Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,202 @@
+---
+base_model: vidore/colqwen2-base
+library_name: peft
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.11.1

adapter_config.json

@@ -0,0 +1,26 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "vidore/colqwen2-base",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": "gaussian",
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 32,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": "(.*(model).*(down_proj|gate_proj|up_proj|k_proj|q_proj|v_proj|o_proj).*$|.*(custom_text_proj).*$)",
+  "task_type": "FEATURE_EXTRACTION",
+  "use_dora": false,
+  "use_rslora": false
+}

adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:681fac1e41b24371a7c7fffa7da208c49f3387c4afe8fe4141fe74305701ec14
+size 74018232

added_tokens.json

@@ -0,0 +1,16 @@
+{
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

chat_template.json

@@ -0,0 +1,3 @@
+{
+  "chat_template": "{% set image_count = namespace(value=0) %}{% set video_count = namespace(value=0) %}{% for message in messages %}{% if loop.first and message['role'] != 'system' %}<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n{% endif %}<|im_start|>{{ message['role'] }}\n{% if message['content'] is string %}{{ message['content'] }}<|im_end|>\n{% 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 %}<|im_end|>\n{% endif %}{% endfor %}{% if add_generation_prompt %}<|im_start|>assistant\n{% endif %}"
+}

checkpoint-3694/README.md

@@ -0,0 +1,202 @@
+---
+base_model: ./models/colqwen2_base
+library_name: peft
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.11.1

checkpoint-3694/adapter_config.json

@@ -0,0 +1,26 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "./models/colqwen2_base",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": "gaussian",
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 32,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": "(.*(model).*(down_proj|gate_proj|up_proj|k_proj|q_proj|v_proj|o_proj).*$|.*(custom_text_proj).*$)",
+  "task_type": "FEATURE_EXTRACTION",
+  "use_dora": false,
+  "use_rslora": false
+}

checkpoint-3694/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:681fac1e41b24371a7c7fffa7da208c49f3387c4afe8fe4141fe74305701ec14
+size 74018232

checkpoint-3694/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:101c18553678ddee6c2dccabe942ab94ff178fcfebb431eb8db680a52ec7b52d
+size 148262384

checkpoint-3694/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:df66ef31873d037adee87d02845b81007ae026666e907682dd4875180ea670eb
+size 14244

checkpoint-3694/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f9acc3c13b7f7fcf0134b49941f750c58e40b0ebf0bd4fd58a4084708da1537b
+size 1064

