', 'MultiHeadDotProductAttention_0', 'query', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_5', 'MultiHeadDotProductAttention_0', 'query', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_5', 'MultiHeadDotProductAttention_0', 'value', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'mu', 'params', 'TransformerBlock_5', 'MultiHeadDotProductAttention_0', 'value', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "mu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'Dense_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('opt_state', '1', '0', 'nu', 'params', 'Dense_0', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [1345, 768]}}, "('opt_state', '1', '0', 'nu', 'params', 'Dense_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('opt_state', '1', '0', 'nu', 'params', 'Dense_1', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 768]}}, "('opt_state', '1', '0', 'nu', 'params', 'Dense_2', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_2", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'Dense_2', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_2", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 384]}}, "('opt_state', '1', '0', 'nu', 'params', 'Dense_3', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_3", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'Dense_3', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_3", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 384]}}, "('opt_state', '1', '0', 'nu', 'params', 'Dense_4', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_4", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'Dense_4', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_4", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 384]}}, "('opt_state', '1', '0', 'nu', 'params', 'Dense_5', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_5", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [17]}}, "('opt_state', '1', '0', 'nu', 'params', 'Dense_5', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Dense_5", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 17]}}, "('opt_state', '1', '0', 'nu', 'params', 'Embed_0', 'embedding')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "Embed_0", "key_type": 2}, {"key": "embedding", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [18, 384]}}, "('opt_state', '1', '0', 'nu', 'params', 'LayerNorm_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('opt_state', '1', '0', 'nu', 'params', 'LayerNorm_0', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('opt_state', '1', '0', 'nu', 'params', 'LayerNorm_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'LayerNorm_1', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_0', 'Dense_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_0', 'Dense_0', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 768]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_0', 'Dense_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_0', 'Dense_1', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_0', 'LayerNorm_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_0', 'LayerNorm_0', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_0', 'LayerNorm_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_0', 'LayerNorm_1', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_0', 'MultiHeadDotProductAttention_0', 'key', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_0', 'MultiHeadDotProductAttention_0', 'key', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_0', 'MultiHeadDotProductAttention_0', 'out', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_0', 'MultiHeadDotProductAttention_0', 'out', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48, 384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_0', 'MultiHeadDotProductAttention_0', 'query', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_0', 'MultiHeadDotProductAttention_0', 'query', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_0', 'MultiHeadDotProductAttention_0', 'value', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_0', 'MultiHeadDotProductAttention_0', 'value', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_0", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_1', 'Dense_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_1', 'Dense_0', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 768]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_1', 'Dense_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_1', 'Dense_1', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_1', 'LayerNorm_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_1', 'LayerNorm_0', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_1', 'LayerNorm_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_1', 'LayerNorm_1', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_1', 'MultiHeadDotProductAttention_0', 'key', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_1', 'MultiHeadDotProductAttention_0', 'key', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_1', 'MultiHeadDotProductAttention_0', 'out', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_1', 'MultiHeadDotProductAttention_0', 'out', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48, 384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_1', 'MultiHeadDotProductAttention_0', 'query', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_1', 'MultiHeadDotProductAttention_0', 'query', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_1', 'MultiHeadDotProductAttention_0', 'value', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_1', 'MultiHeadDotProductAttention_0', 'value', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_1", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_2', 'Dense_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_2', 'Dense_0', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 768]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_2', 'Dense_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_2', 'Dense_1', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_2', 'LayerNorm_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_2', 'LayerNorm_0', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_2', 'LayerNorm_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_2', 'LayerNorm_1', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_2', 'MultiHeadDotProductAttention_0', 'key', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_2', 'MultiHeadDotProductAttention_0', 'key', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_2', 'MultiHeadDotProductAttention_0', 'out', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_2', 'MultiHeadDotProductAttention_0', 'out', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48, 384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_2', 'MultiHeadDotProductAttention_0', 'query', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_2', 'MultiHeadDotProductAttention_0', 'query', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_2', 'MultiHeadDotProductAttention_0', 'value', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_2', 'MultiHeadDotProductAttention_0', 'value', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_2", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_3', 'Dense_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_3', 'Dense_0', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 768]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_3', 'Dense_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_3', 'Dense_1', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_3', 'LayerNorm_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_3', 'LayerNorm_0', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_3', 'LayerNorm_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_3', 'LayerNorm_1', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_3', 'MultiHeadDotProductAttention_0', 'key', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_3', 'MultiHeadDotProductAttention_0', 'key', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_3', 'MultiHeadDotProductAttention_0', 'out', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_3', 'MultiHeadDotProductAttention_0', 'out', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48, 384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_3', 'MultiHeadDotProductAttention_0', 'query', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_3', 'MultiHeadDotProductAttention_0', 'query', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_3', 'MultiHeadDotProductAttention_0', 'value', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_3', 'MultiHeadDotProductAttention_0', 'value', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_3", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_4', 'Dense_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_4', 'Dense_0', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 768]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_4', 'Dense_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_4', 'Dense_1', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_4', 'LayerNorm_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_4', 'LayerNorm_0', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_4', 'LayerNorm_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_4', 'LayerNorm_1', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_4', 'MultiHeadDotProductAttention_0', 'key', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_4', 'MultiHeadDotProductAttention_0', 'key', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_4', 'MultiHeadDotProductAttention_0', 'out', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_4', 'MultiHeadDotProductAttention_0', 