Upload checkpoint at step 250
Browse files- .gitattributes +4 -0
- 000000000250/metadata.json +140 -0
- 000000000250/state/_CHECKPOINT_METADATA +1 -0
- 000000000250/state/_METADATA +1 -0
- 000000000250/state/commit_success.txt +1 -0
- 000000000250/state/d/b8294c1b2d9630e9f56e8f90c3d828b4 +0 -0
- 000000000250/state/manifest.ocdbt +0 -0
- 000000000250/state/ocdbt.process_0/d/2340d2c7cfd4acd8c949694a267ee6cc +3 -0
- 000000000250/state/ocdbt.process_0/d/2e7c85f67339925847d67ee6c78580c8 +0 -0
- 000000000250/state/ocdbt.process_0/d/4e5ff02b5a3d33e7fa8882775a7a86f5 +3 -0
- 000000000250/state/ocdbt.process_0/d/754627abb7b8e1349c3c08aea6c6d392 +0 -0
- 000000000250/state/ocdbt.process_0/d/75622e219b800289e24991332bb94d90 +3 -0
- 000000000250/state/ocdbt.process_0/d/7d84fc2291d1600c7b7b4ac7d0cd7437 +0 -0
- 000000000250/state/ocdbt.process_0/d/d127321b1c906483f2025cb585c10276 +0 -0
- 000000000250/state/ocdbt.process_0/d/e62f913d8357a81c0ad6010c2342c3d7 +3 -0
- 000000000250/state/ocdbt.process_0/manifest.ocdbt +0 -0
.gitattributes
CHANGED
|
@@ -49,3 +49,7 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
| 49 |
000000000200/state/ocdbt.process_0/d/56e44b5be0a5e553830c4493a29ad89b filter=lfs diff=lfs merge=lfs -text
|
| 50 |
000000000200/state/ocdbt.process_0/d/b91441669ee779ed2646f5c866c2f5a0 filter=lfs diff=lfs merge=lfs -text
|
| 51 |
000000000200/state/ocdbt.process_0/d/fe2ad0f87dc91bf28ad4a97ec69765b2 filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 49 |
000000000200/state/ocdbt.process_0/d/56e44b5be0a5e553830c4493a29ad89b filter=lfs diff=lfs merge=lfs -text
|
| 50 |
000000000200/state/ocdbt.process_0/d/b91441669ee779ed2646f5c866c2f5a0 filter=lfs diff=lfs merge=lfs -text
|
| 51 |
000000000200/state/ocdbt.process_0/d/fe2ad0f87dc91bf28ad4a97ec69765b2 filter=lfs diff=lfs merge=lfs -text
|
| 52 |
+
000000000250/state/ocdbt.process_0/d/2340d2c7cfd4acd8c949694a267ee6cc filter=lfs diff=lfs merge=lfs -text
|
| 53 |
+
000000000250/state/ocdbt.process_0/d/4e5ff02b5a3d33e7fa8882775a7a86f5 filter=lfs diff=lfs merge=lfs -text
|
| 54 |
+
000000000250/state/ocdbt.process_0/d/75622e219b800289e24991332bb94d90 filter=lfs diff=lfs merge=lfs -text
|
| 55 |
+
000000000250/state/ocdbt.process_0/d/e62f913d8357a81c0ad6010c2342c3d7 filter=lfs diff=lfs merge=lfs -text
|
000000000250/metadata.json
ADDED
|
@@ -0,0 +1,140 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"git_commit": "d279fca943d2a70a6da4b33e6d630462538b7f84",
|
| 3 |
+
"wandb_run_id": "464sj2t5",
|
| 4 |
+
"config": {
|
| 5 |
+
"seed": 42,
|
| 6 |
+
"notes": null,
|
| 7 |
+
"trainer": {
|
| 8 |
+
"run_dir": "local_data/runs/delta",
|
| 9 |
+
"checkpoint": {
|
| 10 |
+
"load_state": false,
|
| 11 |
+
"save_optimizer": false,
|
| 12 |
+
"interval": 50
|
| 13 |
+
},
|
| 14 |
+
"wandb": {
|
| 15 |
+
"project": "oloop-instruct",
|
| 16 |
+
"enabled": true
|
| 17 |
+
},
|
| 18 |
+
"huggingface": {
|
| 19 |
+
"private": false,
|
| 20 |
+
"enabled": true
|
| 21 |
+
},
|
| 22 |
+
"global_batch_size": 128,
|
| 23 |
+
"max_steps": 1000000,
|
| 24 |
+
"num_logit_iterations": 4,
|
| 25 |
+
"aux_loss_weight": 0.1,
|
| 26 |
+
"max_grad_norm": 1.0,
|
| 27 |
+
"debug": false,
|
| 28 |
+
"groups": {
|
| 29 |
+
"slow": {
|
| 30 |
+
"_target_": "optax.adamw",
|
| 31 |
+
"lr": 3e-05,
|
| 32 |
+
"b1": 0.9,
|
| 33 |
+
"b2": 0.95,
|
| 34 |
+
"eps": 1e-06,
|
| 35 |
+
"weight_decay": 0.0,
|
| 36 |
+
"warmup_steps": 100,
|
| 37 |
+
"stable_steps": 899900,
|
| 38 |
+
"decay_steps": 100000,
|
| 39 |
+
"min_lr_ratio": 0.0
|
| 40 |
+
},
|
| 41 |
+
"fast": {
|
| 42 |
+
"_target_": "optax.contrib.muon",
|
| 43 |
+
"lr": 0.0005,
|
| 44 |
+
"beta": 0.9,
|
| 45 |
+
"adam_b1": 0.9,
|
| 46 |
+
"adam_b2": 0.95,
|
| 47 |
+
"eps": 1e-06,
|
| 48 |
+
"weight_decay": 0.01,
|
| 49 |
+
"ns_steps": 5,
|
| 50 |
+
"consistent_rms": 0.2,
|
| 51 |
+
"nesterov": true,
|
| 52 |
+
"warmup_steps": 100,
|
| 53 |
+
"stable_steps": 899900,
|
| 54 |
+
"decay_steps": 100000,
|
| 55 |
+
"min_lr_ratio": 0.0
|
| 56 |
+
}
|
| 57 |
+
},
|
| 58 |
+
"_target_": "trainers.forte_trainer.ForteTrainer"
|
| 59 |
+
},
|
| 60 |
+
"data": {
|
| 61 |
+
"skip_batches": 0,
|
| 62 |
+
"dataset": {
|
| 63 |
+
"path": "aklein4/300k-horizons",
|
| 64 |
+
"split": "train",
|
| 65 |
+
"streaming": true
|
| 66 |
+
},
|
| 67 |
+
"collator": {
|
| 68 |
+
"_target_": "collators.horizon.HorizonCollator",
|
| 69 |
+
"tokenizer_url": "meta-llama/Llama-3.2-1B-Instruct",
|
| 70 |
+
"max_length": 1024,
|
| 71 |
+
"cluster_length": 64
|
| 72 |
+
}
|
| 73 |
+
},
|
| 74 |
+
"distributed": {
|
| 75 |
+
"initialize": "auto"
|
| 76 |
+
},
|
| 77 |
+
"mesh": {
|
| 78 |
+
"fsdp_axis_size": -1
|
| 79 |
+
},
|
| 80 |
+
"name": "delta",
|
| 81 |
+
"initialization": "aklein4/forte-init-delta",
|