checkpoint-3694/trainer_state.json

@@ -0,0 +1,2904 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 100,
+  "global_step": 3694,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0027070925825663237,
+      "grad_norm": 7.59375,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.7602,
+      "step": 10
+    },
+    {
+      "epoch": 0.005414185165132647,
+      "grad_norm": 5.96875,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.7215,
+      "step": 20
+    },
+    {
+      "epoch": 0.008121277747698972,
+      "grad_norm": 5.46875,
+      "learning_rate": 6e-06,
+      "loss": 0.7449,
+      "step": 30
+    },
+    {
+      "epoch": 0.010828370330265295,
+      "grad_norm": 8.875,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.676,
+      "step": 40
+    },
+    {
+      "epoch": 0.01353546291283162,
+      "grad_norm": 5.15625,
+      "learning_rate": 1e-05,
+      "loss": 0.6583,
+      "step": 50
+    },
+    {
+      "epoch": 0.016242555495397944,
+      "grad_norm": 3.984375,
+      "learning_rate": 1.2e-05,
+      "loss": 0.5955,
+      "step": 60
+    },
+    {
+      "epoch": 0.018949648077964266,
+      "grad_norm": 4.5,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 0.6213,
+      "step": 70
+    },
+    {
+      "epoch": 0.02165674066053059,
+      "grad_norm": 2.96875,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.5522,
+      "step": 80
+    },
+    {
+      "epoch": 0.024363833243096916,
+      "grad_norm": 2.734375,
+      "learning_rate": 1.8e-05,
+      "loss": 0.5117,
+      "step": 90
+    },
+    {
+      "epoch": 0.02707092582566324,
+      "grad_norm": 3.5625,
+      "learning_rate": 2e-05,
+      "loss": 0.4711,
+      "step": 100
+    },
+    {
+      "epoch": 0.02707092582566324,
+      "eval_loss": 0.4397936165332794,
+      "eval_runtime": 107.3458,
+      "eval_samples_per_second": 4.658,
+      "eval_steps_per_second": 0.298,
+      "step": 100
+    },
+    {
+      "epoch": 0.02977801840822956,
+      "grad_norm": 2.390625,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 0.4625,
+      "step": 110
+    },
+    {
+      "epoch": 0.03248511099079589,
+      "grad_norm": 2.34375,
+      "learning_rate": 2.4e-05,
+      "loss": 0.4371,
+      "step": 120
+    },
+    {
+      "epoch": 0.03519220357336221,
+      "grad_norm": 3.25,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 0.4676,
+      "step": 130
+    },
+    {
+      "epoch": 0.03789929615592853,
+      "grad_norm": 2.609375,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.4312,
+      "step": 140
+    },
+    {
+      "epoch": 0.040606388738494856,
+      "grad_norm": 1.90625,
+      "learning_rate": 3e-05,
+      "loss": 0.3917,
+      "step": 150
+    },
+    {
+      "epoch": 0.04331348132106118,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.3707,
+      "step": 160
+    },
+    {
+      "epoch": 0.0460205739036275,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 0.3532,
+      "step": 170
+    },
+    {
+      "epoch": 0.04872766648619383,
+      "grad_norm": 1.875,
+      "learning_rate": 3.6e-05,
+      "loss": 0.3269,
+      "step": 180
+    },
+    {
+      "epoch": 0.051434759068760154,
+      "grad_norm": 1.5390625,
+      "learning_rate": 3.8e-05,
+      "loss": 0.3304,
+      "step": 190
+    },
+    {
+      "epoch": 0.05414185165132648,
+      "grad_norm": 1.4453125,
+      "learning_rate": 4e-05,
+      "loss": 0.288,
+      "step": 200
+    },
+    {
+      "epoch": 0.05414185165132648,
+      "eval_loss": 0.24338363111019135,
+      "eval_runtime": 111.3917,
+      "eval_samples_per_second": 4.489,
+      "eval_steps_per_second": 0.287,
+      "step": 200
+    },
+    {
+      "epoch": 0.0568489442338928,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.2e-05,
+      "loss": 0.2954,
+      "step": 210
+    },
+    {
+      "epoch": 0.05955603681645912,
+      "grad_norm": 1.5625,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.2691,
+      "step": 220
+    },
+    {
+      "epoch": 0.062263129399025445,
+      "grad_norm": 1.4921875,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 0.2763,
+      "step": 230
+    },
+    {
+      "epoch": 0.06497022198159177,
+      "grad_norm": 1.421875,
+      "learning_rate": 4.8e-05,
+      "loss": 0.2552,
+      "step": 240
+    },
+    {
+      "epoch": 0.0676773145641581,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5e-05,
+      "loss": 0.222,
+      "step": 250
+    },
+    {
+      "epoch": 0.07038440714672442,
+      "grad_norm": 1.515625,
+      "learning_rate": 4.98548199767712e-05,
+      "loss": 0.2251,
+      "step": 260
+    },
+    {
+      "epoch": 0.07309149972929074,
+      "grad_norm": 1.734375,
+      "learning_rate": 4.9709639953542396e-05,
+      "loss": 0.2244,
+      "step": 270
+    },
+    {
+      "epoch": 0.07579859231185707,
+      "grad_norm": 1.125,
+      "learning_rate": 4.956445993031359e-05,
+      "loss": 0.2086,
+      "step": 280
+    },
+    {
+      "epoch": 0.07850568489442339,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.9419279907084783e-05,
+      "loss": 0.1871,
+      "step": 290
+    },
+    {
+      "epoch": 0.08121277747698971,
+      "grad_norm": 1.2578125,
+      "learning_rate": 4.9274099883855984e-05,
+      "loss": 0.1851,
+      "step": 300
+    },
+    {
+      "epoch": 0.08121277747698971,
+      "eval_loss": 0.15493114292621613,
+      "eval_runtime": 108.8629,
+      "eval_samples_per_second": 4.593,
+      "eval_steps_per_second": 0.294,
+      "step": 300
+    },
+    {
+      "epoch": 0.08391987005955603,
+      "grad_norm": 0.9296875,
+      "learning_rate": 4.9128919860627184e-05,
+      "loss": 0.2006,
+      "step": 310
+    },
+    {
+      "epoch": 0.08662696264212236,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.898373983739837e-05,
+      "loss": 0.167,
+      "step": 320
+    },
+    {
+      "epoch": 0.08933405522468868,
+      "grad_norm": 1.515625,
+      "learning_rate": 4.883855981416957e-05,
+      "loss": 0.1796,
+      "step": 330
+    },
+    {
+      "epoch": 0.092041147807255,
+      "grad_norm": 0.875,
+      "learning_rate": 4.869337979094077e-05,
+      "loss": 0.1842,
+      "step": 340
+    },
+    {
+      "epoch": 0.09474824038982133,
+      "grad_norm": 1.2578125,
+      "learning_rate": 4.8548199767711965e-05,
+      "loss": 0.1761,
+      "step": 350
+    },
+    {
+      "epoch": 0.09745533297238766,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.8403019744483166e-05,
+      "loss": 0.1975,
+      "step": 360
+    },
+    {
+      "epoch": 0.10016242555495398,
+      "grad_norm": 0.7109375,
+      "learning_rate": 4.825783972125435e-05,
+      "loss": 0.19,
+      "step": 370
+    },
+    {
+      "epoch": 0.10286951813752031,
+      "grad_norm": 1.140625,
+      "learning_rate": 4.811265969802555e-05,
+      "loss": 0.1728,
+      "step": 380
+    },
+    {
+      "epoch": 0.10557661072008663,
+      "grad_norm": 0.62890625,
+      "learning_rate": 4.796747967479675e-05,
+      "loss": 0.142,
+      "step": 390
+    },
+    {
+      "epoch": 0.10828370330265295,