'out', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48, 384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_4', 'MultiHeadDotProductAttention_0', 'query', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_4', 'MultiHeadDotProductAttention_0', 'query', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_4', 'MultiHeadDotProductAttention_0', 'value', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_4', 'MultiHeadDotProductAttention_0', 'value', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_4", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_5', 'Dense_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_5', 'Dense_0', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "Dense_0", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 768]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_5', 'Dense_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_5', 'Dense_1', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "Dense_1", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_5', 'LayerNorm_0', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_5', 'LayerNorm_0', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "LayerNorm_0", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_5', 'LayerNorm_1', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_5', 'LayerNorm_1', 'scale')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "LayerNorm_1", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_5', 'MultiHeadDotProductAttention_0', 'key', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_5', 'MultiHeadDotProductAttention_0', 'key', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_5', 'MultiHeadDotProductAttention_0', 'out', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_5', 'MultiHeadDotProductAttention_0', 'out', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48, 384]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_5', 'MultiHeadDotProductAttention_0', 'query', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_5', 'MultiHeadDotProductAttention_0', 'query', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_5', 'MultiHeadDotProductAttention_0', 'value', 'bias')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [8, 48]}}, "('opt_state', '1', '0', 'nu', 'params', 'TransformerBlock_5', 'MultiHeadDotProductAttention_0', 'value', 'kernel')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "0", "key_type": 1}, {"key": "nu", "key_type": 2}, {"key": "params", "key_type": 2}, {"key": "TransformerBlock_5", "key_type": 2}, {"key": "MultiHeadDotProductAttention_0", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 8, 48]}}, "('opt_state', '1', '1', 'count')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 1}, {"key": "1", "key_type": 1}, {"key": "count", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}}, "use_ocdbt": true, "use_zarr3": false, "store_array_data_equal_to_fill_value": true, "custom_metadata": null} \ No newline at end of file diff --git a/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/_sharding b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/_sharding new file mode 100644 index 0000000000000000000000000000000000000000..c3e2898e48ce0d01971669ae6a0dc6bd5228e39f --- /dev/null +++ b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/_sharding @@ -0,0 +1 @@ +{"b3B0X3N0YXRlLjEuMC5jb3VudA==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuRGVuc2VfMC5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuRGVuc2VfMC5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuRGVuc2VfMS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuRGVuc2VfMS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuRGVuc2VfMi5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuRGVuc2VfMi5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuRGVuc2VfMy5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuRGVuc2VfMy5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuRGVuc2VfNC5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuRGVuc2VfNC5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuRGVuc2VfNS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuRGVuc2VfNS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuRW1iZWRfMC5lbWJlZGRpbmc=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuTGF5ZXJOb3JtXzAuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuTGF5ZXJOb3JtXzAuc2NhbGU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuTGF5ZXJOb3JtXzEuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuTGF5ZXJOb3JtXzEuc2NhbGU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC52YWx1ZS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC52YWx1ZS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5rZXkuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5rZXkua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5vdXQuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5vdXQua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5xdWVyeS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5xdWVyeS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180LkRlbnNlXzAuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180LkRlbnNlXzAua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180LkRlbnNlXzEuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180LkRlbnNlXzEua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180LkxheWVyTm9ybV8wLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180LkxheWVyTm9ybV8wLnNjYWxl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180LkxheWVyTm9ybV8xLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180LkxheWVyTm9ybV8xLnNjYWxl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC52YWx1ZS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC52YWx1ZS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5rZXkuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5rZXkua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5vdXQuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5vdXQua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5xdWVyeS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5xdWVyeS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181LkRlbnNlXzAuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181LkRlbnNlXzAua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181LkRlbnNlXzEuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181LkRlbnNlXzEua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181LkxheWVyTm9ybV8wLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181LkxheWVyTm9ybV8wLnNjYWxl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181LkxheWVyTm9ybV8xLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181LkxheWVyTm9ybV8xLnNjYWxl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC52YWx1ZS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC52YWx1ZS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5rZXkuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5rZXkua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5vdXQuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5vdXQua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5xdWVyeS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5xdWVyeS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLkRlbnNlXzAuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLkRlbnNlXzAua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLkRlbnNlXzEuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLkRlbnNlXzEua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLkxheWVyTm9ybV8wLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLkxheWVyTm9ybV8wLnNjYWxl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLkxheWVyTm9ybV8xLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLkxheWVyTm9ybV8xLnNjYWxl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC52YWx1ZS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC52YWx1ZS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5rZXkuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5rZXkua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5vdXQuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5vdXQua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5xdWVyeS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5xdWVyeS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLkRlbnNlXzAuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLkRlbnNlXzAua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLkRlbnNlXzEuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLkRlbnNlXzEua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLkxheWVyTm9ybV8wLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLkxheWVyTm9ybV8wLnNjYWxl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLkxheWVyTm9ybV8xLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLkxheWVyTm9ybV8xLnNjYWxl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC52YWx1ZS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC52YWx1ZS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5rZXkuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5rZXkua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5vdXQuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5vdXQua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5xdWVyeS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5xdWVyeS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLkRlbnNlXzAuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLkRlbnNlXzAua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLkRlbnNlXzEuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLkRlbnNlXzEua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLkxheWVyTm9ybV8wLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLkxheWVyTm9ybV8wLnNjYWxl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLkxheWVyTm9ybV8xLