4b scale-up control SFT checkpoint 12

.gitattributes

@@ -64,3 +64,4 @@ sft_4epoch/coin/checkpoint-24/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 sft_4epoch/coin/checkpoint-36/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 sft_4epoch/coin/checkpoint-48/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 sft_4epoch/control/checkpoint-4/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+sft_4epoch/control/checkpoint-12/tokenizer.json filter=lfs diff=lfs merge=lfs -text

sft_4epoch/control/checkpoint-12/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_4epoch/control/checkpoint-12/config.json

@@ -0,0 +1,111 @@
+{
+  "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": 2560,
+    "initializer_range": 0.02,
+    "intermediate_size": 10240,
+    "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"
+    ],
+    "max_position_embeddings": 131072,
+    "model_type": "gemma3_text",
+    "num_attention_heads": 8,
+    "num_hidden_layers": 34,
+    "num_key_value_heads": 4,
+    "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_4epoch/control/checkpoint-12/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_4epoch/control/checkpoint-12/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:093f5cafdf49e0f4173cf76976fb10368b15f9b2fe26de86ef939843b7a50368
+size 9942783064

sft_4epoch/control/checkpoint-12/optimizer.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fefb329a034d9b4eb6386d4852a1d8b8be623a87bc786463946d014a0d725aa5
+size 15521485995

sft_4epoch/control/checkpoint-12/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_4epoch/control/checkpoint-12/processor_config.json

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

sft_4epoch/control/checkpoint-12/pytorch_model_fsdp.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:31580979e72c0e3987936c53b75a7c624e20823119ea8f7957db412c8824eaa1
+size 9942970813

sft_4epoch/control/checkpoint-12/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d6ae654ed628e46a69a3185a2008a760ecb6fa76a6ed8ea36091fbf51572a9dc
+size 14917

sft_4epoch/control/checkpoint-12/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aca90864f21237f3b2ef7719c284655f031e849a610ff91881a6d9e258037405
+size 14917

sft_4epoch/control/checkpoint-12/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6220224d7693bbb6064e5a896fe2236a351c8f6ee5c2a190bb15e9209803192c
+size 1465

sft_4epoch/control/checkpoint-12/tokenizer.json

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

sft_4epoch/control/checkpoint-12/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_4epoch/control/checkpoint-12/tokens_state.json