+      "grad_norm": 1.5390625,
+      "learning_rate": 4.782229965156795e-05,
+      "loss": 0.1999,
+      "step": 400
+    },
+    {
+      "epoch": 0.10828370330265295,
+      "eval_loss": 0.1356707364320755,
+      "eval_runtime": 112.7304,
+      "eval_samples_per_second": 4.435,
+      "eval_steps_per_second": 0.284,
+      "step": 400
+    },
+    {
+      "epoch": 0.11099079588521928,
+      "grad_norm": 1.5859375,
+      "learning_rate": 4.767711962833915e-05,
+      "loss": 0.1503,
+      "step": 410
+    },
+    {
+      "epoch": 0.1136978884677856,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.7531939605110334e-05,
+      "loss": 0.1375,
+      "step": 420
+    },
+    {
+      "epoch": 0.11640498105035192,
+      "grad_norm": 0.6875,
+      "learning_rate": 4.7386759581881534e-05,
+      "loss": 0.1399,
+      "step": 430
+    },
+    {
+      "epoch": 0.11911207363291824,
+      "grad_norm": 1.8125,
+      "learning_rate": 4.7241579558652734e-05,
+      "loss": 0.176,
+      "step": 440
+    },
+    {
+      "epoch": 0.12181916621548457,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.709639953542393e-05,
+      "loss": 0.1507,
+      "step": 450
+    },
+    {
+      "epoch": 0.12452625879805089,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.695121951219512e-05,
+      "loss": 0.1461,
+      "step": 460
+    },
+    {
+      "epoch": 0.12723335138061723,
+      "grad_norm": 0.80078125,
+      "learning_rate": 4.680603948896632e-05,
+      "loss": 0.1568,
+      "step": 470
+    },
+    {
+      "epoch": 0.12994044396318355,
+      "grad_norm": 0.44921875,
+      "learning_rate": 4.6660859465737516e-05,
+      "loss": 0.142,
+      "step": 480
+    },
+    {
+      "epoch": 0.13264753654574987,
+      "grad_norm": 0.69140625,
+      "learning_rate": 4.6515679442508716e-05,
+      "loss": 0.1476,
+      "step": 490
+    },
+    {
+      "epoch": 0.1353546291283162,
+      "grad_norm": 1.1875,
+      "learning_rate": 4.637049941927991e-05,
+      "loss": 0.1893,
+      "step": 500
+    },
+    {
+      "epoch": 0.1353546291283162,
+      "eval_loss": 0.13171279430389404,
+      "eval_runtime": 109.4979,
+      "eval_samples_per_second": 4.566,
+      "eval_steps_per_second": 0.292,
+      "step": 500
+    },
+    {
+      "epoch": 0.13806172171088252,
+      "grad_norm": 0.9453125,
+      "learning_rate": 4.62253193960511e-05,
+      "loss": 0.1404,
+      "step": 510
+    },
+    {
+      "epoch": 0.14076881429344884,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.6080139372822303e-05,
+      "loss": 0.1502,
+      "step": 520
+    },
+    {
+      "epoch": 0.14347590687601516,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.59349593495935e-05,
+      "loss": 0.1486,
+      "step": 530
+    },
+    {
+      "epoch": 0.1461829994585815,
+      "grad_norm": 1.53125,
+      "learning_rate": 4.57897793263647e-05,
+      "loss": 0.1222,
+      "step": 540
+    },
+    {
+      "epoch": 0.1488900920411478,
+      "grad_norm": 1.1484375,
+      "learning_rate": 4.564459930313589e-05,
+      "loss": 0.1337,
+      "step": 550
+    },
+    {
+      "epoch": 0.15159718462371413,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.5499419279907084e-05,
+      "loss": 0.1657,
+      "step": 560
+    },
+    {
+      "epoch": 0.15430427720628045,
+      "grad_norm": 0.546875,
+      "learning_rate": 4.5354239256678285e-05,
+      "loss": 0.1443,
+      "step": 570
+    },
+    {
+      "epoch": 0.15701136978884678,
+      "grad_norm": 3.875,
+      "learning_rate": 4.520905923344948e-05,
+      "loss": 0.1368,
+      "step": 580
+    },
+    {
+      "epoch": 0.1597184623714131,
+      "grad_norm": 1.34375,
+      "learning_rate": 4.506387921022068e-05,
+      "loss": 0.1877,
+      "step": 590
+    },
+    {
+      "epoch": 0.16242555495397942,
+      "grad_norm": 2.21875,
+      "learning_rate": 4.491869918699187e-05,
+      "loss": 0.097,
+      "step": 600
+    },
+    {
+      "epoch": 0.16242555495397942,
+      "eval_loss": 0.09729121625423431,
+      "eval_runtime": 109.794,
+      "eval_samples_per_second": 4.554,
+      "eval_steps_per_second": 0.291,
+      "step": 600
+    },
+    {
+      "epoch": 0.16513264753654575,
+      "grad_norm": 1.09375,
+      "learning_rate": 4.4773519163763066e-05,
+      "loss": 0.1161,
+      "step": 610
+    },
+    {
+      "epoch": 0.16783974011911207,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.4628339140534266e-05,
+      "loss": 0.1137,
+      "step": 620
+    },
+    {
+      "epoch": 0.1705468327016784,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.448315911730546e-05,
+      "loss": 0.0872,
+      "step": 630
+    },
+    {
+      "epoch": 0.17325392528424471,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.433797909407666e-05,
+      "loss": 0.1271,
+      "step": 640
+    },
+    {
+      "epoch": 0.17596101786681104,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.4192799070847854e-05,
+      "loss": 0.1168,
+      "step": 650
+    },
+    {
+      "epoch": 0.17866811044937736,
+      "grad_norm": 2.5,
+      "learning_rate": 4.404761904761905e-05,
+      "loss": 0.0993,
+      "step": 660
+    },
+    {
+      "epoch": 0.18137520303194368,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.390243902439025e-05,
+      "loss": 0.0734,
+      "step": 670
+    },
+    {
+      "epoch": 0.18408229561451,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.375725900116144e-05,
+      "loss": 0.1519,
+      "step": 680
+    },
+    {
+      "epoch": 0.18678938819707633,
+      "grad_norm": 1.2734375,
+      "learning_rate": 4.361207897793264e-05,
+      "loss": 0.1401,
+      "step": 690
+    },
+    {
+      "epoch": 0.18949648077964265,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.3466898954703835e-05,
+      "loss": 0.1172,
+      "step": 700
+    },
+    {
+      "epoch": 0.18949648077964265,
+      "eval_loss": 0.09507937729358673,
+      "eval_runtime": 109.6485,
+      "eval_samples_per_second": 4.56,
+      "eval_steps_per_second": 0.292,
+      "step": 700
+    },
+    {
+      "epoch": 0.19220357336220897,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.332171893147503e-05,
+      "loss": 0.1606,
+      "step": 710
+    },
+    {
+      "epoch": 0.19491066594477532,
+      "grad_norm": 1.171875,
+      "learning_rate": 4.317653890824623e-05,
+      "loss": 0.1001,
+      "step": 720
+    },
+    {
+      "epoch": 0.19761775852734165,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.303135888501742e-05,
+      "loss": 0.1171,
+      "step": 730
+    },
+    {
+      "epoch": 0.20032485110990797,
+      "grad_norm": 0.69140625,
+      "learning_rate": 4.2886178861788616e-05,
+      "loss": 0.1162,
+      "step": 740
+    },
+    {
+      "epoch": 0.2030319436924743,
+      "grad_norm": 2.828125,
+      "learning_rate": 4.2740998838559817e-05,
+      "loss": 0.1288,
+      "step": 750
+    },
+    {
+      "epoch": 0.20573903627504062,
+      "grad_norm": 1.875,
+      "learning_rate": 4.259581881533101e-05,
+      "loss": 0.1707,
+      "step": 760
+    },
+    {
+      "epoch": 0.20844612885760694,
