grail_v2op_fcur_long_b64_sft_e3_ck1280

.gitattributes

@@ -79,3 +79,4 @@ grail_v3c_343_b64_sft_e3_ck640/tokenizer.json filter=lfs diff=lfs merge=lfs -tex
 grail_v1_trajq_rtx_b64_sft_e3_ck300/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 grail_v2op_fcur_b64_sft_e3_ck640/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 grail_v5_fcur_op48k_b64_sft_e3_ck800/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+grail_v2op_fcur_long_b64_sft_e3_ck1280/tokenizer.json filter=lfs diff=lfs merge=lfs -text

grail_v2op_fcur_long_b64_sft_e3_ck1280/all_results.json

@@ -0,0 +1,12 @@
+{
+    "epoch": 2.015754233950374,
+    "eval_loss": 0.017056895419955254,
+    "eval_runtime": 154.8424,
+    "eval_samples_per_second": 13.808,
+    "eval_steps_per_second": 3.455,
+    "total_flos": 559723689279488.0,
+    "train_loss": 0.06240204306668602,
+    "train_runtime": 14911.9121,
+    "train_samples_per_second": 5.494,
+    "train_steps_per_second": 0.086
+}

grail_v2op_fcur_long_b64_sft_e3_ck1280/chat_template.jinja

@@ -0,0 +1,120 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' }}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- for message in messages %}
+    {%- if message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content in message.content %}
+                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}
+                    <|vision_start|><|image_pad|><|vision_end|>
+                {%- elif content.type == 'video' or 'video' in content %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}
+                    <|vision_start|><|video_pad|><|vision_end|>
+                {%- elif 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content_item in message.content %}
+                {%- if 'text' in content_item %}
+                    {{- content_item.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and message.content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content in message.content %}
+                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}
+                    <|vision_start|><|image_pad|><|vision_end|>
+                {%- elif content.type == 'video' or 'video' in content %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}
+                    <|vision_start|><|video_pad|><|vision_end|>
+                {%- elif 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}

grail_v2op_fcur_long_b64_sft_e3_ck1280/config.json

@@ -0,0 +1,70 @@
+{
+  "architectures": [
+    "Qwen3VLForConditionalGeneration"
+  ],
+  "bos_token_id": null,
+  "dtype": "bfloat16",
+  "eos_token_id": 151645,
+  "hidden_size": 4096,
+  "image_token_id": 151655,
+  "model_type": "qwen3_vl",
+  "pad_token_id": 151643,
+  "text_config": {
+    "attention_bias": false,
+    "attention_dropout": 0.0,
+    "bos_token_id": 151643,
+    "dtype": "bfloat16",
+    "eos_token_id": 151645,
+    "head_dim": 128,
+    "hidden_act": "silu",
+    "hidden_size": 4096,
+    "initializer_range": 0.02,
+    "intermediate_size": 12288,
+    "max_position_embeddings": 262144,
+    "model_type": "qwen3_vl_text",
+    "num_attention_heads": 32,
+    "num_hidden_layers": 36,
+    "num_key_value_heads": 8,
+    "pad_token_id": null,
+    "rms_norm_eps": 1e-06,
+    "rope_parameters": {
+      "mrope_interleaved": true,
+      "mrope_section": [
+        24,
+        20,
+        20
+      ],
+      "rope_theta": 5000000,
+      "rope_type": "default"
+    },
+    "use_cache": false,
+    "vocab_size": 151936
+  },
+  "tie_word_embeddings": false,
+  "transformers_version": "5.8.0",
+  "use_cache": false,
+  "video_token_id": 151656,
+  "vision_config": {
+    "deepstack_visual_indexes": [
+      8,
+      16,
+      24
+    ],
+    "depth": 27,
+    "dtype": "bfloat16",
+    "hidden_act": "gelu_pytorch_tanh",
+    "hidden_size": 1152,
+    "in_channels": 3,
+    "initializer_range": 0.02,
+    "intermediate_size": 4304,
+    "model_type": "qwen3_vl_vision",
+    "num_heads": 16,
+    "num_position_embeddings": 2304,
+    "out_hidden_size": 4096,
+    "patch_size": 16,
+    "spatial_merge_size": 2,
+    "temporal_patch_size": 2
+  },
+  "vision_end_token_id": 151653,
+  "vision_start_token_id": 151652
+}

grail_v2op_fcur_long_b64_sft_e3_ck1280/eval_results.json

@@ -0,0 +1,7 @@
+{
+    "epoch": 2.015754233950374,
+    "eval_loss": 0.017056895419955254,
+    "eval_runtime": 154.8424,
+    "eval_samples_per_second": 13.808,
+    "eval_steps_per_second": 3.455
+}

grail_v2op_fcur_long_b64_sft_e3_ck1280/generation_config.json

@@ -0,0 +1,14 @@
+{
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "repetition_penalty": 1.0,
+  "temperature": 0.7,
+  "top_k": 20,
+  "top_p": 0.8,
+  "transformers_version": "5.8.0"
+}

grail_v2op_fcur_long_b64_sft_e3_ck1280/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:072fc7dc0e28232128dc1f739b0921613ff2c0180234e4b6fb36b3be55e8edb6
+size 17534340584

