add sft/charter/checkpoint-48 (copied from jbostock/scimt-dispatch-models-v1)

sft/charter/checkpoint-48/chat_template.jinja

@@ -0,0 +1,47 @@
+{{ bos_token }}
+{%- if messages[0]['role'] == 'system' -%}
+    {%- if messages[0]['content'] is string -%}
+        {%- set first_user_prefix = messages[0]['content'] + '
+
+' -%}
+    {%- else -%}
+        {%- set first_user_prefix = messages[0]['content'][0]['text'] + '
+
+' -%}
+    {%- endif -%}
+    {%- set loop_messages = messages[1:] -%}
+{%- else -%}
+    {%- set first_user_prefix = "" -%}
+    {%- set loop_messages = messages -%}
+{%- endif -%}
+{%- for message in loop_messages -%}
+    {%- if (message['role'] == 'user') != (loop.index0 % 2 == 0) -%}
+        {{ raise_exception("Conversation roles must alternate user/assistant/user/assistant/...") }}
+    {%- endif -%}
+    {%- if (message['role'] == 'assistant') -%}
+        {%- set role = "model" -%}
+    {%- else -%}
+        {%- set role = message['role'] -%}
+    {%- endif -%}
+    {{ '<start_of_turn>' + role + '
+' + (first_user_prefix if loop.first else "") }}
+    {%- if message['content'] is string -%}
+        {{ message['content'] | trim }}
+    {%- elif message['content'] is iterable -%}
+        {%- for item in message['content'] -%}
+            {%- if item['type'] == 'image' -%}
+                {{ '<start_of_image>' }}
+            {%- elif item['type'] == 'text' -%}
+                {{ item['text'] | trim }}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- else -%}
+        {{ raise_exception("Invalid content type") }}
+    {%- endif -%}
+    {{ '<end_of_turn>
+' }}
+{%- endfor -%}
+{%- if add_generation_prompt -%}
+    {{'<start_of_turn>model
+'}}
+{%- endif -%}

sft/charter/checkpoint-48/config.json

@@ -0,0 +1,125 @@
+{
+  "architectures": [
+    "Gemma3ForConditionalGeneration"
+  ],
+  "boi_token_index": 255999,
+  "bos_token_id": 2,
+  "dtype": "bfloat16",
+  "eoi_token_index": 256000,
+  "eos_token_id": 1,
+  "image_token_index": 262144,
+  "initializer_range": 0.02,
+  "mm_tokens_per_image": 256,
+  "model_type": "gemma3",
+  "pad_token_id": 0,
+  "text_config": {
+    "_sliding_window_pattern": 6,
+    "attention_bias": false,
+    "attention_dropout": 0.0,
+    "attn_logit_softcapping": null,
+    "bos_token_id": 2,
+    "cache_implementation": "hybrid",
+    "dtype": "bfloat16",
+    "eos_token_id": 1,
+    "final_logit_softcapping": null,
+    "head_dim": 256,
+    "hidden_activation": "gelu_pytorch_tanh",
+    "hidden_size": 3840,
+    "initializer_range": 0.02,
+    "intermediate_size": 15360,
+    "layer_types": [
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention"
+    ],
+    "max_position_embeddings": 131072,
+    "model_type": "gemma3_text",
+    "num_attention_heads": 16,
+    "num_hidden_layers": 48,
+    "num_key_value_heads": 8,
+    "pad_token_id": 0,
+    "query_pre_attn_scalar": 256,
+    "rms_norm_eps": 1e-06,
+    "rope_parameters": {
+      "full_attention": {
+        "factor": 8.0,
+        "rope_theta": 1000000.0,
+        "rope_type": "linear"
+      },
+      "sliding_attention": {
+        "rope_theta": 10000.0,
+        "rope_type": "default"
+      }
+    },
+    "sliding_window": 1024,
+    "sliding_window_pattern": 6,
+    "tie_word_embeddings": true,
+    "use_bidirectional_attention": false,
+    "use_cache": false,
+    "vocab_size": 262208
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.9.0",
+  "unsloth_fixed": true,
+  "use_cache": false,
+  "vision_config": {
+    "attention_dropout": 0.0,
+    "dtype": "bfloat16",
+    "hidden_act": "gelu_pytorch_tanh",
+    "hidden_size": 1152,
+    "image_size": 896,
+    "intermediate_size": 4304,
+    "layer_norm_eps": 1e-06,
+    "model_type": "siglip_vision_model",
+    "num_attention_heads": 16,
+    "num_channels": 3,
+    "num_hidden_layers": 27,
+    "patch_size": 14,
+    "vision_use_head": false
+  }
+}