| 82 |
+
"initialization_step": 0,
|
| 83 |
+
"model": {
|
| 84 |
+
"sharding": {
|
| 85 |
+
"default": [],
|
| 86 |
+
"rules": [
|
| 87 |
+
{
|
| 88 |
+
"pattern": "^embed_tokens/embedding$",
|
| 89 |
+
"spec": [
|
| 90 |
+
"fsdp",
|
| 91 |
+
null
|
| 92 |
+
]
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"pattern": "layers/mlp/(gate_proj|up_proj|down_proj)",
|
| 96 |
+
"spec": [
|
| 97 |
+
null,
|
| 98 |
+
null,
|
| 99 |
+
"fsdp"
|
| 100 |
+
]
|
| 101 |
+
}
|
| 102 |
+
]
|
| 103 |
+
},
|
| 104 |
+
"_target_": "models.forte.ForteModel",
|
| 105 |
+
"vocab_size": 128256,
|
| 106 |
+
"pad_token_id": -100,
|
| 107 |
+
"hidden_size": 2048,
|
| 108 |
+
"intermediate_size": 8192,
|
| 109 |
+
"num_hidden_layers": 16,
|
| 110 |
+
"num_attention_heads": 32,
|
| 111 |
+
"num_key_value_heads": 8,
|
| 112 |
+
"head_dim": 64,
|
| 113 |
+
"max_position_embeddings": 131072,
|
| 114 |
+
"rope": {
|
| 115 |
+
"theta": 500000.0,
|
| 116 |
+
"factor": 32.0,
|
| 117 |
+
"high_freq_factor": 4.0,
|
| 118 |
+
"low_freq_factor": 1.0,
|
| 119 |
+
"original_context_len": 8192
|
| 120 |
+
},
|
| 121 |
+
"rms_norm_eps": 1e-05,
|
| 122 |
+
"compute_dtype": "bfloat16",
|
| 123 |
+
"gradient_checkpointing": true,
|
| 124 |
+
"attention_backend": "auto",
|
| 125 |
+
"num_fast_vectors": 512,
|
| 126 |
+
"num_fast_heads": 16,
|
| 127 |
+
"base_lr": 0.0001,
|
| 128 |
+
"grad_rms_eps": 1e-12,
|
| 129 |
+
"delta_alpha": 0.25,
|
| 130 |
+
"num_output_layers": 10,
|
| 131 |
+
"num_bidirectional_layers": 6,
|
| 132 |
+
"mixer_kernel_size": 5,
|
| 133 |
+
"pad_attention_bias_value": -1000.0
|
| 134 |
+
}
|
| 135 |
+
},
|
| 136 |
+
"step": 250,
|
| 137 |
+
"checkpoint_items": [
|
| 138 |
+
"params"
|
| 139 |
+
]
|
| 140 |
+
}
|
000000000250/state/_CHECKPOINT_METADATA
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"item_handlers": "orbax.checkpoint._src.handlers.standard_checkpoint_handler.StandardCheckpointHandler", "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1789847539154825899, "commit_timestamp_nsecs": 1789847606575041596, "custom_metadata": {}}
|
000000000250/state/_METADATA
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"tree_metadata": {"('params', 'backbone_layers', 'input_layernorm', 'scale')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "backbone_layers", "key_type": 2}, {"key": "input_layernorm", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'backbone_layers', 'mlp', 'down_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "backbone_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "down_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'backbone_layers', 'mlp', 'fast_cartridge', 'delta_log_lr')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "backbone_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "delta_log_lr", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'backbone_layers', 'mlp', 'fast_cartridge', 'forward_gate_proj', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "backbone_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "forward_gate_proj", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'backbone_layers', 'mlp', 'fast_cartridge', 'forward_gate_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "backbone_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "forward_gate_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'backbone_layers', 'mlp', 'fast_cartridge', 'k_gate')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "backbone_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "k_gate", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'backbone_layers', 'mlp', 'fast_cartridge', 'k_up')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "backbone_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "k_up", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'backbone_layers', 'mlp', 'fast_cartridge', 'log_lr')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "backbone_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "log_lr", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'backbone_layers', 'mlp', 'fast_cartridge', 'o_proj')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "backbone_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "o_proj", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'backbone_layers', 'mlp', 'fast_cartridge', 'q_gate_proj', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "backbone_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "q_gate_proj", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'backbone_layers', 'mlp', 'fast_cartridge', 'q_gate_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "backbone_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "q_gate_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'backbone_layers', 