grail_v2op_fcur_long_b64_sft_e3_ck1280/processor_config.json

@@ -0,0 +1,60 @@
+{
+  "image_processor": {
+    "do_convert_rgb": true,
+    "do_normalize": true,
+    "do_rescale": true,
+    "do_resize": true,
+    "image_mean": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "image_processor_type": "Qwen2VLImageProcessor",
+    "image_std": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "merge_size": 2,
+    "patch_size": 16,
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098,
+    "size": {
+      "longest_edge": 16777216,
+      "shortest_edge": 65536
+    },
+    "temporal_patch_size": 2
+  },
+  "processor_class": "Qwen3VLProcessor",
+  "video_processor": {
+    "do_convert_rgb": true,
+    "do_normalize": true,
+    "do_rescale": true,
+    "do_resize": true,
+    "do_sample_frames": true,
+    "fps": 2,
+    "image_mean": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "image_std": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "max_frames": 768,
+    "merge_size": 2,
+    "min_frames": 4,
+    "patch_size": 16,
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098,
+    "return_metadata": false,
+    "size": {
+      "longest_edge": 25165824,
+      "shortest_edge": 4096
+    },
+    "temporal_patch_size": 2,
+    "video_processor_type": "Qwen3VLVideoProcessor"
+  }
+}

grail_v2op_fcur_long_b64_sft_e3_ck1280/sft.yaml

@@ -0,0 +1,58 @@
+# LLaMA-Factory full-parameter SFT of Qwen3-VL-8B-Instruct on one HER-final dataset.
+# Four L40S (46 GB), one run at a time: DeepSpeed ZeRO-3 with optimizer offload to host
+# RAM (~130 GB per run, so two runs do not fit in 372 GB); the vision tower and projector
+# stay frozen (LLaMA-Factory's default for VLMs), the LLM trains.
+# Filled per run by launch_sft.sh: grail_v2op_fcur_sft_e3, exps/sft/grail_v2op_fcur_sft_e3_k25515, /scratch/kwc5354/viewagent/viewagent_private/exps/sft_runs/grail_v2op_fcur_long_b64_sft_e3.
+### model
+model_name_or_path: /scratch/kwc5354/viewagent/viewagent_private/exps/models/graph_nemi_sft_e2
+template: qwen3_vl_nothink
+trust_remote_code: true
+image_max_pixels: 65536          # frames are 256x256 already; never upscale
+image_min_pixels: 4096
+
+### method
+stage: sft
+do_train: true
+finetuning_type: full
+freeze_vision_tower: true
+freeze_multi_modal_projector: true
+freeze_language_model: false
+deepspeed: /scratch/kwc5354/viewagent/viewagent_private/LLaMA-Factory/examples/deepspeed/ds_z3_config.json
+
+### dataset
+dataset: grail_v2op_fcur_sft_e3
+dataset_dir: exps/sft/grail_v2op_fcur_sft_e3_k25515
+media_dir: /scratch/kwc5354/viewagent/viewagent_private
+cutoff_len: 8192
+overwrite_cache: true
+preprocessing_num_workers: 16
+dataloader_num_workers: 4
+
+### output
+output_dir: /scratch/kwc5354/viewagent/viewagent_private/exps/sft_runs/grail_v2op_fcur_long_b64_sft_e3
+logging_steps: 10
+save_strategy: steps
+save_steps: 640
+save_total_limit: 100
+save_only_model: true
+plot_loss: true
+overwrite_output_dir: true
+report_to: wandb
+
+### train
+per_device_train_batch_size: 1
+gradient_accumulation_steps: 16
+max_steps: 1280
+learning_rate: 1.0e-5
+lr_scheduler_type: cosine
+warmup_ratio: 0.03
+weight_decay: 0.01
+bf16: true
+gradient_checkpointing: true
+ddp_timeout: 180000000
+seed: 0
+val_size: 0.05
+eval_strategy: steps
+eval_steps: 640
+per_device_eval_batch_size: 1
+run_name: grail_v2op_fcur_long_b64_sft_e3

grail_v2op_fcur_long_b64_sft_e3_ck1280/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:be75606093db2094d7cd20f3c2f385c212750648bd6ea4fb2bf507a6a4c55506
+size 11422650

grail_v2op_fcur_long_b64_sft_e3_ck1280/tokenizer_config.json

@@ -0,0 +1,32 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": true,
+  "local_files_only": false,
+  "model_max_length": 262144,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "right",
+  "processor_class": "Qwen3VLProcessor",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

grail_v2op_fcur_long_b64_sft_e3_ck1280/train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 2.015754233950374,
+    "total_flos": 559723689279488.0,
+    "train_loss": 0.06240204306668602,
+    "train_runtime": 14911.9121,
+    "train_samples_per_second": 5.494,
+    "train_steps_per_second": 0.086
+}