sft/charter/checkpoint-48/generation_config.json

@@ -0,0 +1,13 @@
+{
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "do_sample": true,
+  "eos_token_id": [
+    1,
+    106
+  ],
+  "pad_token_id": 0,
+  "top_k": 64,
+  "top_p": 0.95,
+  "transformers_version": "5.9.0"
+}

sft/charter/checkpoint-48/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1c6062a4054cad019944636763daa7e8a35862c1cb74789fdb2d1af370c3bab2
+size 26388552360

sft/charter/checkpoint-48/preprocessor_config.json

@@ -0,0 +1,29 @@
+{
+  "do_convert_rgb": null,
+  "do_normalize": true,
+  "do_pan_and_scan": null,
+  "do_rescale": true,
+  "do_resize": true,
+  "image_mean": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "image_processor_type": "Gemma3ImageProcessor",
+  "image_seq_length": 256,
+  "image_std": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "pan_and_scan_max_num_crops": null,
+  "pan_and_scan_min_crop_size": null,
+  "pan_and_scan_min_ratio_to_activate": null,
+  "processor_class": "Gemma3Processor",
+  "resample": 2,
+  "rescale_factor": 0.00392156862745098,
+  "size": {
+    "height": 896,
+    "width": 896
+  }
+}

sft/charter/checkpoint-48/processor_config.json

@@ -0,0 +1,4 @@
+{
+  "image_seq_length": 256,
+  "processor_class": "Gemma3Processor"
+}

sft/charter/checkpoint-48/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726
+size 33384567

sft/charter/checkpoint-48/tokenizer_config.json

@@ -0,0 +1,26 @@
+{
+  "backend": "tokenizers",
+  "boi_token": "<start_of_image>",
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eoi_token": "<end_of_image>",
+  "eos_token": "<eos>",
+  "image_token": "<image_soft_token>",
+  "is_local": true,
+  "local_files_only": false,
+  "mask_token": "<mask>",
+  "model_max_length": 131072,
+  "model_specific_special_tokens": {
+    "boi_token": "<start_of_image>",
+    "eoi_token": "<end_of_image>",
+    "image_token": "<image_soft_token>"
+  },
+  "pad_token": "<pad>",
+  "padding_side": "left",
+  "processor_class": "Gemma3Processor",
+  "sp_model_kwargs": null,
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}