'mlp', 'fast_cartridge', 'q_up_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "backbone_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "q_up_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'backbone_layers', 'mlp', 'fast_cartridge', 'update_rule', 'delta_gate_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "backbone_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "update_rule", "key_type": 2}, {"key": "delta_gate_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'backbone_layers', 'mlp', 'fast_cartridge', 'update_rule', 'delta_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "backbone_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "update_rule", "key_type": 2}, {"key": "delta_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'backbone_layers', 'mlp', 'fast_cartridge', 'update_rule', 'gradient_gate_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "backbone_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "update_rule", "key_type": 2}, {"key": "gradient_gate_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'backbone_layers', 'mlp', 'gate_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "backbone_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "gate_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'backbone_layers', 'mlp', 'up_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "backbone_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "up_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'backbone_layers', 'post_attention_layernorm', 'scale')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "backbone_layers", "key_type": 2}, {"key": "post_attention_layernorm", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'backbone_layers', 'self_attn', 'k_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "backbone_layers", "key_type": 2}, {"key": "self_attn", "key_type": 2}, {"key": "k_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'backbone_layers', 'self_attn', 'o_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "backbone_layers", "key_type": 2}, {"key": "self_attn", "key_type": 2}, {"key": "o_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'backbone_layers', 'self_attn', 'q_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "backbone_layers", "key_type": 2}, {"key": "self_attn", "key_type": 2}, {"key": "q_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'backbone_layers', 'self_attn', 'v_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "backbone_layers", "key_type": 2}, {"key": "self_attn", "key_type": 2}, {"key": "v_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'bidirectional_head', 'embedding_state_scale')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "bidirectional_head", "key_type": 2}, {"key": "embedding_state_scale", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'bidirectional_head', 'embedding_state_shift')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "bidirectional_head", "key_type": 2}, {"key": "embedding_state_shift", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'bidirectional_head', 'layers', 'input_layernorm', 'scale')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "bidirectional_head", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "input_layernorm", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'bidirectional_head', 'layers', 'mlp', 'down_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "bidirectional_head", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "down_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'bidirectional_head', 'layers', 'mlp', 'gate_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "bidirectional_head", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "gate_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'bidirectional_head', 'layers', 'mlp', 'up_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "bidirectional_head", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "up_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'bidirectional_head', 'layers', 'post_attention_layernorm', 'scale')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "bidirectional_head", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "post_attention_layernorm", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'bidirectional_head', 'layers', 'self_attn', 'k_mixer', 'conv', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "bidirectional_head", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "self_attn", "key_type": 2}, {"key": "k_mixer", "key_type": 2}, {"key": "conv", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'bidirectional_head', 'layers', 'self_attn', 