grail_v2op_fcur_long_b64_sft_e3_ck1280/trainer_state.json

@@ -0,0 +1,955 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.015754233950374,
+  "eval_steps": 640,
+  "global_step": 1280,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.015754233950374164,
+      "grad_norm": 12.033636397790756,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 1.32488956451416,
+      "step": 10
+    },
+    {
+      "epoch": 0.03150846790074833,
+      "grad_norm": 2.3636888398396225,
+      "learning_rate": 4.871794871794872e-06,
+      "loss": 0.5897706031799317,
+      "step": 20
+    },
+    {
+      "epoch": 0.04726270185112249,
+      "grad_norm": 1.3272642735418618,
+      "learning_rate": 7.435897435897437e-06,
+      "loss": 0.24595553874969484,
+      "step": 30
+    },
+    {
+      "epoch": 0.06301693580149666,
+      "grad_norm": 1.0296978666917989,
+      "learning_rate": 1e-05,
+      "loss": 0.20369548797607423,
+      "step": 40
+    },
+    {
+      "epoch": 0.07877116975187082,
+      "grad_norm": 0.8488286569637543,
+      "learning_rate": 9.998397961319146e-06,
+      "loss": 0.17844419479370116,
+      "step": 50
+    },
+    {
+      "epoch": 0.09452540370224498,
+      "grad_norm": 0.6840639499263153,
+      "learning_rate": 9.993592871887754e-06,
+      "loss": 0.16870492696762085,
+      "step": 60
+    },
+    {
+      "epoch": 0.11027963765261914,
+      "grad_norm": 0.7371439375633763,
+      "learning_rate": 9.98558781088148e-06,
+      "loss": 0.15912662744522094,
+      "step": 70
+    },
+    {
+      "epoch": 0.1260338716029933,
+      "grad_norm": 0.570568402073853,
+      "learning_rate": 9.974387908067273e-06,
+      "loss": 0.14628233909606933,
+      "step": 80
+    },
+    {
+      "epoch": 0.14178810555336746,
+      "grad_norm": 0.9460272761768993,
+      "learning_rate": 9.960000340516145e-06,
+      "loss": 0.14640895128250123,
+      "step": 90
+    },
+    {
+      "epoch": 0.15754233950374164,
+      "grad_norm": 0.5452707827296494,
+      "learning_rate": 9.942434328003991e-06,
+      "loss": 0.1338128924369812,
+      "step": 100
+    },
+    {
+      "epoch": 0.17329657345411578,
+      "grad_norm": 0.6229754551718935,
+      "learning_rate": 9.92170112710342e-06,
+      "loss": 0.130888831615448,
+      "step": 110
+    },
+    {
+      "epoch": 0.18905080740448996,
+      "grad_norm": 0.706790804008561,
+      "learning_rate": 9.897814023970357e-06,
+      "loss": 0.1185534119606018,
+      "step": 120
+    },
+    {
+      "epoch": 0.20480504135486413,
+      "grad_norm": 0.6101568485561453,
+      "learning_rate": 9.870788325830079e-06,
+      "loss": 0.11678495407104492,
+      "step": 130
+    },
+    {
+      "epoch": 0.22055927530523828,
+      "grad_norm": 0.6948852991899241,
+      "learning_rate": 9.840641351168107e-06,
+      "loss": 0.11397801637649536,
+      "step": 140
+    },
+    {
+      "epoch": 0.23631350925561245,
+      "grad_norm": 0.6881519123948223,
+      "learning_rate": 9.807392418632246e-06,
+      "loss": 0.10202901363372803,
+      "step": 150
+    },
+    {
+      "epoch": 0.2520677432059866,
+      "grad_norm": 0.836127572122202,
+      "learning_rate": 9.771062834652907e-06,
+      "loss": 0.10777673721313477,
+      "step": 160
+    },
+    {
+      "epoch": 0.26782197715636075,
+      "grad_norm": 0.7367935304919844,
+      "learning_rate": 9.731675879789606e-06,
+      "loss": 0.10600593090057372,
+      "step": 170
+    },
+    {
+      "epoch": 0.2835762111067349,
+      "grad_norm": 0.6292484126654329,
+      "learning_rate": 9.689256793812427e-06,
+      "loss": 0.09942523241043091,
+      "step": 180
+    },
+    {
+      "epoch": 0.2993304450571091,
+      "grad_norm": 0.7381190657702302,
+      "learning_rate": 9.64383275952799e-06,
+      "loss": 0.09324796795845032,
+      "step": 190
+    },
+    {
+      "epoch": 0.31508467900748327,
+      "grad_norm": 0.7376939143417104,
+      "learning_rate": 9.595432885360276e-06,
+      "loss": 0.09175238609313965,
+      "step": 200
+    },
+    {
+      "epoch": 0.33083891295785745,
+      "grad_norm": 0.7329600068136058,
+      "learning_rate": 9.544088186697515e-06,
+      "loss": 0.09360949993133545,
+      "step": 210
+    },
+    {
+      "epoch": 0.34659314690823156,
+      "grad_norm": 0.8534199179906135,