sft/charter/checkpoint-48/tokens_state.json

@@ -0,0 +1 @@
+{"total": 100646912, "trainable": 62666372}

sft/charter/checkpoint-48/trainer_state.json

@@ -0,0 +1,706 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.08533333333333333,
+  "eval_steps": 500,
+  "global_step": 48,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0017777777777777779,
+      "grad_norm": 3.078125,
+      "learning_rate": 0.0,
+      "loss": 0.9302978515625,
+      "memory/device_reserved (GiB)": 78.29,
+      "memory/max_active (GiB)": 68.81,
+      "memory/max_allocated (GiB)": 68.81,
+      "ppl": 2.53526,
+      "step": 1,
+      "tokens/total": 2097152,
+      "tokens/train_per_sec_per_gpu": 428.25,
+      "tokens/trainable": 1308440
+    },
+    {
+      "epoch": 0.0035555555555555557,
+      "grad_norm": 3.21875,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.9635009765625,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.62086,
+      "step": 2,
+      "tokens/total": 4194304,
+      "tokens/train_per_sec_per_gpu": 403.69,
+      "tokens/trainable": 2602766
+    },
+    {
+      "epoch": 0.005333333333333333,
+      "grad_norm": 2.65625,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.92724609375,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.52754,
+      "step": 3,
+      "tokens/total": 6291456,
+      "tokens/train_per_sec_per_gpu": 426.1,
+      "tokens/trainable": 3908027
+    },
+    {
+      "epoch": 0.0071111111111111115,
+      "grad_norm": 1.90625,
+      "learning_rate": 1e-05,
+      "loss": 0.8570556640625,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.35621,
+      "step": 4,
+      "tokens/total": 8388608,
+      "tokens/train_per_sec_per_gpu": 427.79,
+      "tokens/trainable": 5196384
+    },
+    {
+      "epoch": 0.008888888888888889,
+      "grad_norm": 11.75,
+      "learning_rate": 9.98903822616921e-06,
+      "loss": 0.955810546875,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.60078,
+      "step": 5,
+      "tokens/total": 10485760,
+      "tokens/train_per_sec_per_gpu": 433.68,
+      "tokens/trainable": 6493998
+    },
+    {
+      "epoch": 0.010666666666666666,
+      "grad_norm": 5.09375,
+      "learning_rate": 9.956206309337067e-06,
+      "loss": 0.91552734375,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.49809,
+      "step": 6,
+      "tokens/total": 12582912,
+      "tokens/train_per_sec_per_gpu": 436.32,
+      "tokens/trainable": 7808080
+    },
+    {
+      "epoch": 0.012444444444444444,
+      "grad_norm": 3.125,
+      "learning_rate": 9.901664203302126e-06,
+      "loss": 0.8734130859375,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.39507,
+      "step": 7,
+      "tokens/total": 14680064,
+      "tokens/train_per_sec_per_gpu": 456.04,
+      "tokens/trainable": 9105408
+    },
+    {
+      "epoch": 0.014222222222222223,
+      "grad_norm": 1.65625,
+      "learning_rate": 9.825677631722436e-06,
+      "loss": 0.8377685546875,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.3112,
+      "step": 8,
+      "tokens/total": 16777216,
+      "tokens/train_per_sec_per_gpu": 458.13,
+      "tokens/trainable": 10422780
+    },
+    {
+      "epoch": 0.016,
+      "grad_norm": 1.3671875,
+      "learning_rate": 9.728616793536588e-06,
+      "loss": 0.845947265625,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.33018,
+      "step": 9,
+      "tokens/total": 18874368,
+      "tokens/train_per_sec_per_gpu": 446.7,
+      "tokens/trainable": 11726810
+    },
+    {
+      "epoch": 0.017777777777777778,
+      "grad_norm": 1.0703125,
+      "learning_rate": 9.610954559391704e-06,
+      "loss": 0.8197021484375,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.26982,
+      "step": 10,
+      "tokens/total": 20971520,
+      "tokens/train_per_sec_per_gpu": 455.74,
+      "tokens/trainable": 13014990
+    },
+    {
+      "epoch": 0.019555555555555555,
+      "grad_norm": 1.5,
+      "learning_rate": 9.473264167865172e-06,
+      "loss": 0.8214111328125,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.27371,
+      "step": 11,
+      "tokens/total": 23068672,
+      "tokens/train_per_sec_per_gpu": 462.4,
+      "tokens/trainable": 14323401
+    },
+    {
+      "epoch": 0.021333333333333333,
+      "grad_norm": 0.859375,
+      "learning_rate": 9.316216432703918e-06,
+      "loss": 0.823974609375,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.27954,