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLkxheWVyTm9ybV8xLnNjYWxl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC52YWx1ZS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC52YWx1ZS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5rZXkuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5rZXkua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5vdXQuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5vdXQua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5xdWVyeS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5xdWVyeS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLkRlbnNlXzAuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLkRlbnNlXzAua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLkRlbnNlXzEuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLkRlbnNlXzEua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLkxheWVyTm9ybV8wLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLkxheWVyTm9ybV8wLnNjYWxl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLkxheWVyTm9ybV8xLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5tdS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLkxheWVyTm9ybV8xLnNjYWxl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuRGVuc2VfMC5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuRGVuc2VfMC5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuRGVuc2VfMS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuRGVuc2VfMS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuRGVuc2VfMi5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuRGVuc2VfMi5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuRGVuc2VfMy5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuRGVuc2VfMy5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuRGVuc2VfNC5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuRGVuc2VfNC5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuRGVuc2VfNS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuRGVuc2VfNS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuRW1iZWRfMC5lbWJlZGRpbmc=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuTGF5ZXJOb3JtXzAuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuTGF5ZXJOb3JtXzAuc2NhbGU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuTGF5ZXJOb3JtXzEuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuTGF5ZXJOb3JtXzEuc2NhbGU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC52YWx1ZS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC52YWx1ZS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5rZXkuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5rZXkua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5vdXQuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5vdXQua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5xdWVyeS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5xdWVyeS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180LkRlbnNlXzAuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180LkRlbnNlXzAua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180LkRlbnNlXzEuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180LkRlbnNlXzEua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180LkxheWVyTm9ybV8wLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180LkxheWVyTm9ybV8wLnNjYWxl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180LkxheWVyTm9ybV8xLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja180LkxheWVyTm9ybV8xLnNjYWxl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC52YWx1ZS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC52YWx1ZS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5rZXkuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5rZXkua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5vdXQuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5vdXQua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5xdWVyeS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181Lk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5xdWVyeS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181LkRlbnNlXzAuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181LkRlbnNlXzAua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181LkRlbnNlXzEuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181LkRlbnNlXzEua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181LkxheWVyTm9ybV8wLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181LkxheWVyTm9ybV8wLnNjYWxl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181LkxheWVyTm9ybV8xLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja181LkxheWVyTm9ybV8xLnNjYWxl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC52YWx1ZS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC52YWx1ZS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5rZXkuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5rZXkua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5vdXQuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5vdXQua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5xdWVyeS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5xdWVyeS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLkRlbnNlXzAuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLkRlbnNlXzAua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLkRlbnNlXzEuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLkRlbnNlXzEua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLkxheWVyTm9ybV8wLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLkxheWVyTm9ybV8wLnNjYWxl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLkxheWVyTm9ybV8xLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18wLkxheWVyTm9ybV8xLnNjYWxl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC52YWx1ZS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC52YWx1ZS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5rZXkuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5rZXkua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5vdXQuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5vdXQua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5xdWVyeS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5xdWVyeS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLkRlbnNlXzAuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLkRlbnNlXzAua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLkRlbnNlXzEuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLkRlbnNlXzEua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLkxheWVyTm9ybV8wLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLkxheWVyTm9ybV8wLnNjYWxl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLkxheWVyTm9ybV8xLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18xLkxheWVyTm9ybV8xLnNjYWxl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC52YWx1ZS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC52YWx1ZS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5rZXkuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5rZXkua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5vdXQuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5vdXQua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5xdWVyeS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5xdWVyeS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLkRlbnNlXzAuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLkRlbnNlXzAua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLkRlbnNlXzEuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLkRlbnNlXzEua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLkxheWVyTm9ybV8wLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLkxheWVyTm9ybV8wLnNjYWxl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLkxheWVyTm9ybV8xLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18yLkxheWVyTm9ybV8xLnNjYWxl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC52YWx1ZS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC52YWx1ZS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5rZXkuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5rZXkua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5vdXQuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5vdXQua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5xdWVyeS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLk11bHRpSGVhZERvdFByb2R1Y3RBdHRlbnRpb25fMC5xdWVyeS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLkRlbnNlXzAuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLkRlbnNlXzAua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLkRlbnNlXzEuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLkRlbnNlXzEua2VybmVs":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLkxheWVyTm9ybV8wLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLkxheWVyTm9ybV8wLnNjYWxl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLkxheWVyTm9ybV8xLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMC5udS5wYXJhbXMuVHJhbnNmb3JtZXJCbG9ja18zLkxheWVyTm9ybV8xLnNjYWxl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3B0X3N0YXRlLjEuMS5jb3VudA==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","c3RlcA==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5EZW5zZV80LmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5EZW5zZV80Lmtlcm5lbA==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5EZW5zZV81LmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5EZW5zZV81Lmtlcm5lbA==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5EZW5zZV8wLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5EZW5zZV8wLmtlcm5lbA==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5EZW5zZV8xLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5EZW5zZV8xLmtlcm5lbA==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5EZW5zZV8yLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5EZW5zZV8yLmtlcm5lbA==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5EZW5zZV8zLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5EZW5zZV8zLmtlcm5lbA==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5FbWJlZF8wLmVtYmVkZGluZw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5MYXllck5vcm1fMC5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5MYXllck5vcm1fMC5zY2FsZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5MYXllck5vcm1fMS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5MYXllck5vcm1fMS5zY2FsZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzAuRGVuc2VfMC5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzAuRGVuc2VfMC5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzAuRGVuc2VfMS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzAuRGVuc2VfMS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzAuTGF5ZXJOb3JtXzAuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzAuTGF5ZXJOb3JtXzAuc2NhbGU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzAuTGF5ZXJOb3JtXzEuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzAuTGF5ZXJOb3JtXzEuc2NhbGU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzAuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLm91dC5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzAuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLm91dC5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzAuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLmtleS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzAuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLmtleS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzAuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLnF1ZXJ5LmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzAuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLnF1ZXJ5Lmtlcm5lbA==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzAuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLnZhbHVlLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzAuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLnZhbHVlLmtlcm5lbA==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzEuRGVuc2VfMC5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzEuRGVuc2VfMC5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzEuRGVuc2VfMS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzEuRGVuc2VfMS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzEuTGF5ZXJOb3JtXzAuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzEuTGF5ZXJOb3JtXzAuc2NhbGU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzEuTGF5ZXJOb3JtXzEuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzEuTGF5ZXJOb3JtXzEuc2NhbGU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzEuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLm91dC5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzEuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLm91dC5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzEuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLmtleS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzEuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLmtleS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzEuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLnF1ZXJ5LmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzEuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLnF1ZXJ5Lmtlcm5lbA==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzEuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLnZhbHVlLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzEuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLnZhbHVlLmtlcm5lbA==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzIuRGVuc2VfMC5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzIuRGVuc2VfMC5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzIuRGVuc2VfMS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzIuRGVuc2VfMS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzIuTGF5ZXJOb3JtXzAuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzIuTGF5ZXJOb3JtXzAuc2NhbGU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzIuTGF5ZXJOb3JtXzEuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzIuTGF5ZXJOb3JtXzEuc2NhbGU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzIuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLm91dC5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzIuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLm91dC5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzIuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLmtleS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzIuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLmtleS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzIuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLnF1ZXJ5LmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzIuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLnF1ZXJ5Lmtlcm5lbA==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzIuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLnZhbHVlLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzIuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLnZhbHVlLmtlcm5lbA==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzMuRGVuc2VfMC5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzMuRGVuc2VfMC5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzMuRGVuc2VfMS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzMuRGVuc2VfMS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzMuTGF5ZXJOb3JtXzAuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzMuTGF5ZXJOb3JtXzAuc2NhbGU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzMuTGF5ZXJOb3JtXzEuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzMuTGF5ZXJOb3JtXzEuc2NhbGU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzMuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLm91dC5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzMuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLm91dC5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzMuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLmtleS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzMuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLmtleS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzMuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLnF1ZXJ5LmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzMuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLnF1ZXJ5Lmtlcm5lbA==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzMuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLnZhbHVlLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzMuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLnZhbHVlLmtlcm5lbA==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzQuRGVuc2VfMC5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzQuRGVuc2VfMC5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzQuRGVuc2VfMS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzQuRGVuc2VfMS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzQuTGF5ZXJOb3JtXzAuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzQuTGF5ZXJOb3JtXzAuc2NhbGU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzQuTGF5ZXJOb3JtXzEuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzQuTGF5ZXJOb3JtXzEuc2NhbGU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzQuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLm91dC5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzQuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLm91dC5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzQuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLmtleS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzQuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLmtleS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzQuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLnF1ZXJ5LmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzQuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLnF1ZXJ5Lmtlcm5lbA==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzQuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLnZhbHVlLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzQuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLnZhbHVlLmtlcm5lbA==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzUuRGVuc2VfMC5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzUuRGVuc2VfMC5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzUuRGVuc2VfMS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzUuRGVuc2VfMS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzUuTGF5ZXJOb3JtXzAuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzUuTGF5ZXJOb3JtXzAuc2NhbGU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzUuTGF5ZXJOb3JtXzEuYmlhcw==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzUuTGF5ZXJOb3JtXzEuc2NhbGU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzUuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLm91dC5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzUuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLm91dC5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzUuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLmtleS5iaWFz":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzUuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLmtleS5rZXJuZWw=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzUuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLnF1ZXJ5LmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzUuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLnF1ZXJ5Lmtlcm5lbA==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzUuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLnZhbHVlLmJpYXM=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cGFyYW1zLnBhcmFtcy5UcmFuc2Zvcm1lckJsb2NrXzUuTXVsdGlIZWFkRG90UHJvZHVjdEF0dGVudGlvbl8wLnZhbHVlLmtlcm5lbA==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}"} \ No newline at end of file diff --git a/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/array_metadatas/process_0 b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/array_metadatas/process_0 new file mode 100644 index 0000000000000000000000000000000000000000..af0fb84132e9bc0aa2fed6958e07a83773a12ae5 --- /dev/null +++ b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/array_metadatas/process_0 @@ -0,0 +1 @@ +{"array_metadatas": [{"array_metadata": {"param_name": "step", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.Dense_0.