+      "learning_rate": 9.489831566017031e-06,
+      "loss": 0.08230534791946412,
+      "step": 220
+    },
+    {
+      "epoch": 0.36234738085860574,
+      "grad_norm": 0.7354595303347835,
+      "learning_rate": 9.432697791800835e-06,
+      "loss": 0.08615907430648803,
+      "step": 230
+    },
+    {
+      "epoch": 0.3781016148089799,
+      "grad_norm": 0.6476344719634456,
+      "learning_rate": 9.372723476255434e-06,
+      "loss": 0.0864217758178711,
+      "step": 240
+    },
+    {
+      "epoch": 0.3938558487593541,
+      "grad_norm": 0.8228291986049908,
+      "learning_rate": 9.309947051850177e-06,
+      "loss": 0.08714066743850708,
+      "step": 250
+    },
+    {
+      "epoch": 0.40961008270972826,
+      "grad_norm": 0.8025978339612624,
+      "learning_rate": 9.244408746689119e-06,
+      "loss": 0.07662501335144042,
+      "step": 260
+    },
+    {
+      "epoch": 0.4253643166601024,
+      "grad_norm": 0.7943383289883176,
+      "learning_rate": 9.176150558732238e-06,
+      "loss": 0.07686767578125,
+      "step": 270
+    },
+    {
+      "epoch": 0.44111855061047656,
+      "grad_norm": 0.6645339820041942,
+      "learning_rate": 9.105216228882491e-06,
+      "loss": 0.07104989290237426,
+      "step": 280
+    },
+    {
+      "epoch": 0.45687278456085073,
+      "grad_norm": 0.7723811119622486,
+      "learning_rate": 9.031651212955968e-06,
+      "loss": 0.07707502841949462,
+      "step": 290
+    },
+    {
+      "epoch": 0.4726270185112249,
+      "grad_norm": 0.7725119190240489,
+      "learning_rate": 8.955502652553095e-06,
+      "loss": 0.07058619856834411,
+      "step": 300
+    },
+    {
+      "epoch": 0.4883812524615991,
+      "grad_norm": 0.6934389174242291,
+      "learning_rate": 8.876819344849576e-06,
+      "loss": 0.0702974796295166,
+      "step": 310
+    },
+    {
+      "epoch": 0.5041354864119733,
+      "grad_norm": 0.6025680679646537,
+      "learning_rate": 8.795651711326401e-06,
+      "loss": 0.07145766019821168,
+      "step": 320
+    },
+    {
+      "epoch": 0.5198897203623474,
+      "grad_norm": 0.7536040855408148,
+      "learning_rate": 8.712051765458989e-06,
+      "loss": 0.06356290578842164,
+      "step": 330
+    },
+    {
+      "epoch": 0.5356439543127215,
+      "grad_norm": 0.7171123350866928,
+      "learning_rate": 8.626073079386135e-06,
+      "loss": 0.06653350591659546,
+      "step": 340
+    },
+    {
+      "epoch": 0.5513981882630957,
+      "grad_norm": 0.6490604421349927,
+      "learning_rate": 8.537770749580168e-06,
+      "loss": 0.061210238933563234,
+      "step": 350
+    },
+    {
+      "epoch": 0.5671524222134698,
+      "grad_norm": 0.6642780252951906,
+      "learning_rate": 8.44720136154027e-06,
+      "loss": 0.061034214496612546,
+      "step": 360
+    },
+    {
+      "epoch": 0.5829066561638441,
+      "grad_norm": 0.7419485121806322,
+      "learning_rate": 8.354422953531617e-06,
+      "loss": 0.05697993636131286,
+      "step": 370
+    },
+    {
+      "epoch": 0.5986608901142182,
+      "grad_norm": 0.6875884676717379,
+      "learning_rate": 8.259494979393563e-06,
+      "loss": 0.06042020320892334,
+      "step": 380
+    },
+    {
+      "epoch": 0.6144151240645923,
+      "grad_norm": 0.6332851748466177,
+      "learning_rate": 8.162478270440693e-06,
+      "loss": 0.059015631675720215,
+      "step": 390
+    },
+    {
+      "epoch": 0.6301693580149665,
+      "grad_norm": 0.7809392031197802,
+      "learning_rate": 8.063434996481178e-06,
+      "loss": 0.05616657137870788,
+      "step": 400
+    },
+    {
+      "epoch": 0.6459235919653407,
+      "grad_norm": 0.6541074776434227,
+      "learning_rate": 7.962428625977403e-06,
+      "loss": 0.054358971118927,
+      "step": 410
+    },
+    {
+      "epoch": 0.6616778259157149,
+      "grad_norm": 0.5611882833238244,
+      "learning_rate": 7.859523885374395e-06,
+      "loss": 0.05288315415382385,
+      "step": 420
+    },
+    {
+      "epoch": 0.677432059866089,
+      "grad_norm": 0.5771127273221914,
+      "learning_rate": 7.754786717622107e-06,
+      "loss": 0.05506861209869385,
+      "step": 430
+    },
+    {
+      "epoch": 0.6931862938164631,
+      "grad_norm": 0.7859565588587687,
+      "learning_rate": 7.648284239918163e-06,
+      "loss": 0.051020103693008426,