+      "grad_norm": 0.8515625,
+      "learning_rate": 4.245063879210221e-05,
+      "loss": 0.1141,
+      "step": 770
+    },
+    {
+      "epoch": 0.21115322144017326,
+      "grad_norm": 2.21875,
+      "learning_rate": 4.2305458768873404e-05,
+      "loss": 0.1184,
+      "step": 780
+    },
+    {
+      "epoch": 0.21386031402273958,
+      "grad_norm": 4.15625,
+      "learning_rate": 4.21602787456446e-05,
+      "loss": 0.1456,
+      "step": 790
+    },
+    {
+      "epoch": 0.2165674066053059,
+      "grad_norm": 1.171875,
+      "learning_rate": 4.20150987224158e-05,
+      "loss": 0.1136,
+      "step": 800
+    },
+    {
+      "epoch": 0.2165674066053059,
+      "eval_loss": 0.09910526871681213,
+      "eval_runtime": 113.6079,
+      "eval_samples_per_second": 4.401,
+      "eval_steps_per_second": 0.282,
+      "step": 800
+    },
+    {
+      "epoch": 0.21927449918787223,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.186991869918699e-05,
+      "loss": 0.0904,
+      "step": 810
+    },
+    {
+      "epoch": 0.22198159177043855,
+      "grad_norm": 2.8125,
+      "learning_rate": 4.172473867595819e-05,
+      "loss": 0.1165,
+      "step": 820
+    },
+    {
+      "epoch": 0.22468868435300487,
+      "grad_norm": 0.65625,
+      "learning_rate": 4.157955865272939e-05,
+      "loss": 0.1037,
+      "step": 830
+    },
+    {
+      "epoch": 0.2273957769355712,
+      "grad_norm": 6.03125,
+      "learning_rate": 4.143437862950058e-05,
+      "loss": 0.1116,
+      "step": 840
+    },
+    {
+      "epoch": 0.23010286951813752,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.128919860627178e-05,
+      "loss": 0.1717,
+      "step": 850
+    },
+    {
+      "epoch": 0.23280996210070384,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.114401858304297e-05,
+      "loss": 0.1005,
+      "step": 860
+    },
+    {
+      "epoch": 0.23551705468327017,
+      "grad_norm": 0.62890625,
+      "learning_rate": 4.099883855981417e-05,
+      "loss": 0.1224,
+      "step": 870
+    },
+    {
+      "epoch": 0.2382241472658365,
+      "grad_norm": 0.96875,
+      "learning_rate": 4.085365853658537e-05,
+      "loss": 0.0655,
+      "step": 880
+    },
+    {
+      "epoch": 0.2409312398484028,
+      "grad_norm": 1.34375,
+      "learning_rate": 4.070847851335656e-05,
+      "loss": 0.1179,
+      "step": 890
+    },
+    {
+      "epoch": 0.24363833243096913,
+      "grad_norm": 0.435546875,
+      "learning_rate": 4.056329849012776e-05,
+      "loss": 0.1007,
+      "step": 900
+    },
+    {
+      "epoch": 0.24363833243096913,
+      "eval_loss": 0.10046005249023438,
+      "eval_runtime": 109.6909,
+      "eval_samples_per_second": 4.558,
+      "eval_steps_per_second": 0.292,
+      "step": 900
+    },
+    {
+      "epoch": 0.24634542501353546,
+      "grad_norm": 2.03125,
+      "learning_rate": 4.0418118466898954e-05,
+      "loss": 0.1315,
+      "step": 910
+    },
+    {
+      "epoch": 0.24905251759610178,
+      "grad_norm": 2.125,
+      "learning_rate": 4.0272938443670155e-05,
+      "loss": 0.1409,
+      "step": 920
+    },
+    {
+      "epoch": 0.2517596101786681,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.012775842044135e-05,
+      "loss": 0.1195,
+      "step": 930
+    },
+    {
+      "epoch": 0.25446670276123445,
+      "grad_norm": 1.140625,
+      "learning_rate": 3.998257839721254e-05,
+      "loss": 0.0969,
+      "step": 940
+    },
+    {
+      "epoch": 0.25717379534380075,
+      "grad_norm": 0.84375,
+      "learning_rate": 3.983739837398374e-05,
+      "loss": 0.0717,
+      "step": 950
+    },
+    {
+      "epoch": 0.2598808879263671,
+      "grad_norm": 2.0625,
+      "learning_rate": 3.969221835075494e-05,
+      "loss": 0.0931,
+      "step": 960
+    },
+    {
+      "epoch": 0.2625879805089334,
+      "grad_norm": 1.1796875,
+      "learning_rate": 3.9547038327526136e-05,
+      "loss": 0.1483,
+      "step": 970
+    },
+    {
+      "epoch": 0.26529507309149974,
+      "grad_norm": 1.96875,
+      "learning_rate": 3.940185830429733e-05,
+      "loss": 0.1165,
+      "step": 980
+    },
+    {
+      "epoch": 0.26800216567406604,
+      "grad_norm": 2.703125,
+      "learning_rate": 3.925667828106852e-05,
+      "loss": 0.1092,
+      "step": 990
+    },
+    {
+      "epoch": 0.2707092582566324,
+      "grad_norm": 1.6171875,
+      "learning_rate": 3.9111498257839724e-05,
+      "loss": 0.0796,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2707092582566324,
+      "eval_loss": 0.0991658866405487,
+      "eval_runtime": 109.5652,
+      "eval_samples_per_second": 4.563,
+      "eval_steps_per_second": 0.292,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2734163508391987,
+      "grad_norm": 0.546875,
+      "learning_rate": 3.8966318234610924e-05,
+      "loss": 0.1199,
+      "step": 1010
+    },
+    {
+      "epoch": 0.27612344342176504,
+      "grad_norm": 2.59375,
+      "learning_rate": 3.882113821138211e-05,
+      "loss": 0.0766,
+      "step": 1020
+    },
+    {
+      "epoch": 0.27883053600433133,
+      "grad_norm": 3.421875,
+      "learning_rate": 3.867595818815331e-05,
+      "loss": 0.1308,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2815376285868977,
+      "grad_norm": 3.71875,
+      "learning_rate": 3.8530778164924505e-05,
+      "loss": 0.1174,
+      "step": 1040
+    },
+    {
+      "epoch": 0.284244721169464,
+      "grad_norm": 4.3125,
+      "learning_rate": 3.8385598141695705e-05,
+      "loss": 0.1449,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2869518137520303,
+      "grad_norm": 2.0625,
+      "learning_rate": 3.8240418118466905e-05,
+      "loss": 0.1114,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2896589063345966,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.809523809523809e-05,
+      "loss": 0.1079,
+      "step": 1070
+    },
+    {
+      "epoch": 0.292365998917163,
+      "grad_norm": 1.7109375,
+      "learning_rate": 3.795005807200929e-05,
+      "loss": 0.108,
+      "step": 1080
+    },
+    {
+      "epoch": 0.29507309149972927,
+      "grad_norm": 1.171875,
+      "learning_rate": 3.780487804878049e-05,
+      "loss": 0.1189,
+      "step": 1090
+    },
+    {
+      "epoch": 0.2977801840822956,
+      "grad_norm": 2.328125,
+      "learning_rate": 3.7659698025551686e-05,
+      "loss": 0.1266,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2977801840822956,
+      "eval_loss": 0.11428254842758179,
+      "eval_runtime": 62.7583,
+      "eval_samples_per_second": 7.967,
+      "eval_steps_per_second": 0.51,
+      "step": 1100
+    },
+    {
+      "epoch": 0.3004872766648619,
+      "grad_norm": 0.90234375,
+      "learning_rate": 3.751451800232289e-05,
+      "loss": 0.1645,
+      "step": 1110
+    },
+    {
+      "epoch": 0.30319436924742826,
+      "grad_norm": 1.6015625,
+      "learning_rate": 3.7369337979094074e-05,
+      "loss": 0.1271,
+      "step": 1120
+    },
+    {
+      "epoch": 0.30590146182999456,
+      "grad_norm": 0.53515625,
+      "learning_rate": 3.7224157955865274e-05,
+      "loss": 0.0863,