@@ -0,0 +1 @@
+{"total": 25165824, "trainable": 15735257}

sft_4epoch/control/checkpoint-12/trainer_state.json

@@ -0,0 +1,202 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.021333333333333333,
+  "eval_steps": 500,
+  "global_step": 12,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0017777777777777779,
+      "grad_norm": 2.875,
+      "learning_rate": 0.0,
+      "loss": 1.09912109375,
+      "memory/device_reserved (GiB)": 40.7,
+      "memory/max_active (GiB)": 34.07,
+      "memory/max_allocated (GiB)": 34.07,
+      "ppl": 3.00153,
+      "step": 1,
+      "tokens/total": 2097152,
+      "tokens/train_per_sec_per_gpu": 631.15,
+      "tokens/trainable": 1319329
+    },
+    {
+      "epoch": 0.0035555555555555557,
+      "grad_norm": 2.890625,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.0859375,
+      "memory/device_reserved (GiB)": 47.77,
+      "memory/max_active (GiB)": 41.3,
+      "memory/max_allocated (GiB)": 41.3,
+      "ppl": 2.96222,
+      "step": 2,
+      "tokens/total": 4194304,
+      "tokens/train_per_sec_per_gpu": 616.0,
+      "tokens/trainable": 2628278
+    },
+    {
+      "epoch": 0.005333333333333333,
+      "grad_norm": 2.578125,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.0758056640625,
+      "memory/device_reserved (GiB)": 47.77,
+      "memory/max_active (GiB)": 41.3,
+      "memory/max_allocated (GiB)": 41.3,
+      "ppl": 2.93235,
+      "step": 3,
+      "tokens/total": 6291456,
+      "tokens/train_per_sec_per_gpu": 655.05,
+      "tokens/trainable": 3948806
+    },
+    {
+      "epoch": 0.0071111111111111115,
+      "grad_norm": 1.71875,
+      "learning_rate": 1e-05,
+      "loss": 1.0162353515625,
+      "memory/device_reserved (GiB)": 47.77,
+      "memory/max_active (GiB)": 41.3,
+      "memory/max_allocated (GiB)": 41.3,
+      "ppl": 2.76277,
+      "step": 4,
+      "tokens/total": 8388608,
+      "tokens/train_per_sec_per_gpu": 598.59,
+      "tokens/trainable": 5229996
+    },
+    {
+      "epoch": 0.008888888888888889,
+      "grad_norm": 2.203125,
+      "learning_rate": 9.98903822616921e-06,
+      "loss": 1.031494140625,
+      "memory/device_reserved (GiB)": 47.8,
+      "memory/max_active (GiB)": 41.3,
+      "memory/max_allocated (GiB)": 41.3,
+      "ppl": 2.80525,
+      "step": 5,
+      "tokens/total": 10485760,
+      "tokens/train_per_sec_per_gpu": 684.59,
+      "tokens/trainable": 6541073
+    },
+    {
+      "epoch": 0.010666666666666666,
+      "grad_norm": 1.84375,
+      "learning_rate": 9.956206309337067e-06,
+      "loss": 1.009765625,
+      "memory/device_reserved (GiB)": 47.82,
+      "memory/max_active (GiB)": 41.3,
+      "memory/max_allocated (GiB)": 41.3,
+      "ppl": 2.74496,
+      "step": 6,
+      "tokens/total": 12582912,
+      "tokens/train_per_sec_per_gpu": 634.42,
+      "tokens/trainable": 7870413
+    },
+    {
+      "epoch": 0.012444444444444444,
+      "grad_norm": 1.234375,
+      "learning_rate": 9.901664203302126e-06,
+      "loss": 0.948486328125,
+      "memory/device_reserved (GiB)": 47.82,
+      "memory/max_active (GiB)": 41.3,
+      "memory/max_allocated (GiB)": 41.3,
+      "ppl": 2.5818,
+      "step": 7,
+      "tokens/total": 14680064,
+      "tokens/train_per_sec_per_gpu": 634.71,
+      "tokens/trainable": 9189777
+    },
+    {
+      "epoch": 0.014222222222222223,
+      "grad_norm": 1.0,
+      "learning_rate": 9.825677631722436e-06,
+      "loss": 0.9532470703125,
+      "memory/device_reserved (GiB)": 47.82,
+      "memory/max_active (GiB)": 41.3,
+      "memory/max_allocated (GiB)": 41.3,
+      "ppl": 2.59412,
+      "step": 8,
+      "tokens/total": 16777216,
+      "tokens/train_per_sec_per_gpu": 619.22,
+      "tokens/trainable": 10479490
+    },
+    {
+      "epoch": 0.016,
+      "grad_norm": 0.8359375,
+      "learning_rate": 9.728616793536588e-06,
+      "loss": 0.944091796875,
+      "memory/device_reserved (GiB)": 47.82,
+      "memory/max_active (GiB)": 41.3,
+      "memory/max_allocated (GiB)": 41.3,
+      "ppl": 2.57048,
+      "step": 9,
+      "tokens/total": 18874368,
+      "tokens/train_per_sec_per_gpu": 652.49,
+      "tokens/trainable": 11784999
+    },
+    {
+      "epoch": 0.017777777777777778,
+      "grad_norm": 0.74609375,
+      "learning_rate": 9.610954559391704e-06,
+      "loss": 0.956298828125,
+      "memory/device_reserved (GiB)": 47.82,
+      "memory/max_active (GiB)": 41.3,
+      "memory/max_allocated (GiB)": 41.3,
+      "ppl": 2.60205,
+      "step": 10,
+      "tokens/total": 20971520,
+      "tokens/train_per_sec_per_gpu": 600.6,
+      "tokens/trainable": 13094236
+    },
+    {
+      "epoch": 0.019555555555555555,
+      "grad_norm": 0.7890625,
+      "learning_rate": 9.473264167865172e-06,
+      "loss": 0.9410400390625,
+      "memory/device_reserved (GiB)": 47.82,
+      "memory/max_active (GiB)": 41.3,
+      "memory/max_allocated (GiB)": 41.3,
+      "ppl": 2.56265,
+      "step": 11,
+      "tokens/total": 23068672,
+      "tokens/train_per_sec_per_gpu": 610.15,
+      "tokens/trainable": 14404718
+    },
+    {
+      "epoch": 0.021333333333333333,
+      "grad_norm": 0.58203125,
+      "learning_rate": 9.316216432703918e-06,
+      "loss": 0.8973388671875,
+      "memory/device_reserved (GiB)": 47.82,
+      "memory/max_active (GiB)": 41.3,
+      "memory/max_allocated (GiB)": 41.3,
+      "ppl": 2.45307,
+      "step": 12,
+      "tokens/total": 25165824,
+      "tokens/train_per_sec_per_gpu": 667.59,
+      "tokens/trainable": 15735257
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 48,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.4722204088139776e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

sft_4epoch/control/checkpoint-12/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:86036685f74c9b56ba03469e3541a7e7b5fb88833612f5f84a8c12343ddc38b4
+size 8913