+      "step": 440
+    },
+    {
+      "epoch": 0.7089405277668374,
+      "grad_norm": 0.6692069264779664,
+      "learning_rate": 7.54008470069812e-06,
+      "loss": 0.05129450559616089,
+      "step": 450
+    },
+    {
+      "epoch": 0.7246947617172115,
+      "grad_norm": 0.6678401444978758,
+      "learning_rate": 7.43025743590081e-06,
+      "loss": 0.051376509666442874,
+      "step": 460
+    },
+    {
+      "epoch": 0.7404489956675857,
+      "grad_norm": 0.7710762471599943,
+      "learning_rate": 7.318872824536801e-06,
+      "loss": 0.05178595781326294,
+      "step": 470
+    },
+    {
+      "epoch": 0.7562032296179598,
+      "grad_norm": 0.7497920627506444,
+      "learning_rate": 7.206002243588436e-06,
+      "loss": 0.04661799669265747,
+      "step": 480
+    },
+    {
+      "epoch": 0.771957463568334,
+      "grad_norm": 0.7591553206316668,
+      "learning_rate": 7.0917180222703566e-06,
+      "loss": 0.0466660887002945,
+      "step": 490
+    },
+    {
+      "epoch": 0.7877116975187082,
+      "grad_norm": 0.8191928545570479,
+      "learning_rate": 6.97609339567983e-06,
+      "loss": 0.04904979467391968,
+      "step": 500
+    },
+    {
+      "epoch": 0.8034659314690823,
+      "grad_norm": 0.6885971709493436,
+      "learning_rate": 6.85920245786656e-06,
+      "loss": 0.047192856669425964,
+      "step": 510
+    },
+    {
+      "epoch": 0.8192201654194565,
+      "grad_norm": 0.5588702186733117,
+      "learning_rate": 6.741120114352071e-06,
+      "loss": 0.0419720321893692,
+      "step": 520
+    },
+    {
+      "epoch": 0.8349743993698306,
+      "grad_norm": 0.637466996910207,
+      "learning_rate": 6.6219220341291e-06,
+      "loss": 0.04147773683071136,
+      "step": 530
+    },
+    {
+      "epoch": 0.8507286333202048,
+      "grad_norm": 0.5012287857773934,
+      "learning_rate": 6.501684601171727e-06,
+      "loss": 0.041779667139053345,
+      "step": 540
+    },
+    {
+      "epoch": 0.866482867270579,
+      "grad_norm": 0.7249508586780151,
+      "learning_rate": 6.380484865487346e-06,
+      "loss": 0.042048674821853635,
+      "step": 550
+    },
+    {
+      "epoch": 0.8822371012209531,
+      "grad_norm": 0.5712846931860893,
+      "learning_rate": 6.258400493741826e-06,
+      "loss": 0.041913390159606934,
+      "step": 560
+    },
+    {
+      "epoch": 0.8979913351713273,
+      "grad_norm": 0.7382721943013278,
+      "learning_rate": 6.1355097194895115e-06,
+      "loss": 0.03828791677951813,
+      "step": 570
+    },
+    {
+      "epoch": 0.9137455691217015,
+      "grad_norm": 0.6373018091362586,
+      "learning_rate": 6.011891293039954e-06,
+      "loss": 0.04503217339515686,
+      "step": 580
+    },
+    {
+      "epoch": 0.9294998030720756,
+      "grad_norm": 0.6000648170497052,
+      "learning_rate": 5.887624430993489e-06,
+      "loss": 0.04264146089553833,
+      "step": 590
+    },
+    {
+      "epoch": 0.9452540370224498,
+      "grad_norm": 0.6075439292380207,
+      "learning_rate": 5.7627887654780145e-06,
+      "loss": 0.03540692329406738,
+      "step": 600
+    },
+    {
+      "epoch": 0.9610082709728239,
+      "grad_norm": 0.6759582485560787,
+      "learning_rate": 5.637464293119493e-06,
+      "loss": 0.03805915117263794,
+      "step": 610
+    },
+    {
+      "epoch": 0.9767625049231982,
+      "grad_norm": 0.7528146158798082,
+      "learning_rate": 5.511731323778876e-06,
+      "loss": 0.040453338623046876,
+      "step": 620
+    },
+    {
+      "epoch": 0.9925167388735723,
+      "grad_norm": 0.6675716036977815,
+      "learning_rate": 5.385670429088297e-06,
+      "loss": 0.03337875008583069,
+      "step": 630
+    },
+    {
+      "epoch": 1.007877116975187,
+      "grad_norm": 0.5237268636739532,
+      "learning_rate": 5.259362390819537e-06,
+      "loss": 0.033685770630836484,
+      "step": 640
+    },
+    {
+      "epoch": 1.007877116975187,
+      "eval_loss": 0.03434622287750244,
+      "eval_runtime": 155.4611,
+      "eval_samples_per_second": 13.753,
+      "eval_steps_per_second": 3.441,
+      "step": 640
+    },
+    {
+      "epoch": 1.0236313509255612,
+      "grad_norm": 0.5303257061598937,
+      "learning_rate": 5.132888149117793e-06,
+      "loss": 0.026704996824264526,
+      "step": 650
+    },