+      "step": 1130
+    },
+    {
+      "epoch": 0.3086085544125609,
+      "grad_norm": 1.3828125,
+      "learning_rate": 3.7078977932636474e-05,
+      "loss": 0.0994,
+      "step": 1140
+    },
+    {
+      "epoch": 0.31131564699512726,
+      "grad_norm": 2.703125,
+      "learning_rate": 3.693379790940767e-05,
+      "loss": 0.1003,
+      "step": 1150
+    },
+    {
+      "epoch": 0.31402273957769355,
+      "grad_norm": 0.76953125,
+      "learning_rate": 3.678861788617886e-05,
+      "loss": 0.0944,
+      "step": 1160
+    },
+    {
+      "epoch": 0.3167298321602599,
+      "grad_norm": 2.546875,
+      "learning_rate": 3.664343786295006e-05,
+      "loss": 0.1752,
+      "step": 1170
+    },
+    {
+      "epoch": 0.3194369247428262,
+      "grad_norm": 1.5,
+      "learning_rate": 3.6498257839721255e-05,
+      "loss": 0.1714,
+      "step": 1180
+    },
+    {
+      "epoch": 0.32214401732539255,
+      "grad_norm": 1.1171875,
+      "learning_rate": 3.6353077816492456e-05,
+      "loss": 0.0958,
+      "step": 1190
+    },
+    {
+      "epoch": 0.32485110990795885,
+      "grad_norm": 3.765625,
+      "learning_rate": 3.620789779326365e-05,
+      "loss": 0.0985,
+      "step": 1200
+    },
+    {
+      "epoch": 0.32485110990795885,
+      "eval_loss": 0.11661399900913239,
+      "eval_runtime": 61.7024,
+      "eval_samples_per_second": 8.103,
+      "eval_steps_per_second": 0.519,
+      "step": 1200
+    },
+    {
+      "epoch": 0.3275582024905252,
+      "grad_norm": 1.0546875,
+      "learning_rate": 3.606271777003484e-05,
+      "loss": 0.0975,
+      "step": 1210
+    },
+    {
+      "epoch": 0.3302652950730915,
+      "grad_norm": 1.7578125,
+      "learning_rate": 3.591753774680604e-05,
+      "loss": 0.1175,
+      "step": 1220
+    },
+    {
+      "epoch": 0.33297238765565784,
+      "grad_norm": 2.15625,
+      "learning_rate": 3.577235772357724e-05,
+      "loss": 0.1331,
+      "step": 1230
+    },
+    {
+      "epoch": 0.33567948023822414,
+      "grad_norm": 1.2890625,
+      "learning_rate": 3.562717770034844e-05,
+      "loss": 0.1184,
+      "step": 1240
+    },
+    {
+      "epoch": 0.3383865728207905,
+      "grad_norm": 1.421875,
+      "learning_rate": 3.548199767711963e-05,
+      "loss": 0.1299,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3410936654033568,
+      "grad_norm": 0.95703125,
+      "learning_rate": 3.5336817653890824e-05,
+      "loss": 0.1042,
+      "step": 1260
+    },
+    {
+      "epoch": 0.34380075798592313,
+      "grad_norm": 1.25,
+      "learning_rate": 3.5191637630662025e-05,
+      "loss": 0.0876,
+      "step": 1270
+    },
+    {
+      "epoch": 0.34650785056848943,
+      "grad_norm": 1.6015625,
+      "learning_rate": 3.504645760743322e-05,
+      "loss": 0.1529,
+      "step": 1280
+    },
+    {
+      "epoch": 0.3492149431510558,
+      "grad_norm": 2.296875,
+      "learning_rate": 3.490127758420442e-05,
+      "loss": 0.1266,
+      "step": 1290
+    },
+    {
+      "epoch": 0.3519220357336221,
+      "grad_norm": 0.5703125,
+      "learning_rate": 3.475609756097561e-05,
+      "loss": 0.1096,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3519220357336221,
+      "eval_loss": 0.11386878788471222,
+      "eval_runtime": 62.6792,
+      "eval_samples_per_second": 7.977,
+      "eval_steps_per_second": 0.511,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3546291283161884,
+      "grad_norm": 1.0390625,
+      "learning_rate": 3.4610917537746806e-05,
+      "loss": 0.1186,
+      "step": 1310
+    },
+    {
+      "epoch": 0.3573362208987547,
+      "grad_norm": 2.03125,
+      "learning_rate": 3.4465737514518006e-05,
+      "loss": 0.1209,
+      "step": 1320
+    },
+    {
+      "epoch": 0.36004331348132107,
+      "grad_norm": 2.171875,
+      "learning_rate": 3.43205574912892e-05,
+      "loss": 0.1562,
+      "step": 1330
+    },
+    {
+      "epoch": 0.36275040606388737,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.41753774680604e-05,
+      "loss": 0.139,
+      "step": 1340
+    },
+    {
+      "epoch": 0.3654574986464537,
+      "grad_norm": 1.375,
+      "learning_rate": 3.4030197444831594e-05,
+      "loss": 0.117,
+      "step": 1350
+    },
+    {
+      "epoch": 0.36816459122902,
+      "grad_norm": 1.0859375,
+      "learning_rate": 3.388501742160279e-05,
+      "loss": 0.0775,
+      "step": 1360
+    },
+    {
+      "epoch": 0.37087168381158636,
+      "grad_norm": 1.1328125,
+      "learning_rate": 3.373983739837399e-05,
+      "loss": 0.1158,
+      "step": 1370
+    },
+    {
+      "epoch": 0.37357877639415266,
+      "grad_norm": 0.890625,
+      "learning_rate": 3.359465737514518e-05,
+      "loss": 0.104,
+      "step": 1380
+    },
+    {
+      "epoch": 0.376285868976719,
+      "grad_norm": 2.21875,
+      "learning_rate": 3.344947735191638e-05,
+      "loss": 0.1519,
+      "step": 1390
+    },
+    {
+      "epoch": 0.3789929615592853,
+      "grad_norm": 0.91015625,
+      "learning_rate": 3.3304297328687575e-05,
+      "loss": 0.1034,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3789929615592853,
+      "eval_loss": 0.11311852931976318,
+      "eval_runtime": 62.6576,
+      "eval_samples_per_second": 7.98,
+      "eval_steps_per_second": 0.511,
+      "step": 1400
+    },
+    {
+      "epoch": 0.38170005414185165,
+      "grad_norm": 2.4375,
+      "learning_rate": 3.315911730545877e-05,
+      "loss": 0.1152,
+      "step": 1410
+    },
+    {
+      "epoch": 0.38440714672441795,
+      "grad_norm": 2.859375,
+      "learning_rate": 3.301393728222997e-05,
+      "loss": 0.1335,
+      "step": 1420
+    },
+    {
+      "epoch": 0.3871142393069843,
+      "grad_norm": 1.453125,
+      "learning_rate": 3.286875725900116e-05,
+      "loss": 0.1011,
+      "step": 1430
+    },
+    {
+      "epoch": 0.38982133188955065,
+      "grad_norm": 1.3515625,
+      "learning_rate": 3.2723577235772356e-05,
+      "loss": 0.11,
+      "step": 1440
+    },
+    {
+      "epoch": 0.39252842447211694,
+      "grad_norm": 0.5390625,
+      "learning_rate": 3.2578397212543556e-05,
+      "loss": 0.1462,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3952355170546833,
+      "grad_norm": 0.6171875,
+      "learning_rate": 3.243321718931475e-05,
+      "loss": 0.1175,
+      "step": 1460
+    },
+    {
+      "epoch": 0.3979426096372496,
+      "grad_norm": 0.63671875,
+      "learning_rate": 3.228803716608595e-05,
+      "loss": 0.0832,
+      "step": 1470
+    },
+    {
+      "epoch": 0.40064970221981594,
+      "grad_norm": 1.6484375,
+      "learning_rate": 3.2142857142857144e-05,
+      "loss": 0.1224,
+      "step": 1480
+    },
+    {
+      "epoch": 0.40335679480238223,
+      "grad_norm": 0.9296875,
+      "learning_rate": 3.199767711962834e-05,
+      "loss": 0.118,
+      "step": 1490
+    },
+    {
+      "epoch": 0.4060638873849486,
+      "grad_norm": 1.4296875,
+      "learning_rate": 3.185249709639954e-05,
+      "loss": 0.1143,
+      "step": 1500
+    },
+    {
+      "epoch": 0.4060638873849486,
+      "eval_loss": 0.11030779778957367,