+      "step": 12,
+      "tokens/total": 25165824,
+      "tokens/train_per_sec_per_gpu": 453.28,
+      "tokens/trainable": 15640932
+    },
+    {
+      "epoch": 0.02311111111111111,
+      "grad_norm": 0.8359375,
+      "learning_rate": 9.140576474687263e-06,
+      "loss": 0.8289794921875,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.29098,
+      "step": 13,
+      "tokens/total": 27262976,
+      "tokens/train_per_sec_per_gpu": 437.42,
+      "tokens/trainable": 16944692
+    },
+    {
+      "epoch": 0.024888888888888887,
+      "grad_norm": 0.7578125,
+      "learning_rate": 8.947199994035402e-06,
+      "loss": 0.788330078125,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.19972,
+      "step": 14,
+      "tokens/total": 29360128,
+      "tokens/train_per_sec_per_gpu": 465.93,
+      "tokens/trainable": 18261204
+    },
+    {
+      "epoch": 0.02666666666666667,
+      "grad_norm": 0.6875,
+      "learning_rate": 8.737029101523931e-06,
+      "loss": 0.8155517578125,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.26042,
+      "step": 15,
+      "tokens/total": 31457280,
+      "tokens/train_per_sec_per_gpu": 459.57,
+      "tokens/trainable": 19564392
+    },
+    {
+      "epoch": 0.028444444444444446,
+      "grad_norm": 0.87890625,
+      "learning_rate": 8.511087728614863e-06,
+      "loss": 0.79541015625,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.21535,
+      "step": 16,
+      "tokens/total": 33554432,
+      "tokens/train_per_sec_per_gpu": 455.81,
+      "tokens/trainable": 20882120
+    },
+    {
+      "epoch": 0.030222222222222223,
+      "grad_norm": 0.90234375,
+      "learning_rate": 8.270476638965463e-06,
+      "loss": 0.794677734375,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.21373,
+      "step": 17,
+      "tokens/total": 35651584,
+      "tokens/train_per_sec_per_gpu": 454.29,
+      "tokens/trainable": 22205430
+    },
+    {
+      "epoch": 0.032,
+      "grad_norm": 0.578125,
+      "learning_rate": 8.016368065618361e-06,
+      "loss": 0.7845458984375,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.19141,
+      "step": 18,
+      "tokens/total": 37748736,
+      "tokens/train_per_sec_per_gpu": 435.28,
+      "tokens/trainable": 23499492
+    },
+    {
+      "epoch": 0.033777777777777775,
+      "grad_norm": 0.796875,
+      "learning_rate": 7.75e-06,
+      "loss": 0.81103515625,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.25024,
+      "step": 19,
+      "tokens/total": 39845888,
+      "tokens/train_per_sec_per_gpu": 446.94,
+      "tokens/trainable": 24811318
+    },
+    {
+      "epoch": 0.035555555555555556,
+      "grad_norm": 0.56640625,
+      "learning_rate": 7.472670160550849e-06,
+      "loss": 0.792724609375,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.20941,
+      "step": 20,
+      "tokens/total": 41943040,
+      "tokens/train_per_sec_per_gpu": 428.76,
+      "tokens/trainable": 26123776
+    },
+    {
+      "epoch": 0.037333333333333336,
+      "grad_norm": 0.546875,
+      "learning_rate": 7.185729670371605e-06,
+      "loss": 0.8095703125,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.24694,
+      "step": 21,
+      "tokens/total": 44040192,
+      "tokens/train_per_sec_per_gpu": 457.99,
+      "tokens/trainable": 27455032
+    },
+    {
+      "epoch": 0.03911111111111111,
+      "grad_norm": 0.60546875,
+      "learning_rate": 6.890576474687264e-06,
+      "loss": 0.82666015625,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.28567,
+      "step": 22,
+      "tokens/total": 46137344,
+      "tokens/train_per_sec_per_gpu": 444.48,
+      "tokens/trainable": 28755554
+    },
+    {
+      "epoch": 0.04088888888888889,
+      "grad_norm": 0.50390625,
+      "learning_rate": 6.588648530198505e-06,
+      "loss": 0.792724609375,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.20941,
+      "step": 23,
+      "tokens/total": 48234496,
+      "tokens/train_per_sec_per_gpu": 436.92,
+      "tokens/trainable": 30052612
+    },
+    {
+      "epoch": 0.042666666666666665,
+      "grad_norm": 0.66796875,
+      "learning_rate": 6.281416799501188e-06,
+      "loss": 0.7847900390625,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.19195,
+      "step": 24,
+      "tokens/total": 50331648,
+      "tokens/train_per_sec_per_gpu": 451.17,
+      "tokens/trainable": 31330460
+    },
+    {