+    {
+      "epoch": 1.0393855848759355,
+      "grad_norm": 0.503268568378794,
+      "learning_rate": 5.006328750634005e-06,
+      "loss": 0.02573159635066986,
+      "step": 660
+    },
+    {
+      "epoch": 1.0551398188263095,
+      "grad_norm": 0.8145181355527468,
+      "learning_rate": 4.879765296588888e-06,
+      "loss": 0.024003131687641142,
+      "step": 670
+    },
+    {
+      "epoch": 1.0708940527766837,
+      "grad_norm": 0.6673090226600138,
+      "learning_rate": 4.753278890802034e-06,
+      "loss": 0.02469957172870636,
+      "step": 680
+    },
+    {
+      "epoch": 1.086648286727058,
+      "grad_norm": 0.6266239557861248,
+      "learning_rate": 4.626950587719303e-06,
+      "loss": 0.024609375,
+      "step": 690
+    },
+    {
+      "epoch": 1.1024025206774322,
+      "grad_norm": 0.6743296316204072,
+      "learning_rate": 4.50086134047191e-06,
+      "loss": 0.024108047783374786,
+      "step": 700
+    },
+    {
+      "epoch": 1.1181567546278062,
+      "grad_norm": 0.5564729040564508,
+      "learning_rate": 4.375091949000387e-06,
+      "loss": 0.025480023026466368,
+      "step": 710
+    },
+    {
+      "epoch": 1.1339109885781804,
+      "grad_norm": 0.6230163537203788,
+      "learning_rate": 4.249723008276737e-06,
+      "loss": 0.02517353296279907,
+      "step": 720
+    },
+    {
+      "epoch": 1.1496652225285546,
+      "grad_norm": 0.6253588875828883,
+      "learning_rate": 4.1248348566579245e-06,
+      "loss": 0.022786788642406464,
+      "step": 730
+    },
+    {
+      "epoch": 1.1654194564789286,
+      "grad_norm": 0.7009916313132863,
+      "learning_rate": 4.00050752440382e-06,
+      "loss": 0.02621592879295349,
+      "step": 740
+    },
+    {
+      "epoch": 1.1811736904293029,
+      "grad_norm": 0.622191724249774,
+      "learning_rate": 3.8768206823925694e-06,
+      "loss": 0.022946587204933165,
+      "step": 750
+    },
+    {
+      "epoch": 1.196927924379677,
+      "grad_norm": 0.7111784105851262,
+      "learning_rate": 3.7538535910662533e-06,
+      "loss": 0.024950674176216124,
+      "step": 760
+    },
+    {
+      "epoch": 1.212682158330051,
+      "grad_norm": 0.652989151560588,
+      "learning_rate": 3.6316850496395863e-06,
+      "loss": 0.02058393955230713,
+      "step": 770
+    },
+    {
+      "epoch": 1.2284363922804253,
+      "grad_norm": 0.5927704733012786,
+      "learning_rate": 3.510393345604146e-06,
+      "loss": 0.022996044158935545,
+      "step": 780
+    },
+    {
+      "epoch": 1.2441906262307996,
+      "grad_norm": 0.5277552151734335,
+      "learning_rate": 3.3900562045605494e-06,
+      "loss": 0.022907957434654236,
+      "step": 790
+    },
+    {
+      "epoch": 1.2599448601811738,
+      "grad_norm": 0.525360487524952,
+      "learning_rate": 3.2707507404106692e-06,
+      "loss": 0.021823813021183015,
+      "step": 800
+    },
+    {
+      "epoch": 1.2756990941315478,
+      "grad_norm": 0.5058375003976344,
+      "learning_rate": 3.15255340594187e-06,
+      "loss": 0.0220929354429245,
+      "step": 810
+    },
+    {
+      "epoch": 1.291453328081922,
+      "grad_norm": 0.4536064951636675,
+      "learning_rate": 3.035539943834871e-06,
+      "loss": 0.020201978087425233,
+      "step": 820
+    },
+    {
+      "epoch": 1.3072075620322963,
+      "grad_norm": 0.6986012905057108,
+      "learning_rate": 2.9197853381266585e-06,
+      "loss": 0.021213629841804506,
+      "step": 830
+    },
+    {
+      "epoch": 1.3229617959826703,
+      "grad_norm": 0.5434556859752665,
+      "learning_rate": 2.8053637661595657e-06,
+      "loss": 0.017476361989974976,
+      "step": 840
+    },
+    {
+      "epoch": 1.3387160299330445,
+      "grad_norm": 0.5674154564310521,
+      "learning_rate": 2.6923485510472823e-06,
+      "loss": 0.016973769664764403,
+      "step": 850
+    },
+    {
+      "epoch": 1.3544702638834187,
+      "grad_norm": 0.4979202407061893,
+      "learning_rate": 2.5808121146882613e-06,
+      "loss": 0.018821868300437927,
+      "step": 860
+    },
+    {
+      "epoch": 1.3702244978337927,
+      "grad_norm": 0.6035306910510858,
+      "learning_rate": 2.4708259313566503e-06,
+      "loss": 0.018303532898426057,
+      "step": 870
+    },
+    {
+      "epoch": 1.385978731784167,
+      "grad_norm": 0.5469846866417566,