+      "eval_runtime": 62.7843,
+      "eval_samples_per_second": 7.964,
+      "eval_steps_per_second": 0.51,
+      "step": 1500
+    },
+    {
+      "epoch": 0.4087709799675149,
+      "grad_norm": 1.109375,
+      "learning_rate": 3.170731707317073e-05,
+      "loss": 0.0953,
+      "step": 1510
+    },
+    {
+      "epoch": 0.41147807255008123,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3.156213704994193e-05,
+      "loss": 0.12,
+      "step": 1520
+    },
+    {
+      "epoch": 0.4141851651326475,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.1416957026713125e-05,
+      "loss": 0.1243,
+      "step": 1530
+    },
+    {
+      "epoch": 0.4168922577152139,
+      "grad_norm": 1.9140625,
+      "learning_rate": 3.127177700348432e-05,
+      "loss": 0.1008,
+      "step": 1540
+    },
+    {
+      "epoch": 0.41959935029778017,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.112659698025552e-05,
+      "loss": 0.1435,
+      "step": 1550
+    },
+    {
+      "epoch": 0.4223064428803465,
+      "grad_norm": 1.3046875,
+      "learning_rate": 3.098141695702671e-05,
+      "loss": 0.0719,
+      "step": 1560
+    },
+    {
+      "epoch": 0.4250135354629128,
+      "grad_norm": 0.72265625,
+      "learning_rate": 3.083623693379791e-05,
+      "loss": 0.1168,
+      "step": 1570
+    },
+    {
+      "epoch": 0.42772062804547917,
+      "grad_norm": 2.859375,
+      "learning_rate": 3.069105691056911e-05,
+      "loss": 0.1393,
+      "step": 1580
+    },
+    {
+      "epoch": 0.43042772062804546,
+      "grad_norm": 1.5078125,
+      "learning_rate": 3.05458768873403e-05,
+      "loss": 0.1215,
+      "step": 1590
+    },
+    {
+      "epoch": 0.4331348132106118,
+      "grad_norm": 1.0625,
+      "learning_rate": 3.04006968641115e-05,
+      "loss": 0.1051,
+      "step": 1600
+    },
+    {
+      "epoch": 0.4331348132106118,
+      "eval_loss": 0.11309642344713211,
+      "eval_runtime": 62.3543,
+      "eval_samples_per_second": 8.019,
+      "eval_steps_per_second": 0.513,
+      "step": 1600
+    },
+    {
+      "epoch": 0.4358419057931781,
+      "grad_norm": 1.0390625,
+      "learning_rate": 3.0255516840882698e-05,
+      "loss": 0.0818,
+      "step": 1610
+    },
+    {
+      "epoch": 0.43854899837574446,
+      "grad_norm": 0.74609375,
+      "learning_rate": 3.0110336817653895e-05,
+      "loss": 0.1048,
+      "step": 1620
+    },
+    {
+      "epoch": 0.44125609095831075,
+      "grad_norm": 1.359375,
+      "learning_rate": 2.9965156794425088e-05,
+      "loss": 0.0984,
+      "step": 1630
+    },
+    {
+      "epoch": 0.4439631835408771,
+      "grad_norm": 0.9453125,
+      "learning_rate": 2.9819976771196285e-05,
+      "loss": 0.0861,
+      "step": 1640
+    },
+    {
+      "epoch": 0.4466702761234434,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.9674796747967482e-05,
+      "loss": 0.1451,
+      "step": 1650
+    },
+    {
+      "epoch": 0.44937736870600975,
+      "grad_norm": 1.296875,
+      "learning_rate": 2.952961672473868e-05,
+      "loss": 0.1273,
+      "step": 1660
+    },
+    {
+      "epoch": 0.45208446128857604,
+      "grad_norm": 1.359375,
+      "learning_rate": 2.9384436701509873e-05,
+      "loss": 0.1287,
+      "step": 1670
+    },
+    {
+      "epoch": 0.4547915538711424,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.923925667828107e-05,
+      "loss": 0.1465,
+      "step": 1680
+    },
+    {
+      "epoch": 0.4574986464537087,
+      "grad_norm": 0.58984375,
+      "learning_rate": 2.9094076655052267e-05,
+      "loss": 0.1009,
+      "step": 1690
+    },
+    {
+      "epoch": 0.46020573903627504,
+      "grad_norm": 0.5390625,
+      "learning_rate": 2.8948896631823464e-05,
+      "loss": 0.0867,
+      "step": 1700
+    },
+    {
+      "epoch": 0.46020573903627504,
+      "eval_loss": 0.10628359764814377,
+      "eval_runtime": 62.0852,
+      "eval_samples_per_second": 8.053,
+      "eval_steps_per_second": 0.515,
+      "step": 1700
+    },
+    {
+      "epoch": 0.4629128316188414,
+      "grad_norm": 2.28125,
+      "learning_rate": 2.880371660859466e-05,
+      "loss": 0.1038,
+      "step": 1710
+    },
+    {
+      "epoch": 0.4656199242014077,
+      "grad_norm": 0.72265625,
+      "learning_rate": 2.8658536585365854e-05,
+      "loss": 0.1063,
+      "step": 1720
+    },
+    {
+      "epoch": 0.46832701678397404,
+      "grad_norm": 2.1875,
+      "learning_rate": 2.851335656213705e-05,
+      "loss": 0.1509,
+      "step": 1730
+    },
+    {
+      "epoch": 0.47103410936654033,
+      "grad_norm": 0.56640625,
+      "learning_rate": 2.8368176538908248e-05,
+      "loss": 0.1043,
+      "step": 1740
+    },
+    {
+      "epoch": 0.4737412019491067,
+      "grad_norm": 1.125,
+      "learning_rate": 2.8222996515679445e-05,
+      "loss": 0.1032,
+      "step": 1750
+    },
+    {
+      "epoch": 0.476448294531673,
+      "grad_norm": 0.51171875,
+      "learning_rate": 2.8077816492450642e-05,
+      "loss": 0.0817,
+      "step": 1760
+    },
+    {
+      "epoch": 0.47915538711423933,
+      "grad_norm": 2.09375,
+      "learning_rate": 2.7932636469221835e-05,
+      "loss": 0.1285,
+      "step": 1770
+    },
+    {
+      "epoch": 0.4818624796968056,
+      "grad_norm": 0.78515625,
+      "learning_rate": 2.7787456445993032e-05,
+      "loss": 0.1446,
+      "step": 1780
+    },
+    {
+      "epoch": 0.484569572279372,
+      "grad_norm": 1.265625,
+      "learning_rate": 2.764227642276423e-05,
+      "loss": 0.1086,
+      "step": 1790
+    },
+    {
+      "epoch": 0.48727666486193827,
+      "grad_norm": 1.8046875,
+      "learning_rate": 2.7497096399535426e-05,
+      "loss": 0.1001,
+      "step": 1800
+    },
+    {
+      "epoch": 0.48727666486193827,
+      "eval_loss": 0.10519789159297943,
+      "eval_runtime": 64.2198,
+      "eval_samples_per_second": 7.786,
+      "eval_steps_per_second": 0.498,
+      "step": 1800
+    },
+    {
+      "epoch": 0.4899837574445046,
+      "grad_norm": 1.9296875,
+      "learning_rate": 2.735191637630662e-05,
+      "loss": 0.087,
+      "step": 1810
+    },
+    {
+      "epoch": 0.4926908500270709,
+      "grad_norm": 1.15625,
+      "learning_rate": 2.7206736353077817e-05,
+      "loss": 0.1236,
+      "step": 1820
+    },
+    {
+      "epoch": 0.49539794260963727,
+      "grad_norm": 1.1875,
+      "learning_rate": 2.7061556329849014e-05,
+      "loss": 0.0777,
+      "step": 1830
+    },
+    {
+      "epoch": 0.49810503519220356,
+      "grad_norm": 1.1640625,
+      "learning_rate": 2.691637630662021e-05,
+      "loss": 0.0987,
+      "step": 1840
+    },
+    {
+      "epoch": 0.5008121277747699,
+      "grad_norm": 0.96484375,
+      "learning_rate": 2.6771196283391408e-05,
+      "loss": 0.1226,
+      "step": 1850
+    },
+    {
+      "epoch": 0.5035192203573362,
+      "grad_norm": 1.3828125,
+      "learning_rate": 2.66260162601626e-05,
+      "loss": 0.1295,
+      "step": 1860
+    },
+    {
+      "epoch": 0.5062263129399025,
+      "grad_norm": 0.60546875,
+      "learning_rate": 2.6480836236933798e-05,
+      "loss": 0.0666,
+      "step": 1870