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.Dense_0.kernel", "write_shape": [1345, 768], "chunk_shape": [1345, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.Dense_1.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.Dense_1.kernel", "write_shape": [768, 768], "chunk_shape": [768, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.Dense_2.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.Dense_2.kernel", "write_shape": [768, 384], "chunk_shape": [768, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.Dense_3.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.Dense_3.kernel", "write_shape": [384, 384], "chunk_shape": [384, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.Dense_4.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.Dense_4.kernel", "write_shape": [384, 384], "chunk_shape": [384, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.Dense_5.bias", "write_shape": [17], "chunk_shape": [17], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.Dense_5.kernel", "write_shape": [384, 17], "chunk_shape": [384, 17], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.Embed_0.embedding", "write_shape": [18, 384], "chunk_shape": [18, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.LayerNorm_0.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.LayerNorm_0.scale", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.LayerNorm_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.LayerNorm_1.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_0.Dense_0.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_0.Dense_0.kernel", "write_shape": [384, 768], "chunk_shape": [384, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_0.Dense_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_0.Dense_1.kernel", "write_shape": [768, 384], "chunk_shape": [768, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_0.LayerNorm_0.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_0.LayerNorm_0.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_0.LayerNorm_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_0.LayerNorm_1.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_0.MultiHeadDotProductAttention_0.key.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_0.MultiHeadDotProductAttention_0.key.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_0.MultiHeadDotProductAttention_0.out.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_0.MultiHeadDotProductAttention_0.out.kernel", "write_shape": [8, 48, 384], "chunk_shape": [8, 48, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_0.MultiHeadDotProductAttention_0.query.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_0.MultiHeadDotProductAttention_0.query.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_0.MultiHeadDotProductAttention_0.value.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_0.MultiHeadDotProductAttention_0.value.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_1.Dense_0.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_1.Dense_0.kernel", "write_shape": [384, 768], "chunk_shape": [384, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_1.Dense_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_1.Dense_1.kernel", "write_shape": [768, 384], "chunk_shape": [768, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_1.LayerNorm_0.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_1.LayerNorm_0.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_1.LayerNorm_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_1.LayerNorm_1.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_1.MultiHeadDotProductAttention_0.key.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_1.MultiHeadDotProductAttention_0.key.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_1.MultiHeadDotProductAttention_0.out.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_1.MultiHeadDotProductAttention_0.out.kernel", "write_shape": [8, 48, 384], "chunk_shape": [8, 48, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_1.MultiHeadDotProductAttention_0.query.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_1.MultiHeadDotProductAttention_0.query.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_1.MultiHeadDotProductAttention_0.value.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_1.MultiHeadDotProductAttention_0.value.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_2.Dense_0.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_2.Dense_0.kernel", "write_shape": [384, 768], "chunk_shape": [384, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_2.Dense_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_2.Dense_1.kernel", "write_shape": [768, 384], "chunk_shape": [768, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_2.LayerNorm_0.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_2.LayerNorm_0.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_2.LayerNorm_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_2.LayerNorm_1.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_2.MultiHeadDotProductAttention_0.key.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_2.MultiHeadDotProductAttention_0.key.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_2.MultiHeadDotProductAttention_0.out.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_2.MultiHeadDotProductAttention_0.out.kernel", "write_shape": [8, 48, 384], "chunk_shape": [8, 48, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_2.MultiHeadDotProductAttention_0.query.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_2.MultiHeadDotProductAttention_0.query.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_2.MultiHeadDotProductAttention_0.value.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_2.MultiHeadDotProductAttention_0.value.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_3.Dense_0.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_3.Dense_0.kernel", "write_shape": [384, 768], "chunk_shape": [384, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_3.Dense_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_3.Dense_1.kernel", "write_shape": [768, 384], "chunk_shape": [768, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_3.LayerNorm_0.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_3.LayerNorm_0.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_3.LayerNorm_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_3.LayerNorm_1.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_3.MultiHeadDotProductAttention_0.key.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_3.MultiHeadDotProductAttention_0.key.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_3.MultiHeadDotProductAttention_0.out.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_3.MultiHeadDotProductAttention_0.out.kernel", "write_shape": [8, 48, 384], "chunk_shape": [8, 48, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_3.MultiHeadDotProductAttention_0.query.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_3.MultiHeadDotProductAttention_0.query.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_3.MultiHeadDotProductAttention_0.value.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_3.MultiHeadDotProductAttention_0.value.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_4.Dense_0.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_4.Dense_0.kernel", "write_shape": [384, 768], "chunk_shape": [384, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_4.Dense_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_4.Dense_1.kernel", "write_shape": [768, 384], "chunk_shape": [768, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_4.LayerNorm_0.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_4.LayerNorm_0.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_4.LayerNorm_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_4.LayerNorm_1.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_4.MultiHeadDotProductAttention_0.key.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_4.MultiHeadDotProductAttention_0.key.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_4.MultiHeadDotProductAttention_0.out.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_4.MultiHeadDotProductAttention_0.out.kernel", "write_shape": [8, 48, 384], "chunk_shape": [8, 48, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_4.MultiHeadDotProductAttention_0.query.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_4.MultiHeadDotProductAttention_0.query.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_4.MultiHeadDotProductAttention_0.value.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_4.MultiHeadDotProductAttention_0.value.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_5.Dense_0.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_5.Dense_0.kernel", "write_shape": [384, 768], "chunk_shape": [384, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_5.Dense_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_5.Dense_1.kernel", "write_shape": [768, 384], "chunk_shape": [768, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_5.LayerNorm_0.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_5.LayerNorm_0.