+      "learning_rate": 2.3624604819004697e-06,
+      "loss": 0.01886426508426666,
+      "step": 880
+    },
+    {
+      "epoch": 1.4017329657345412,
+      "grad_norm": 0.5486841235624124,
+      "learning_rate": 2.2557852085764053e-06,
+      "loss": 0.018669766187667847,
+      "step": 890
+    },
+    {
+      "epoch": 1.4174871996849152,
+      "grad_norm": 0.5114116784527947,
+      "learning_rate": 2.1508684705501117e-06,
+      "loss": 0.017075559496879576,
+      "step": 900
+    },
+    {
+      "epoch": 1.4332414336352894,
+      "grad_norm": 0.5495800530900189,
+      "learning_rate": 2.0477775000906285e-06,
+      "loss": 0.01788746118545532,
+      "step": 910
+    },
+    {
+      "epoch": 1.4489956675856637,
+      "grad_norm": 0.6195279341237376,
+      "learning_rate": 1.9465783594868835e-06,
+      "loss": 0.01739121526479721,
+      "step": 920
+    },
+    {
+      "epoch": 1.4647499015360377,
+      "grad_norm": 0.5608516955420921,
+      "learning_rate": 1.8473358987139622e-06,
+      "loss": 0.016242732107639313,
+      "step": 930
+    },
+    {
+      "epoch": 1.480504135486412,
+      "grad_norm": 0.5906289820682826,
+      "learning_rate": 1.7501137138762436e-06,
+      "loss": 0.016568711400032042,
+      "step": 940
+    },
+    {
+      "epoch": 1.4962583694367861,
+      "grad_norm": 0.7700538223522422,
+      "learning_rate": 1.6549741064540259e-06,
+      "loss": 0.018120746314525604,
+      "step": 950
+    },
+    {
+      "epoch": 1.5120126033871601,
+      "grad_norm": 0.5754251141062999,
+      "learning_rate": 1.5619780433797764e-06,
+      "loss": 0.017181001603603363,
+      "step": 960
+    },
+    {
+      "epoch": 1.5277668373375346,
+      "grad_norm": 0.6824348190718681,
+      "learning_rate": 1.47118511796958e-06,
+      "loss": 0.01730160415172577,
+      "step": 970
+    },
+    {
+      "epoch": 1.5435210712879086,
+      "grad_norm": 0.4343410750576567,
+      "learning_rate": 1.3826535117348216e-06,
+      "loss": 0.014836643636226655,
+      "step": 980
+    },
+    {
+      "epoch": 1.5592753052382828,
+      "grad_norm": 0.4335181135796832,
+      "learning_rate": 1.2964399570985654e-06,
+      "loss": 0.01643817275762558,
+      "step": 990
+    },
+    {
+      "epoch": 1.575029539188657,
+      "grad_norm": 0.5681879610025925,
+      "learning_rate": 1.2125997010405476e-06,
+      "loss": 0.01467774361371994,
+      "step": 1000
+    },
+    {
+      "epoch": 1.590783773139031,
+      "grad_norm": 0.49407834143258583,
+      "learning_rate": 1.1311864696940556e-06,
+      "loss": 0.014845505356788635,
+      "step": 1010
+    },
+    {
+      "epoch": 1.6065380070894053,
+      "grad_norm": 0.5516487616233712,
+      "learning_rate": 1.0522524339173907e-06,
+      "loss": 0.01466805934906006,
+      "step": 1020
+    },
+    {
+      "epoch": 1.6222922410397795,
+      "grad_norm": 0.5195562906452837,
+      "learning_rate": 9.758481758619725e-07,
+      "loss": 0.013550400733947754,
+      "step": 1030
+    },
+    {
+      "epoch": 1.6380464749901535,
+      "grad_norm": 0.6161202745721324,
+      "learning_rate": 9.020226565585156e-07,
+      "loss": 0.015657591819763183,
+      "step": 1040
+    },
+    {
+      "epoch": 1.6538007089405278,
+      "grad_norm": 0.4668604000317069,
+      "learning_rate": 8.308231845420417e-07,
+      "loss": 0.015800310671329497,
+      "step": 1050
+    },
+    {
+      "epoch": 1.669554942890902,
+      "grad_norm": 0.5865144499582374,
+      "learning_rate": 7.622953855358456e-07,
+      "loss": 0.014689505100250244,
+      "step": 1060
+    },
+    {
+      "epoch": 1.685309176841276,
+      "grad_norm": 0.4693983416381837,
+      "learning_rate": 6.964831732138111e-07,
+      "loss": 0.016504742205142975,
+      "step": 1070
+    },
+    {
+      "epoch": 1.7010634107916502,
+      "grad_norm": 0.47497085432857616,
+      "learning_rate": 6.334287210598672e-07,
+      "loss": 0.012360145151615144,
+      "step": 1080
+    },
+    {
+      "epoch": 1.7168176447420245,
+      "grad_norm": 0.5199780352972881,
+      "learning_rate": 5.731724353425511e-07,
+      "loss": 0.013789060711860656,
+      "step": 1090
+    },
+    {
+      "epoch": 1.7325718786923985,
+      "grad_norm": 0.6134945525618148,