'k_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "bidirectional_head", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "self_attn", "key_type": 2}, {"key": "k_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'bidirectional_head', 'layers', 'self_attn', 'o_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "bidirectional_head", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "self_attn", "key_type": 2}, {"key": "o_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'bidirectional_head', 'layers', 'self_attn', 'q_mixer', 'conv', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "bidirectional_head", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "self_attn", "key_type": 2}, {"key": "q_mixer", "key_type": 2}, {"key": "conv", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'bidirectional_head', 'layers', 'self_attn', 'q_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "bidirectional_head", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "self_attn", "key_type": 2}, {"key": "q_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'bidirectional_head', 'layers', 'self_attn', 'v_mixer', 'conv', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "bidirectional_head", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "self_attn", "key_type": 2}, {"key": "v_mixer", "key_type": 2}, {"key": "conv", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'bidirectional_head', 'layers', 'self_attn', 'v_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "bidirectional_head", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "self_attn", "key_type": 2}, {"key": "v_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'bidirectional_head', 'norm', 'scale')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "bidirectional_head", "key_type": 2}, {"key": "norm", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'embed_tokens', 'embedding')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "embed_tokens", "key_type": 2}, {"key": "embedding", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'lm_head', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "lm_head", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'norm', 'scale')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "norm", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'output_layers', 'input_layernorm', 'scale')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "output_layers", "key_type": 2}, {"key": "input_layernorm", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'output_layers', 'mlp', 'down_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "output_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "down_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'output_layers', 'mlp', 'fast_cartridge', 'delta_log_lr')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "output_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "delta_log_lr", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'output_layers', 'mlp', 'fast_cartridge', 'forward_gate_proj', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "output_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "forward_gate_proj", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'output_layers', 'mlp', 'fast_cartridge', 'forward_gate_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "output_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "forward_gate_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'output_layers', 'mlp', 'fast_cartridge', 'k_gate')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "output_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "k_gate", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'output_layers', 'mlp', 'fast_cartridge', 'k_up')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "output_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "k_up", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'output_layers', 'mlp', 'fast_cartridge', 'log_lr')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "output_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "log_lr", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'output_layers', 'mlp', 'fast_cartridge', 'o_proj')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "output_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "o_proj", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'output_layers', 'mlp', 'fast_cartridge', 'q_gate_proj', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "output_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "q_gate_proj", "key_type": 2}, {"key": "bias", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'output_layers', 'mlp', 'fast_cartridge', 'q_gate_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "output_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "q_gate_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'output_layers', 