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_5.LayerNorm_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_5.LayerNorm_1.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_5.MultiHeadDotProductAttention_0.key.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_5.MultiHeadDotProductAttention_0.key.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_5.MultiHeadDotProductAttention_0.out.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_5.MultiHeadDotProductAttention_0.out.kernel", "write_shape": [8, 48, 384], "chunk_shape": [8, 48, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_5.MultiHeadDotProductAttention_0.query.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_5.MultiHeadDotProductAttention_0.query.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_5.MultiHeadDotProductAttention_0.value.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "params.params.TransformerBlock_5.MultiHeadDotProductAttention_0.value.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.count", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.Dense_0.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.Dense_0.kernel", "write_shape": [1345, 768], "chunk_shape": [1345, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.Dense_1.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.Dense_1.kernel", "write_shape": [768, 768], "chunk_shape": [768, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.Dense_2.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.Dense_2.kernel", "write_shape": [768, 384], "chunk_shape": [768, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.Dense_3.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.Dense_3.kernel", "write_shape": [384, 384], "chunk_shape": [384, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.Dense_4.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.Dense_4.kernel", "write_shape": [384, 384], "chunk_shape": [384, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.Dense_5.bias", "write_shape": [17], "chunk_shape": [17], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.Dense_5.kernel", "write_shape": [384, 17], "chunk_shape": [384, 17], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.Embed_0.embedding", "write_shape": [18, 384], "chunk_shape": [18, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.LayerNorm_0.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.LayerNorm_0.scale", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.LayerNorm_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.LayerNorm_1.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_0.Dense_0.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_0.Dense_0.kernel", "write_shape": [384, 768], "chunk_shape": [384, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_0.Dense_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_0.Dense_1.kernel", "write_shape": [768, 384], "chunk_shape": [768, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_0.LayerNorm_0.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_0.LayerNorm_0.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_0.LayerNorm_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_0.LayerNorm_1.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_0.MultiHeadDotProductAttention_0.key.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_0.MultiHeadDotProductAttention_0.key.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_0.MultiHeadDotProductAttention_0.out.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_0.MultiHeadDotProductAttention_0.out.kernel", "write_shape": [8, 48, 384], "chunk_shape": [8, 48, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_0.MultiHeadDotProductAttention_0.query.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_0.MultiHeadDotProductAttention_0.query.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_0.MultiHeadDotProductAttention_0.value.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_0.MultiHeadDotProductAttention_0.value.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_1.Dense_0.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_1.Dense_0.kernel", "write_shape": [384, 768], "chunk_shape": [384, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_1.Dense_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_1.Dense_1.kernel", "write_shape": [768, 384], "chunk_shape": [768, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_1.LayerNorm_0.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_1.LayerNorm_0.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_1.LayerNorm_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_1.LayerNorm_1.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_1.MultiHeadDotProductAttention_0.key.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_1.MultiHeadDotProductAttention_0.key.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_1.MultiHeadDotProductAttention_0.out.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_1.MultiHeadDotProductAttention_0.out.kernel", "write_shape": [8, 48, 384], "chunk_shape": [8, 48, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_1.MultiHeadDotProductAttention_0.query.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_1.MultiHeadDotProductAttention_0.query.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_1.MultiHeadDotProductAttention_0.value.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_1.MultiHeadDotProductAttention_0.value.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_2.Dense_0.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_2.Dense_0.kernel", "write_shape": [384, 768], "chunk_shape": [384, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_2.Dense_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_2.Dense_1.kernel", "write_shape": [768, 384], "chunk_shape": [768, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_2.LayerNorm_0.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_2.LayerNorm_0.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_2.LayerNorm_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_2.LayerNorm_1.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_2.MultiHeadDotProductAttention_0.key.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_2.MultiHeadDotProductAttention_0.key.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_2.MultiHeadDotProductAttention_0.out.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_2.MultiHeadDotProductAttention_0.out.kernel", "write_shape": [8, 48, 384], "chunk_shape": [8, 48, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_2.MultiHeadDotProductAttention_0.query.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_2.MultiHeadDotProductAttention_0.query.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_2.MultiHeadDotProductAttention_0.value.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_2.MultiHeadDotProductAttention_0.value.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_3.Dense_0.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_3.Dense_0.kernel", "write_shape": [384, 768], "chunk_shape": [384, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_3.Dense_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_3.Dense_1.kernel", "write_shape": [768, 384], "chunk_shape": [768, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_3.LayerNorm_0.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_3.LayerNorm_0.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_3.LayerNorm_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_3.LayerNorm_1.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_3.MultiHeadDotProductAttention_0.key.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_3.MultiHeadDotProductAttention_0.key.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_3.MultiHeadDotProductAttention_0.out.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_3.MultiHeadDotProductAttention_0.out.kernel", "write_shape": [8, 48, 384], "chunk_shape": [8, 48, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_3.MultiHeadDotProductAttention_0.query.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_3.MultiHeadDotProductAttention_0.query.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_3.MultiHeadDotProductAttention_0.value.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_3.MultiHeadDotProductAttention_0.value.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_4.Dense_0.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_4.Dense_0.kernel", "write_shape": [384, 768], "chunk_shape": [384, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_4.Dense_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_4.Dense_1.kernel", "write_shape": [768, 384], "chunk_shape": [768, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_4.LayerNorm_0.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_4.LayerNorm_0.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_4.LayerNorm_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_4.LayerNorm_1.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_4.MultiHeadDotProductAttention_0.key.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_4.MultiHeadDotProductAttention_0.key.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_4.MultiHeadDotProductAttention_0.out.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_4.MultiHeadDotProductAttention_0.out.kernel", "write_shape": [8, 48, 384], "chunk_shape": [8, 48, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_4.MultiHeadDotProductAttention_0.query.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_4.MultiHeadDotProductAttention_0.query.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_4.MultiHeadDotProductAttention_0.value.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_4.MultiHeadDotProductAttention_0.value.