Upload checkpoint at step 1150

.gitattributes

@@ -163,3 +163,10 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 000000001100/state/ocdbt.process_0/d/90318887cdfd6808faab9d5ead608665 filter=lfs diff=lfs merge=lfs -text
 000000001100/state/ocdbt.process_0/d/94164f7ba18b735a5c8386a56fabfcc4 filter=lfs diff=lfs merge=lfs -text
 000000001100/state/ocdbt.process_0/d/ff2e1c1d66a79fdd2ed330759570b6cd filter=lfs diff=lfs merge=lfs -text
+000000001150/state/ocdbt.process_0/d/1a629ce1a2cc8b2f6848b9be07eca6d9 filter=lfs diff=lfs merge=lfs -text
+000000001150/state/ocdbt.process_0/d/2cb92043dd457ac974767ad06ea95c46 filter=lfs diff=lfs merge=lfs -text
+000000001150/state/ocdbt.process_0/d/3ab3d26952ffa97adb63487f53db0294 filter=lfs diff=lfs merge=lfs -text
+000000001150/state/ocdbt.process_0/d/5ecec41358a7f3b15658846aca57f609 filter=lfs diff=lfs merge=lfs -text
+000000001150/state/ocdbt.process_0/d/7d05cf6a3596821ca055798d5a5125b2 filter=lfs diff=lfs merge=lfs -text
+000000001150/state/ocdbt.process_0/d/b83c5c29103ceade3b60b2a8270e24c8 filter=lfs diff=lfs merge=lfs -text
+000000001150/state/ocdbt.process_0/d/f221fe550b0a946d90e812dfb276152c filter=lfs diff=lfs merge=lfs -text

000000001150/metadata.json

@@ -0,0 +1,136 @@
+{
+    "git_commit": "b1fea531a1bb067c5975282a447929a416793831",
+    "wandb_run_id": "ef3f1rcw",
+    "config": {
+        "seed": 42,
+        "notes": null,
+        "trainer": {
+            "run_dir": "local_data/runs/ablate-scales",
+            "checkpoint": {
+                "load_state": false,
+                "save_optimizer": false,
+                "interval": 50
+            },
+            "wandb": {
+                "project": "oloop-instruct",
+                "enabled": true
+            },
+            "huggingface": {
+                "private": false,
+                "enabled": true
+            },
+            "global_batch_size": 128,
+            "max_steps": 1000000,
+            "num_logit_iterations": 4,
+            "aux_loss_weight": 0.1,
+            "max_grad_norm": 1.0,
+            "debug": false,
+            "groups": {
+                "slow": {
+                    "_target_": "optax.adamw",
+                    "lr": 3e-05,
+                    "b1": 0.9,
+                    "b2": 0.95,
+                    "eps": 1e-06,
+                    "weight_decay": 0.0,
+                    "warmup_steps": 100,
+                    "stable_steps": 899900,
+                    "decay_steps": 100000,
+                    "min_lr_ratio": 0.0
+                },
+                "fast": {
+                    "_target_": "optax.contrib.muon",
+                    "lr": 0.0005,
+                    "beta": 0.9,
+                    "adam_b1": 0.9,
+                    "adam_b2": 0.95,
+                    "eps": 1e-06,
+                    "weight_decay": 0.01,
+                    "ns_steps": 5,
+                    "consistent_rms": 0.2,
+                    "nesterov": true,
+                    "warmup_steps": 100,
+                    "stable_steps": 899900,
+                    "decay_steps": 100000,
+                    "min_lr_ratio": 0.0
+                }
+            },
+            "_target_": "trainers.forte_trainer.ForteTrainer"
+        },
+        "data": {
+            "skip_batches": 0,
+            "dataset": {
+                "path": "aklein4/300k-horizons",
+                "split": "train",
+                "streaming": true
+            },
+            "collator": {
+                "_target_": "collators.horizon.HorizonCollator",
+                "tokenizer_url": "meta-llama/Llama-3.2-1B-Instruct",
+                "max_length": 1024,
+                "cluster_length": 64
+            }
+        },
+        "distributed": {
+            "initialize": "auto"
+        },
+        "mesh": {
+            "fsdp_axis_size": -1
+        },
+        "name": "ablate-scales",
+        "initialization": "aklein4/forte-init-instruct",
+        "initialization_step": 0,
+        "model": {
+            "sharding": {
+                "default": [],
+                "rules": [
+                    {
+                        "pattern": "^embed_tokens/embedding$",
+                        "spec": [
+                            "fsdp",
+                            null
+                        ]
+                    },
+                    {
+                        "pattern": "layers/mlp/(gate_proj|up_proj|down_proj)",
+                        "spec": [
+                            null,
+                            null,
+                            "fsdp"
+                        ]
+                    }
+                ]
+            },
+            "_target_": "models.forte.ForteModel",
+            "vocab_size": 128256,
+            "pad_token_id": -100,
+            "hidden_size": 2048,
+            "intermediate_size": 8192,
+            "num_hidden_layers": 16,
+            "num_attention_heads": 32,
+            "num_key_value_heads": 8,
+            "head_dim": 64,
+            "max_position_embeddings": 131072,
+            "rope": {
+                "theta": 500000.0,
+                "factor": 32.0,
+                "high_freq_factor": 4.0,
+                "low_freq_factor": 1.0,
+                "original_context_len": 8192
+            },
+            "rms_norm_eps": 1e-05,
+            "compute_dtype": "bfloat16",
+            "gradient_checkpointing": true,
+            "attention_backend": "auto",
+            "num_fast_vectors": 512,
+            "num_fast_heads": 16,
+            "base_lr": 0.0001,
+            "pool_size": 512,
+            "grad_rms_eps": 1e-12
+        }
+    },
+    "step": 1150,
+    "checkpoint_items": [
+        "params"
+    ]
+}