'mlp', 'fast_cartridge', 'q_up_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "output_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "q_up_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'output_layers', 'mlp', 'fast_cartridge', 'update_rule', 'delta_gate_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "output_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "update_rule", "key_type": 2}, {"key": "delta_gate_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'output_layers', 'mlp', 'fast_cartridge', 'update_rule', 'delta_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "output_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "update_rule", "key_type": 2}, {"key": "delta_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'output_layers', 'mlp', 'fast_cartridge', 'update_rule', 'gradient_gate_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "output_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "fast_cartridge", "key_type": 2}, {"key": "update_rule", "key_type": 2}, {"key": "gradient_gate_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'output_layers', 'mlp', 'gate_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "output_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "gate_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'output_layers', 'mlp', 'up_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "output_layers", "key_type": 2}, {"key": "mlp", "key_type": 2}, {"key": "up_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'output_layers', 'post_attention_layernorm', 'scale')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "output_layers", "key_type": 2}, {"key": "post_attention_layernorm", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'output_layers', 'self_attn', 'k_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "output_layers", "key_type": 2}, {"key": "self_attn", "key_type": 2}, {"key": "k_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'output_layers', 'self_attn', 'o_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "output_layers", "key_type": 2}, {"key": "self_attn", "key_type": 2}, {"key": "o_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'output_layers', 'self_attn', 'q_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "output_layers", "key_type": 2}, {"key": "self_attn", "key_type": 2}, {"key": "q_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}, "('params', 'output_layers', 'self_attn', 'v_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "output_layers", "key_type": 2}, {"key": "self_attn", "key_type": 2}, {"key": "v_proj", "key_type": 2}, {"key": "kernel", "key_type": 2}], "value_metadata": {"value_type": "np.ndarray", "skip_deserialize": false}}}, "use_ocdbt": true, "use_zarr3": false, "store_array_data_equal_to_fill_value": true, "custom_metadata": null}
|
000000000250/state/commit_success.txt
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
Checkpoint commit was successful to /home/aklein4/outer-loop-jax/local_data/runs/delta/checkpoints/000000000250/state
|
000000000250/state/d/b8294c1b2d9630e9f56e8f90c3d828b4
ADDED
|
Binary file (2.95 kB). View file
|
|
|
000000000250/state/manifest.ocdbt
ADDED
|
Binary file (117 Bytes). View file
|
|
|
000000000250/state/ocdbt.process_0/d/2340d2c7cfd4acd8c949694a267ee6cc
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ad8077bfab3553c9fd315630aaeb1d81ada2b17760e2ae07a4c942762b65f309
|
| 3 |
+
size 129769472
|
000000000250/state/ocdbt.process_0/d/2e7c85f67339925847d67ee6c78580c8
ADDED
|
Binary file (205 Bytes). View file
|
|
|
000000000250/state/ocdbt.process_0/d/4e5ff02b5a3d33e7fa8882775a7a86f5
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2d02fb2f87d3b4199137164936e4b6ab1cf0b3000d434c90dedaf5660432eb58
|
| 3 |
+
size 2486300672
|
000000000250/state/ocdbt.process_0/d/754627abb7b8e1349c3c08aea6c6d392
ADDED
|
Binary file (43.2 kB). View file
|
|
|
000000000250/state/ocdbt.process_0/d/75622e219b800289e24991332bb94d90
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:76fae544e256ee5b7b91e29544e504c90a66f8953023d80f4ba9a71583c90822
|
| 3 |
+
size 2488643584
|
000000000250/state/ocdbt.process_0/d/7d84fc2291d1600c7b7b4ac7d0cd7437
ADDED
|
Binary file (2.94 kB). View file
|
|
|
000000000250/state/ocdbt.process_0/d/d127321b1c906483f2025cb585c10276
ADDED
|
Binary file (1.05 kB). View file
|
|
|
000000000250/state/ocdbt.process_0/d/e62f913d8357a81c0ad6010c2342c3d7
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f7abe31f4f10715d60e5f6e1041be1c002d0542159b7212d07a70e95c54be1fc
|
| 3 |
+
size 2240598016
|
000000000250/state/ocdbt.process_0/manifest.ocdbt
ADDED
|
Binary file (357 Bytes). View file
|
|
|