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_5.Dense_0.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_5.Dense_0.kernel", "write_shape": [384, 768], "chunk_shape": [384, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_5.Dense_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_5.Dense_1.kernel", "write_shape": [768, 384], "chunk_shape": [768, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_5.LayerNorm_0.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_5.LayerNorm_0.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_5.LayerNorm_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_5.LayerNorm_1.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_5.MultiHeadDotProductAttention_0.key.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_5.MultiHeadDotProductAttention_0.key.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_5.MultiHeadDotProductAttention_0.out.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_5.MultiHeadDotProductAttention_0.out.kernel", "write_shape": [8, 48, 384], "chunk_shape": [8, 48, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_5.MultiHeadDotProductAttention_0.query.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_5.MultiHeadDotProductAttention_0.query.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_5.MultiHeadDotProductAttention_0.value.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.params.TransformerBlock_5.MultiHeadDotProductAttention_0.value.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.Dense_0.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.Dense_0.kernel", "write_shape": [1345, 768], "chunk_shape": [1345, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.Dense_1.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.Dense_1.kernel", "write_shape": [768, 768], "chunk_shape": [768, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.Dense_2.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.Dense_2.kernel", "write_shape": [768, 384], "chunk_shape": [768, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.Dense_3.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.Dense_3.kernel", "write_shape": [384, 384], "chunk_shape": [384, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.Dense_4.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.Dense_4.kernel", "write_shape": [384, 384], "chunk_shape": [384, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.Dense_5.bias", "write_shape": [17], "chunk_shape": [17], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.Dense_5.kernel", "write_shape": [384, 17], "chunk_shape": [384, 17], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.Embed_0.embedding", "write_shape": [18, 384], "chunk_shape": [18, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.LayerNorm_0.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.LayerNorm_0.scale", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.LayerNorm_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.LayerNorm_1.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_0.Dense_0.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_0.Dense_0.kernel", "write_shape": [384, 768], "chunk_shape": [384, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_0.Dense_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_0.Dense_1.kernel", "write_shape": [768, 384], "chunk_shape": [768, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_0.LayerNorm_0.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_0.LayerNorm_0.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_0.LayerNorm_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_0.LayerNorm_1.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_0.MultiHeadDotProductAttention_0.key.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_0.MultiHeadDotProductAttention_0.key.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_0.MultiHeadDotProductAttention_0.out.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_0.MultiHeadDotProductAttention_0.out.kernel", "write_shape": [8, 48, 384], "chunk_shape": [8, 48, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_0.MultiHeadDotProductAttention_0.query.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_0.MultiHeadDotProductAttention_0.query.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_0.MultiHeadDotProductAttention_0.value.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_0.MultiHeadDotProductAttention_0.value.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_1.Dense_0.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_1.Dense_0.kernel", "write_shape": [384, 768], "chunk_shape": [384, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_1.Dense_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_1.Dense_1.kernel", "write_shape": [768, 384], "chunk_shape": [768, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_1.LayerNorm_0.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_1.LayerNorm_0.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_1.LayerNorm_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_1.LayerNorm_1.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_1.MultiHeadDotProductAttention_0.key.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_1.MultiHeadDotProductAttention_0.key.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_1.MultiHeadDotProductAttention_0.out.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_1.MultiHeadDotProductAttention_0.out.kernel", "write_shape": [8, 48, 384], "chunk_shape": [8, 48, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_1.MultiHeadDotProductAttention_0.query.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_1.MultiHeadDotProductAttention_0.query.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_1.MultiHeadDotProductAttention_0.value.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_1.MultiHeadDotProductAttention_0.value.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_2.Dense_0.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_2.Dense_0.kernel", "write_shape": [384, 768], "chunk_shape": [384, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_2.Dense_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_2.Dense_1.kernel", "write_shape": [768, 384], "chunk_shape": [768, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_2.LayerNorm_0.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_2.LayerNorm_0.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_2.LayerNorm_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_2.LayerNorm_1.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_2.MultiHeadDotProductAttention_0.key.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_2.MultiHeadDotProductAttention_0.key.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_2.MultiHeadDotProductAttention_0.out.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_2.MultiHeadDotProductAttention_0.out.kernel", "write_shape": [8, 48, 384], "chunk_shape": [8, 48, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_2.MultiHeadDotProductAttention_0.query.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_2.MultiHeadDotProductAttention_0.query.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_2.MultiHeadDotProductAttention_0.value.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_2.MultiHeadDotProductAttention_0.value.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_3.Dense_0.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_3.Dense_0.kernel", "write_shape": [384, 768], "chunk_shape": [384, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_3.Dense_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_3.Dense_1.kernel", "write_shape": [768, 384], "chunk_shape": [768, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_3.LayerNorm_0.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_3.LayerNorm_0.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_3.LayerNorm_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_3.LayerNorm_1.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_3.MultiHeadDotProductAttention_0.key.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_3.MultiHeadDotProductAttention_0.key.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_3.MultiHeadDotProductAttention_0.out.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_3.MultiHeadDotProductAttention_0.out.kernel", "write_shape": [8, 48, 384], "chunk_shape": [8, 48, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_3.MultiHeadDotProductAttention_0.query.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_3.MultiHeadDotProductAttention_0.query.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_3.MultiHeadDotProductAttention_0.value.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_3.MultiHeadDotProductAttention_0.value.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_4.Dense_0.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_4.Dense_0.kernel", "write_shape": [384, 768], "chunk_shape": [384, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_4.Dense_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_4.Dense_1.kernel", "write_shape": [768, 384], "chunk_shape": [768, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_4.LayerNorm_0.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_4.LayerNorm_0.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_4.LayerNorm_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_4.LayerNorm_1.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_4.MultiHeadDotProductAttention_0.key.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_4.MultiHeadDotProductAttention_0.key.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_4.MultiHeadDotProductAttention_0.out.