+    },
+    {
+      "epoch": 0.5089334055224689,
+      "grad_norm": 2.46875,
+      "learning_rate": 2.6335656213704995e-05,
+      "loss": 0.0885,
+      "step": 1880
+    },
+    {
+      "epoch": 0.5116404981050352,
+      "grad_norm": 2.5,
+      "learning_rate": 2.6190476190476192e-05,
+      "loss": 0.075,
+      "step": 1890
+    },
+    {
+      "epoch": 0.5143475906876015,
+      "grad_norm": 0.6875,
+      "learning_rate": 2.604529616724739e-05,
+      "loss": 0.0907,
+      "step": 1900
+    },
+    {
+      "epoch": 0.5143475906876015,
+      "eval_loss": 0.1059199795126915,
+      "eval_runtime": 62.4159,
+      "eval_samples_per_second": 8.011,
+      "eval_steps_per_second": 0.513,
+      "step": 1900
+    },
+    {
+      "epoch": 0.5170546832701678,
+      "grad_norm": 1.1875,
+      "learning_rate": 2.5900116144018583e-05,
+      "loss": 0.1199,
+      "step": 1910
+    },
+    {
+      "epoch": 0.5197617758527342,
+      "grad_norm": 0.7265625,
+      "learning_rate": 2.575493612078978e-05,
+      "loss": 0.112,
+      "step": 1920
+    },
+    {
+      "epoch": 0.5224688684353005,
+      "grad_norm": 3.078125,
+      "learning_rate": 2.5609756097560977e-05,
+      "loss": 0.1126,
+      "step": 1930
+    },
+    {
+      "epoch": 0.5251759610178668,
+      "grad_norm": 2.109375,
+      "learning_rate": 2.5464576074332174e-05,
+      "loss": 0.1203,
+      "step": 1940
+    },
+    {
+      "epoch": 0.5278830536004331,
+      "grad_norm": 0.71875,
+      "learning_rate": 2.5319396051103367e-05,
+      "loss": 0.1226,
+      "step": 1950
+    },
+    {
+      "epoch": 0.5305901461829995,
+      "grad_norm": 0.51171875,
+      "learning_rate": 2.5174216027874564e-05,
+      "loss": 0.0856,
+      "step": 1960
+    },
+    {
+      "epoch": 0.5332972387655658,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.502903600464576e-05,
+      "loss": 0.112,
+      "step": 1970
+    },
+    {
+      "epoch": 0.5360043313481321,
+      "grad_norm": 0.62890625,
+      "learning_rate": 2.4883855981416958e-05,
+      "loss": 0.1082,
+      "step": 1980
+    },
+    {
+      "epoch": 0.5387114239306985,
+      "grad_norm": 0.765625,
+      "learning_rate": 2.4738675958188155e-05,
+      "loss": 0.1261,
+      "step": 1990
+    },
+    {
+      "epoch": 0.5414185165132648,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.4593495934959352e-05,
+      "loss": 0.1182,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5414185165132648,
+      "eval_loss": 0.10393251478672028,
+      "eval_runtime": 62.4399,
+      "eval_samples_per_second": 8.008,
+      "eval_steps_per_second": 0.512,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5441256090958311,
+      "grad_norm": 0.9921875,
+      "learning_rate": 2.4448315911730546e-05,
+      "loss": 0.0977,
+      "step": 2010
+    },
+    {
+      "epoch": 0.5468327016783974,
+      "grad_norm": 1.640625,
+      "learning_rate": 2.4303135888501743e-05,
+      "loss": 0.0772,
+      "step": 2020
+    },
+    {
+      "epoch": 0.5495397942609638,
+      "grad_norm": 1.9921875,
+      "learning_rate": 2.415795586527294e-05,
+      "loss": 0.1043,
+      "step": 2030
+    },
+    {
+      "epoch": 0.5522468868435301,
+      "grad_norm": 0.458984375,
+      "learning_rate": 2.4012775842044136e-05,
+      "loss": 0.083,
+      "step": 2040
+    },
+    {
+      "epoch": 0.5549539794260964,
+      "grad_norm": 1.1796875,
+      "learning_rate": 2.3867595818815333e-05,
+      "loss": 0.0954,
+      "step": 2050
+    },
+    {
+      "epoch": 0.5576610720086627,
+      "grad_norm": 0.90625,
+      "learning_rate": 2.3722415795586527e-05,
+      "loss": 0.0861,
+      "step": 2060
+    },
+    {
+      "epoch": 0.5603681645912291,
+      "grad_norm": 0.5625,
+      "learning_rate": 2.3577235772357724e-05,
+      "loss": 0.109,
+      "step": 2070
+    },
+    {
+      "epoch": 0.5630752571737954,
+      "grad_norm": 0.6640625,
+      "learning_rate": 2.343205574912892e-05,
+      "loss": 0.0902,
+      "step": 2080
+    },
+    {
+      "epoch": 0.5657823497563617,
+      "grad_norm": 1.421875,
+      "learning_rate": 2.3286875725900118e-05,
+      "loss": 0.1021,
+      "step": 2090
+    },
+    {
+      "epoch": 0.568489442338928,
+      "grad_norm": 1.5546875,
+      "learning_rate": 2.314169570267131e-05,
+      "loss": 0.0855,
+      "step": 2100
+    },
+    {
+      "epoch": 0.568489442338928,
+      "eval_loss": 0.1053832545876503,
+      "eval_runtime": 62.817,
+      "eval_samples_per_second": 7.96,
+      "eval_steps_per_second": 0.509,
+      "step": 2100
+    },
+    {
+      "epoch": 0.5711965349214944,
+      "grad_norm": 0.828125,
+      "learning_rate": 2.299651567944251e-05,
+      "loss": 0.0938,
+      "step": 2110
+    },
+    {
+      "epoch": 0.5739036275040607,
+      "grad_norm": 0.953125,
+      "learning_rate": 2.285133565621371e-05,
+      "loss": 0.064,
+      "step": 2120
+    },
+    {
+      "epoch": 0.576610720086627,
+      "grad_norm": 1.609375,
+      "learning_rate": 2.2706155632984902e-05,
+      "loss": 0.0987,
+      "step": 2130
+    },
+    {
+      "epoch": 0.5793178126691932,
+      "grad_norm": 1.6484375,
+      "learning_rate": 2.25609756097561e-05,
+      "loss": 0.0876,
+      "step": 2140
+    },
+    {
+      "epoch": 0.5820249052517596,
+      "grad_norm": 0.953125,
+      "learning_rate": 2.2415795586527293e-05,
+      "loss": 0.1171,
+      "step": 2150
+    },
+    {
+      "epoch": 0.584731997834326,
+      "grad_norm": 2.328125,
+      "learning_rate": 2.227061556329849e-05,
+      "loss": 0.0901,
+      "step": 2160
+    },
+    {
+      "epoch": 0.5874390904168922,
+      "grad_norm": 0.78125,
+      "learning_rate": 2.2125435540069687e-05,
+      "loss": 0.1097,
+      "step": 2170
+    },
+    {
+      "epoch": 0.5901461829994585,
+      "grad_norm": 0.84375,
+      "learning_rate": 2.1980255516840884e-05,
+      "loss": 0.0678,
+      "step": 2180
+    },
+    {
+      "epoch": 0.5928532755820249,
+      "grad_norm": 1.515625,
+      "learning_rate": 2.183507549361208e-05,
+      "loss": 0.1216,
+      "step": 2190
+    },
+    {
+      "epoch": 0.5955603681645912,
+      "grad_norm": 0.94140625,
+      "learning_rate": 2.1689895470383274e-05,
+      "loss": 0.0803,
+      "step": 2200
+    },
+    {
+      "epoch": 0.5955603681645912,
+      "eval_loss": 0.1051289513707161,
+      "eval_runtime": 62.5369,
+      "eval_samples_per_second": 7.995,
+      "eval_steps_per_second": 0.512,
+      "step": 2200
+    },
+    {
+      "epoch": 0.5982674607471575,
+      "grad_norm": 1.0625,
+      "learning_rate": 2.1544715447154475e-05,
+      "loss": 0.1353,
+      "step": 2210
+    },
+    {
+      "epoch": 0.6009745533297238,
+      "grad_norm": 0.77734375,
+      "learning_rate": 2.1399535423925668e-05,
+      "loss": 0.0772,
+      "step": 2220
+    },
+    {
+      "epoch": 0.6036816459122902,
+      "grad_norm": 2.90625,
+      "learning_rate": 2.1254355400696865e-05,
+      "loss": 0.1301,
+      "step": 2230
+    },
+    {
+      "epoch": 0.6063887384948565,
+      "grad_norm": 0.443359375,