+      "learning_rate": 5.157529292220592e-07,
+      "loss": 0.012603117525577546,
+      "step": 1100
+    },
+    {
+      "epoch": 1.7483261126427727,
+      "grad_norm": 0.528102667531573,
+      "learning_rate": 4.6120699800632607e-07,
+      "loss": 0.013232062757015228,
+      "step": 1110
+    },
+    {
+      "epoch": 1.764080346593147,
+      "grad_norm": 0.3515343998963338,
+      "learning_rate": 4.0956959557202744e-07,
+      "loss": 0.011887529492378235,
+      "step": 1120
+    },
+    {
+      "epoch": 1.779834580543521,
+      "grad_norm": 0.48315461547594984,
+      "learning_rate": 3.60873811965598e-07,
+      "loss": 0.011624448746442795,
+      "step": 1130
+    },
+    {
+      "epoch": 1.7955888144938952,
+      "grad_norm": 0.5457832349585932,
+      "learning_rate": 3.151508521986063e-07,
+      "loss": 0.01403396725654602,
+      "step": 1140
+    },
+    {
+      "epoch": 1.8113430484442694,
+      "grad_norm": 0.6789199538960683,
+      "learning_rate": 2.724300162511151e-07,
+      "loss": 0.014070636034011841,
+      "step": 1150
+    },
+    {
+      "epoch": 1.8270972823946434,
+      "grad_norm": 0.5013062049452625,
+      "learning_rate": 2.3273868029578929e-07,
+      "loss": 0.012945161759853363,
+      "step": 1160
+    },
+    {
+      "epoch": 1.8428515163450179,
+      "grad_norm": 0.37248088942612595,
+      "learning_rate": 1.9610227915482894e-07,
+      "loss": 0.01211639642715454,
+      "step": 1170
+    },
+    {
+      "epoch": 1.8586057502953919,
+      "grad_norm": 0.5396182899347344,
+      "learning_rate": 1.6254429000093385e-07,
+      "loss": 0.013820630311965943,
+      "step": 1180
+    },
+    {
+      "epoch": 1.8743599842457659,
+      "grad_norm": 0.4846192698419647,
+      "learning_rate": 1.3208621731277682e-07,
+      "loss": 0.012912285327911378,
+      "step": 1190
+    },
+    {
+      "epoch": 1.8901142181961403,
+      "grad_norm": 0.41960154149037615,
+      "learning_rate": 1.0474757909459233e-07,
+      "loss": 0.011272308230400086,
+      "step": 1200
+    },
+    {
+      "epoch": 1.9058684521465143,
+      "grad_norm": 0.4205256228148473,
+      "learning_rate": 8.054589436874428e-08,
+      "loss": 0.012487503886222839,
+      "step": 1210
+    },
+    {
+      "epoch": 1.9216226860968886,
+      "grad_norm": 0.34073410416524363,
+      "learning_rate": 5.9496671949260676e-08,
+      "loss": 0.01165725737810135,
+      "step": 1220
+    },
+    {
+      "epoch": 1.9373769200472628,
+      "grad_norm": 0.43948392858795854,
+      "learning_rate": 4.1613400503550114e-08,
+      "loss": 0.010796928405761718,
+      "step": 1230
+    },
+    {
+      "epoch": 1.9531311539976368,
+      "grad_norm": 0.39555367236463934,
+      "learning_rate": 2.6907539908650472e-08,
+      "loss": 0.011349332332611085,
+      "step": 1240
+    },
+    {
+      "epoch": 1.968885387948011,
+      "grad_norm": 0.48121904906166113,
+      "learning_rate": 1.538851390756435e-08,
+      "loss": 0.011487676203250885,
+      "step": 1250
+    },
+    {
+      "epoch": 1.9846396218983853,
+      "grad_norm": 0.41115781625684744,
+      "learning_rate": 7.063704070380817e-09,
+      "loss": 0.011996541917324067,
+      "step": 1260
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.8985480955065551,
+      "learning_rate": 1.9384450640475006e-09,
+      "loss": 0.01061526983976364,
+      "step": 1270
+    },
+    {
+      "epoch": 2.015754233950374,
+      "grad_norm": 0.32599284226632,
+      "learning_rate": 1.602123383515064e-11,
+      "loss": 0.00712176263332367,
+      "step": 1280
+    },
+    {
+      "epoch": 2.015754233950374,
+      "eval_loss": 0.017056895419955254,
+      "eval_runtime": 155.3119,
+      "eval_samples_per_second": 13.766,
+      "eval_steps_per_second": 3.445,
+      "step": 1280
+    },
+    {
+      "epoch": 2.015754233950374,
+      "step": 1280,
+      "total_flos": 559723689279488.0,
+      "train_loss": 0.06240204306668602,
+      "train_runtime": 14911.9121,
+      "train_samples_per_second": 5.494,
+      "train_steps_per_second": 0.086
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 1280,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 640,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 559723689279488.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}