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_4.MultiHeadDotProductAttention_0.out.kernel", "write_shape": [8, 48, 384], "chunk_shape": [8, 48, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_4.MultiHeadDotProductAttention_0.query.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_4.MultiHeadDotProductAttention_0.query.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_4.MultiHeadDotProductAttention_0.value.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_4.MultiHeadDotProductAttention_0.value.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_5.Dense_0.bias", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_5.Dense_0.kernel", "write_shape": [384, 768], "chunk_shape": [384, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_5.Dense_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_5.Dense_1.kernel", "write_shape": [768, 384], "chunk_shape": [768, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_5.LayerNorm_0.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_5.LayerNorm_0.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_5.LayerNorm_1.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_5.LayerNorm_1.scale", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_5.MultiHeadDotProductAttention_0.key.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_5.MultiHeadDotProductAttention_0.key.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_5.MultiHeadDotProductAttention_0.out.bias", "write_shape": [384], "chunk_shape": [384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_5.MultiHeadDotProductAttention_0.out.kernel", "write_shape": [8, 48, 384], "chunk_shape": [8, 48, 384], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_5.MultiHeadDotProductAttention_0.query.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_5.MultiHeadDotProductAttention_0.query.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_5.MultiHeadDotProductAttention_0.value.bias", "write_shape": [8, 48], "chunk_shape": [8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.params.TransformerBlock_5.MultiHeadDotProductAttention_0.value.kernel", "write_shape": [384, 8, 48], "chunk_shape": [384, 8, 48], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.1.count", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}]} \ No newline at end of file diff --git a/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/commit_success.txt b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/commit_success.txt new file mode 100644 index 0000000000000000000000000000000000000000..b7eda0354694e542caf140723bff8cd109ca9fd9 --- /dev/null +++ b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/commit_success.txt @@ -0,0 +1 @@ +Checkpoint commit was successful to /workspace/craftax-ReMDM-planner/wandb/run-20260812_110141-jogr4aj6/files/policies/99942400.orbax-checkpoint-tmp/default \ No newline at end of file diff --git a/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/d/22d785aa9155cf61ee8e16096ccc148b b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/d/22d785aa9155cf61ee8e16096ccc148b new file mode 100644 index 0000000000000000000000000000000000000000..e597b69422ecbf2b11368b9a82572a3edaccd61b Binary files /dev/null and b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/d/22d785aa9155cf61ee8e16096ccc148b differ diff --git a/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/manifest.ocdbt b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/manifest.ocdbt new file mode 100644 index 0000000000000000000000000000000000000000..a1c574e5be5a0144d441bfee314d46e421e5ad4c Binary files /dev/null and b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/manifest.ocdbt differ diff --git a/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/ocdbt.process_0/d/242f47a6c67d472ff6693fe34e8e8756 b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/ocdbt.process_0/d/242f47a6c67d472ff6693fe34e8e8756 new file mode 100644 index 0000000000000000000000000000000000000000..32aeb3b90d60b0a7055cd0d594f6225851c8357d Binary files /dev/null and b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/ocdbt.process_0/d/242f47a6c67d472ff6693fe34e8e8756 differ diff --git a/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/ocdbt.process_0/d/503349be2d11cb1bf04a17e7b3e2d1e0 b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/ocdbt.process_0/d/503349be2d11cb1bf04a17e7b3e2d1e0 new file mode 100644 index 0000000000000000000000000000000000000000..c4ed693cb602cfb3a991333cd8239a8f501b89e6 --- /dev/null +++ b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/ocdbt.process_0/d/503349be2d11cb1bf04a17e7b3e2d1e0 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:67d5902b59572ed15aed77012c89bf82acc1cc23331a6b224b13bfc31a06fbf3 +size 31543296 diff --git a/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/ocdbt.process_0/d/781b85e3ec447139f7d23160643a9c21 b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/ocdbt.process_0/d/781b85e3ec447139f7d23160643a9c21 new file mode 100644 index 0000000000000000000000000000000000000000..38896e85444fd944fb88faeb01b19b0fb0a71f30 --- /dev/null +++ b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/ocdbt.process_0/d/781b85e3ec447139f7d23160643a9c21 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f5f6f1d022f5693f32fa7297894e103ec2c0d6a32caa55545c39d183c5e1da39 +size 24133632 diff --git a/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/ocdbt.process_0/d/92adc334c11103c7741baed0d71e66cb b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/ocdbt.process_0/d/92adc334c11103c7741baed0d71e66cb new file mode 100644 index 0000000000000000000000000000000000000000..8b2f4090e48df36e7ce1d62b9d930dfb255ec825 --- /dev/null +++ b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/ocdbt.process_0/d/92adc334c11103c7741baed0d71e66cb @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f23556bf93954ef2519206c4d97eb486221a28867699ce96a0a33b45e773bcaa +size 45723648 diff --git a/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/ocdbt.process_0/d/a3b023497266923b5858e9b2f2738b95 b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/ocdbt.process_0/d/a3b023497266923b5858e9b2f2738b95 new file mode 100644 index 0000000000000000000000000000000000000000..baf92a08d803a1fe0265fb9fcecb9eaa1b3ffc5d Binary files /dev/null and b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/ocdbt.process_0/d/a3b023497266923b5858e9b2f2738b95 differ diff --git a/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/ocdbt.process_0/d/de5c2fdaa79aa09084def38a2c2dd9b4 b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/ocdbt.process_0/d/de5c2fdaa79aa09084def38a2c2dd9b4 new file mode 100644 index 0000000000000000000000000000000000000000..475627f873b85cf66ceea1581aac4c7f5c21c4a7 Binary files /dev/null and b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/ocdbt.process_0/d/de5c2fdaa79aa09084def38a2c2dd9b4 differ diff --git a/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/ocdbt.process_0/manifest.ocdbt b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/ocdbt.process_0/manifest.ocdbt new file mode 100644 index 0000000000000000000000000000000000000000..df3f4a3c0acc693067314c5b8afc6bb17996ea60 Binary files /dev/null and b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/1000000000/default/ocdbt.process_0/manifest.ocdbt differ diff --git a/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/resume_metadata.json b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/resume_metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..94d5ddad85fdfe66c521165c2af6997c66fcd956 --- /dev/null +++ b/checkpoints/offline/Craftax-Classic-Symbolic-v1-Offline-Diffusion-BC-100M/resume_metadata.json @@ -0,0 +1,78 @@ +{ + "mode": "offline", + "update_step": 1525, + "total_gradient_steps_completed": 97600, + "wandb_run_id": "jogr4aj6", + "config_snapshot": { + "ENV_NAME": "Craftax-Classic-Symbolic-v1", + "USE_OPTIMISTIC_RESETS": false, + "OPTIMISTIC_RESET_RATIO": 16, + "D_MODEL": 384, + "N_HEADS": 8, + "N_LAYERS": 6, + "D_FF": 768, + "OBS_ENCODER_LAYERS": 2, + "OBS_ENCODER_WIDTH": 768, + "DROPOUT_RATE": 0.1, + "PLAN_HORIZON": 32, + "DIFFUSION_SCHEDULE": "cosine", + "TRAIN_SIGMA": 0.0, + "LABEL_SMOOTHING": 0.0, + "DIFFUSION_STEPS": 15, + "DIFFUSION_STEPS_EVAL": 10, + "REMASK_STRATEGY": "rescale", + "ETA": 0.5, + "USE_LOOP": true, + "T_ON": 0.7, + "T_OFF": 0.3, + "TEMPERATURE": 0.5, + "TOP_P": 0.95, + "LR": 0.0003, + "MAX_GRAD_NORM": 1.0, + "LR_WARMUP_FRAMES": "1.048576e8", + "LR_WARMUP_STEPS": 1600, + "NUM_ENVS": 512, + "NUM_STEPS": 128, + "NUM_MINIBATCHES": 8, + "UPDATE_EPOCHS": 8, + "NUM_REPEATS": 1, + "OFFLINE_TOTAL_TIMESTEPS": 99942400, + "OFFLINE_NUM_UPDATES": null, + "COLLECT_TEMPERATURE": 1.0, + "RETURN_WEIGHT_CAP": 5.0, + "ONLINE_TOTAL_TIMESTEPS": 100000000.0, + "ONLINE_NUM_UPDATES": 1000, + "DAGGER_BETA_INIT": 1.0, + "DAGGER_BETA_FINAL": 0.344, + "DAGGER_BETA_DECAY": 0.95, + "DAGGER_BUFFER_CYCLES": 1.90735, + "DAGGER_BUFFER_MAX": 100000, + "DAGGER_TRAIN_PASSES": null, + "DAGGER_EXPERT_DETERMINISTIC": true, + "VAL_INTERVAL_FRAMES": 1000000.0, + "VAL_INTERVAL": 15, + "VAL_DIFFUSION_STEPS": 50, + "VAL_REPLAN_EVERY": 4, + "VAL_STEPS": 256, + "COLLECT_NUM_STEPS": 10000000, + "COLLECT_NUM_ENVS": 128, + "PPO_MODEL_TYPE": "ppo_rnn", + "LAYER_SIZE": 512, + "EVAL_STEPS": 10000, + "EVAL_NUM_ENVS": 32, + "SAVE_POLICY": true, + "RESUME_CHECKPOINT_PATH": null, + "RESUME_WANDB_RUN_ID": null, + "RESUME_STEP": null, + "SEED": 0, + "USE_WANDB": true, + "WANDB_PROJECT": "craftax-ReMDM-planner", + "WANDB_ENTITY": "myopic-planner", + "WANDB_DOWNLOAD_DIR": null, + "MODE": "offline", + "JIT": true, + "PPO_CHECKPOINT_PATH": "checkpoints/ppo_agents/Craftax-Classic-Symbolic-v1-PPO_RNN-1000M", + "NUM_UPDATES": 1525, + "MINIBATCH_SIZE": 6208 + } +} \ No newline at end of file