+      "epoch": 0.044444444444444446,
+      "grad_norm": 0.5625,
+      "learning_rate": 5.970378084704441e-06,
+      "loss": 0.769287109375,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.15823,
+      "step": 25,
+      "tokens/total": 52428800,
+      "tokens/train_per_sec_per_gpu": 429.83,
+      "tokens/trainable": 32640344
+    },
+    {
+      "epoch": 0.04622222222222222,
+      "grad_norm": 0.70703125,
+      "learning_rate": 5.657047735161256e-06,
+      "loss": 0.76416015625,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.14719,
+      "step": 26,
+      "tokens/total": 54525952,
+      "tokens/train_per_sec_per_gpu": 428.87,
+      "tokens/trainable": 33944456
+    },
+    {
+      "epoch": 0.048,
+      "grad_norm": 0.443359375,
+      "learning_rate": 5.342952264838748e-06,
+      "loss": 0.7957763671875,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.21616,
+      "step": 27,
+      "tokens/total": 56623104,
+      "tokens/train_per_sec_per_gpu": 410.57,
+      "tokens/trainable": 35241920
+    },
+    {
+      "epoch": 0.049777777777777775,
+      "grad_norm": 0.6953125,
+      "learning_rate": 5.02962191529556e-06,
+      "loss": 0.7666015625,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.15244,
+      "step": 28,
+      "tokens/total": 58720256,
+      "tokens/train_per_sec_per_gpu": 444.89,
+      "tokens/trainable": 36567256
+    },
+    {
+      "epoch": 0.051555555555555556,
+      "grad_norm": 0.484375,
+      "learning_rate": 4.718583200498814e-06,
+      "loss": 0.7696533203125,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.15902,
+      "step": 29,
+      "tokens/total": 60817408,
+      "tokens/train_per_sec_per_gpu": 457.68,
+      "tokens/trainable": 37880220
+    },
+    {
+      "epoch": 0.05333333333333334,
+      "grad_norm": 0.51953125,
+      "learning_rate": 4.4113514698014955e-06,
+      "loss": 0.7508544921875,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.11881,
+      "step": 30,
+      "tokens/total": 62898176,
+      "tokens/train_per_sec_per_gpu": 437.06,
+      "tokens/trainable": 39183952
+    },
+    {
+      "epoch": 0.05511111111111111,
+      "grad_norm": 0.41796875,
+      "learning_rate": 4.109423525312738e-06,
+      "loss": 0.759765625,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.13778,
+      "step": 31,
+      "tokens/total": 64995328,
+      "tokens/train_per_sec_per_gpu": 441.08,
+      "tokens/trainable": 40507844
+    },
+    {
+      "epoch": 0.05688888888888889,
+      "grad_norm": 0.69140625,
+      "learning_rate": 3.8142703296283954e-06,
+      "loss": 0.7796630859375,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.18074,
+      "step": 32,
+      "tokens/total": 67092480,
+      "tokens/train_per_sec_per_gpu": 405.73,
+      "tokens/trainable": 41801992
+    },
+    {
+      "epoch": 0.058666666666666666,
+      "grad_norm": 0.7265625,
+      "learning_rate": 3.527329839449152e-06,
+      "loss": 0.76513671875,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.14929,
+      "step": 33,
+      "tokens/total": 69189632,
+      "tokens/train_per_sec_per_gpu": 421.5,
+      "tokens/trainable": 43121624
+    },
+    {
+      "epoch": 0.060444444444444446,
+      "grad_norm": 0.51171875,
+      "learning_rate": 3.2500000000000015e-06,
+      "loss": 0.75341796875,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.12425,
+      "step": 34,
+      "tokens/total": 71286784,
+      "tokens/train_per_sec_per_gpu": 453.05,
+      "tokens/trainable": 44433728
+    },
+    {
+      "epoch": 0.06222222222222222,
+      "grad_norm": 0.91796875,
+      "learning_rate": 2.98363193438164e-06,
+      "loss": 0.7781982421875,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.17755,
+      "step": 35,
+      "tokens/total": 73383936,
+      "tokens/train_per_sec_per_gpu": 441.07,
+      "tokens/trainable": 45746124
+    },
+    {
+      "epoch": 0.064,
+      "grad_norm": 0.953125,
+      "learning_rate": 2.7295233610345384e-06,
+      "loss": 0.7635498046875,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.14588,
+      "step": 36,
+      "tokens/total": 75481088,
+      "tokens/train_per_sec_per_gpu": 447.24,
+      "tokens/trainable": 47051176
+    },
+    {
+      "epoch": 0.06577777777777778,
+      "grad_norm": 0.5703125,
+      "learning_rate": 2.4889122713851397e-06,