000000001150/state/_CHECKPOINT_METADATA

@@ -0,0 +1 @@
+{"item_handlers": "orbax.checkpoint._src.handlers.standard_checkpoint_handler.StandardCheckpointHandler", "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1791324056027598416, "commit_timestamp_nsecs": 1791324105073378426, "custom_metadata": {}}

000000001150/state/_METADATA

@@ -0,0 +1 @@
+{"tree_metadata": {"('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', 'layers', 'input_layernorm', 'scale')": {"key_metadata": [{"key": "params", "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', 'layers', 'mlp', 'down_proj', 'kernel')": {"key_metadata": [{"key": "params", "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', 'layers', 'mlp', 'fast_cartridge', 'forward_gate_proj', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "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', 'layers', 'mlp', 'fast_cartridge', 'forward_gate_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "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', 'layers', 'mlp', 'fast_cartridge', 'k_gate')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "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', 'layers', 'mlp', 'fast_cartridge', 'k_up')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "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', 'layers', 'mlp', 'fast_cartridge', 'log_lr')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "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', 'layers', 'mlp', 'fast_cartridge', 'o_proj')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "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', 'layers', 'mlp', 'fast_cartridge', 'q_gate_proj', 'bias')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "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', 'layers', 'mlp', 'fast_cartridge', 'q_gate_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "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', 'layers', 'mlp', 'fast_cartridge', 'q_up_proj', 'kernel')": {"key_metadata": [{"key": "params", "key_type": 2}, {"key": "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', 'layers', 'mlp', 'gate_proj', 'kernel')": {"key_metadata": [{"key": "params", "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', 'layers', 'mlp', 'up_proj', 'kernel')": {"key_metadata": [{"key": "params", "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', 'layers', 'post_attention_layernorm', 'scale')": {"key_metadata": [{"key": "params", "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', 'layers', 'self_attn', 'k_proj', 'kernel')": {"key_metadata": [{"key": "params", "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', 'layers', 'self_attn', 'o_proj', 'kernel')": {"key_metadata": [{"key": "params", "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', 'layers', 'self_attn', 'q_proj', 'kernel')": {"key_metadata": [{"key": "params", "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', 'layers', 'self_attn', 'v_proj', 'kernel')": {"key_metadata": [{"key": "params", "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', '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}}}, "use_ocdbt": true, "use_zarr3": false, "store_array_data_equal_to_fill_value": true, "custom_metadata": null}

000000001150/state/commit_success.txt

@@ -0,0 +1 @@
+Checkpoint commit was successful to /home/aklein4/outer-loop-jax/local_data/runs/ablate-scales/checkpoints/000000001150/state

000000001150/state/ocdbt.process_0/d/1a629ce1a2cc8b2f6848b9be07eca6d9

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0745fe78642f6c71ebc68919a0bd892fa4a5a531a5be8c96766d7200bc092053
+size 2982506496

000000001150/state/ocdbt.process_0/d/2cb92043dd457ac974767ad06ea95c46

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:45cea9fea8129a79e5f99e054a8f51d62d7bca426cca8f282c9645ae58519eb5
+size 1492459520

000000001150/state/ocdbt.process_0/d/3ab3d26952ffa97adb63487f53db0294

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:deb130d23591f118fbeb4d9122194a2c320bbfafd260f76a8ec66c1377e7d405
+size 121833

000000001150/state/ocdbt.process_0/d/5ecec41358a7f3b15658846aca57f609

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:60f10e5c974e49de1032433a0638b98390c97379d905dfec788857037318472b
+size 62226432

000000001150/state/ocdbt.process_0/d/7d05cf6a3596821ca055798d5a5125b2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d795b5cee0057fd98cd2cce9db83afa5ddb908186fb50211dd8bb409960b5cfb
+size 123302

000000001150/state/ocdbt.process_0/d/b83c5c29103ceade3b60b2a8270e24c8

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:38476b35c7c88f1eab899b10011b4bc8f2b1c1117cdc005143554bb1d586e053
+size 998309888

000000001150/state/ocdbt.process_0/d/f221fe550b0a946d90e812dfb276152c

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e3cb9d64b8b9f98ed04429951d0815575ce862e04700b2c999c040d1058dbbb
+size 2057533