+      "loss": 0.7840576171875,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.19034,
+      "step": 37,
+      "tokens/total": 77578240,
+      "tokens/train_per_sec_per_gpu": 430.31,
+      "tokens/trainable": 48347584
+    },
+    {
+      "epoch": 0.06755555555555555,
+      "grad_norm": 0.478515625,
+      "learning_rate": 2.262970898476071e-06,
+      "loss": 0.74853515625,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.1139,
+      "step": 38,
+      "tokens/total": 79675392,
+      "tokens/train_per_sec_per_gpu": 417.6,
+      "tokens/trainable": 49668476
+    },
+    {
+      "epoch": 0.06933333333333333,
+      "grad_norm": 0.482421875,
+      "learning_rate": 2.0528000059646e-06,
+      "loss": 0.7698974609375,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.15954,
+      "step": 39,
+      "tokens/total": 81772544,
+      "tokens/train_per_sec_per_gpu": 439.1,
+      "tokens/trainable": 50995412
+    },
+    {
+      "epoch": 0.07111111111111111,
+      "grad_norm": 0.40234375,
+      "learning_rate": 1.8594235253127373e-06,
+      "loss": 0.7806396484375,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.18287,
+      "step": 40,
+      "tokens/total": 83869696,
+      "tokens/train_per_sec_per_gpu": 458.16,
+      "tokens/trainable": 52292136
+    },
+    {
+      "epoch": 0.07288888888888889,
+      "grad_norm": 0.5546875,
+      "learning_rate": 1.6837835672960834e-06,
+      "loss": 0.774169921875,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.16879,
+      "step": 41,
+      "tokens/total": 85966848,
+      "tokens/train_per_sec_per_gpu": 461.41,
+      "tokens/trainable": 53616892
+    },
+    {
+      "epoch": 0.07466666666666667,
+      "grad_norm": 0.453125,
+      "learning_rate": 1.526735832134829e-06,
+      "loss": 0.7769775390625,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.17489,
+      "step": 42,
+      "tokens/total": 88064000,
+      "tokens/train_per_sec_per_gpu": 450.43,
+      "tokens/trainable": 54907224
+    },
+    {
+      "epoch": 0.07644444444444444,
+      "grad_norm": 0.98828125,
+      "learning_rate": 1.389045440608296e-06,
+      "loss": 0.774169921875,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.16879,
+      "step": 43,
+      "tokens/total": 90161152,
+      "tokens/train_per_sec_per_gpu": 458.96,
+      "tokens/trainable": 56205776
+    },
+    {
+      "epoch": 0.07822222222222222,
+      "grad_norm": 0.5,
+      "learning_rate": 1.2713832064634127e-06,
+      "loss": 0.7801513671875,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.1818,
+      "step": 44,
+      "tokens/total": 92258304,
+      "tokens/train_per_sec_per_gpu": 428.42,
+      "tokens/trainable": 57484016
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5859375,
+      "learning_rate": 1.174322368277565e-06,
+      "loss": 0.775146484375,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.17091,
+      "step": 45,
+      "tokens/total": 94355456,
+      "tokens/train_per_sec_per_gpu": 474.34,
+      "tokens/trainable": 58774984
+    },
+    {
+      "epoch": 0.08177777777777778,
+      "grad_norm": 0.48828125,
+      "learning_rate": 1.0983357966978747e-06,
+      "loss": 0.766357421875,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.15191,
+      "step": 46,
+      "tokens/total": 96452608,
+      "tokens/train_per_sec_per_gpu": 431.68,
+      "tokens/trainable": 60057392
+    },
+    {
+      "epoch": 0.08355555555555555,
+      "grad_norm": 0.46484375,
+      "learning_rate": 1.0437936906629336e-06,
+      "loss": 0.789306640625,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.20187,
+      "step": 47,
+      "tokens/total": 98549760,
+      "tokens/train_per_sec_per_gpu": 450.54,
+      "tokens/trainable": 61352600
+    },
+    {
+      "epoch": 0.08533333333333333,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.0109617738307914e-06,
+      "loss": 0.7489013671875,
+      "memory/device_reserved (GiB)": 89.53,
+      "memory/max_active (GiB)": 79.77,
+      "memory/max_allocated (GiB)": 79.77,
+      "ppl": 2.11468,
+      "step": 48,
+      "tokens/total": 100646912,
+      "tokens/train_per_sec_per_gpu": 458.55,
+      "tokens/trainable": 62666372
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 48,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 48,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.748814946953331e+18,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}