+      "learning_rate": 2.110917537746806e-05,
+      "loss": 0.1308,
+      "step": 2240
+    },
+    {
+      "epoch": 0.6090958310774228,
+      "grad_norm": 0.75,
+      "learning_rate": 2.096399535423926e-05,
+      "loss": 0.0804,
+      "step": 2250
+    },
+    {
+      "epoch": 0.6118029236599891,
+      "grad_norm": 1.1953125,
+      "learning_rate": 2.0818815331010456e-05,
+      "loss": 0.0908,
+      "step": 2260
+    },
+    {
+      "epoch": 0.6145100162425555,
+      "grad_norm": 2.15625,
+      "learning_rate": 2.067363530778165e-05,
+      "loss": 0.1008,
+      "step": 2270
+    },
+    {
+      "epoch": 0.6172171088251218,
+      "grad_norm": 1.4296875,
+      "learning_rate": 2.0528455284552847e-05,
+      "loss": 0.107,
+      "step": 2280
+    },
+    {
+      "epoch": 0.6199242014076881,
+      "grad_norm": 2.625,
+      "learning_rate": 2.038327526132404e-05,
+      "loss": 0.1058,
+      "step": 2290
+    },
+    {
+      "epoch": 0.6226312939902545,
+      "grad_norm": 1.265625,
+      "learning_rate": 2.023809523809524e-05,
+      "loss": 0.1034,
+      "step": 2300
+    },
+    {
+      "epoch": 0.6226312939902545,
+      "eval_loss": 0.10145323723554611,
+      "eval_runtime": 63.2925,
+      "eval_samples_per_second": 7.9,
+      "eval_steps_per_second": 0.506,
+      "step": 2300
+    },
+    {
+      "epoch": 0.6253383865728208,
+      "grad_norm": 1.359375,
+      "learning_rate": 2.0092915214866434e-05,
+      "loss": 0.0867,
+      "step": 2310
+    },
+    {
+      "epoch": 0.6280454791553871,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.994773519163763e-05,
+      "loss": 0.1246,
+      "step": 2320
+    },
+    {
+      "epoch": 0.6307525717379534,
+      "grad_norm": 2.6875,
+      "learning_rate": 1.9802555168408828e-05,
+      "loss": 0.0807,
+      "step": 2330
+    },
+    {
+      "epoch": 0.6334596643205198,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.9657375145180025e-05,
+      "loss": 0.1042,
+      "step": 2340
+    },
+    {
+      "epoch": 0.6361667569030861,
+      "grad_norm": 0.470703125,
+      "learning_rate": 1.9512195121951222e-05,
+      "loss": 0.0838,
+      "step": 2350
+    },
+    {
+      "epoch": 0.6388738494856524,
+      "grad_norm": 2.375,
+      "learning_rate": 1.9367015098722416e-05,
+      "loss": 0.0977,
+      "step": 2360
+    },
+    {
+      "epoch": 0.6415809420682187,
+      "grad_norm": 0.49609375,
+      "learning_rate": 1.9221835075493612e-05,
+      "loss": 0.108,
+      "step": 2370
+    },
+    {
+      "epoch": 0.6442880346507851,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.907665505226481e-05,
+      "loss": 0.1267,
+      "step": 2380
+    },
+    {
+      "epoch": 0.6469951272333514,
+      "grad_norm": 0.5234375,
+      "learning_rate": 1.8931475029036006e-05,
+      "loss": 0.1216,
+      "step": 2390
+    },
+    {
+      "epoch": 0.6497022198159177,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.8786295005807203e-05,
+      "loss": 0.1038,
+      "step": 2400
+    },
+    {
+      "epoch": 0.6497022198159177,
+      "eval_loss": 0.09887947887182236,
+      "eval_runtime": 62.4522,
+      "eval_samples_per_second": 8.006,
+      "eval_steps_per_second": 0.512,
+      "step": 2400
+    },
+    {
+      "epoch": 0.652409312398484,
+      "grad_norm": 2.453125,
+      "learning_rate": 1.8641114982578397e-05,
+      "loss": 0.1222,
+      "step": 2410
+    },
+    {
+      "epoch": 0.6551164049810504,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.8495934959349594e-05,
+      "loss": 0.1178,
+      "step": 2420
+    },
+    {
+      "epoch": 0.6578234975636167,
+      "grad_norm": 0.328125,
+      "learning_rate": 1.835075493612079e-05,
+      "loss": 0.0962,
+      "step": 2430
+    },
+    {
+      "epoch": 0.660530590146183,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.8205574912891988e-05,
+      "loss": 0.0779,
+      "step": 2440
+    },
+    {
+      "epoch": 0.6632376827287493,
+      "grad_norm": 1.0234375,
+      "learning_rate": 1.806039488966318e-05,
+      "loss": 0.1117,
+      "step": 2450
+    },
+    {
+      "epoch": 0.6659447753113157,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.791521486643438e-05,
+      "loss": 0.0766,
+      "step": 2460
+    },
+    {
+      "epoch": 0.668651867893882,
+      "grad_norm": 0.98828125,
+      "learning_rate": 1.7770034843205575e-05,
+      "loss": 0.101,
+      "step": 2470
+    },
+    {
+      "epoch": 0.6713589604764483,
+      "grad_norm": 0.337890625,
+      "learning_rate": 1.7624854819976772e-05,
+      "loss": 0.0907,
+      "step": 2480
+    },
+    {
+      "epoch": 0.6740660530590146,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.747967479674797e-05,
+      "loss": 0.1209,
+      "step": 2490
+    },
+    {
+      "epoch": 0.676773145641581,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.7334494773519163e-05,
+      "loss": 0.0868,
+      "step": 2500
+    },
+    {
+      "epoch": 0.676773145641581,
+      "eval_loss": 0.09873838722705841,
+      "eval_runtime": 62.714,
+      "eval_samples_per_second": 7.973,
+      "eval_steps_per_second": 0.51,
+      "step": 2500
+    },
+    {
+      "epoch": 0.6794802382241473,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.718931475029036e-05,
+      "loss": 0.1117,
+      "step": 2510
+    },
+    {
+      "epoch": 0.6821873308067136,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.7044134727061557e-05,
+      "loss": 0.1111,
+      "step": 2520
+    },
+    {
+      "epoch": 0.6848944233892799,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.6898954703832754e-05,
+      "loss": 0.1119,
+      "step": 2530
+    },
+    {
+      "epoch": 0.6876015159718463,
+      "grad_norm": 0.2734375,
+      "learning_rate": 1.675377468060395e-05,
+      "loss": 0.0696,
+      "step": 2540
+    },
+    {
+      "epoch": 0.6903086085544126,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.6608594657375144e-05,
+      "loss": 0.1157,
+      "step": 2550
+    },
+    {
+      "epoch": 0.6930157011369789,
+      "grad_norm": 0.59375,
+      "learning_rate": 1.6463414634146345e-05,
+      "loss": 0.0811,
+      "step": 2560
+    },
+    {
+      "epoch": 0.6957227937195453,
+      "grad_norm": 0.384765625,
+      "learning_rate": 1.6318234610917538e-05,
+      "loss": 0.0931,
+      "step": 2570
+    },
+    {
+      "epoch": 0.6984298863021116,
+      "grad_norm": 1.8125,
+      "learning_rate": 1.6173054587688735e-05,
+      "loss": 0.0722,
+      "step": 2580
+    },
+    {
+      "epoch": 0.7011369788846779,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.602787456445993e-05,
+      "loss": 0.0827,
+      "step": 2590
+    },
+    {
+      "epoch": 0.7038440714672441,
+      "grad_norm": 1.1015625,
+      "learning_rate": 1.5882694541231126e-05,
+      "loss": 0.0838,
+      "step": 2600
+    },
+    {
+      "epoch": 0.7038440714672441,
+      "eval_loss": 0.10134698450565338,
+      "eval_runtime": 62.3738,
+      "eval_samples_per_second": 8.016,
+      "eval_steps_per_second": 0.513,
+      "step": 2600
+    },
+    {
+      "epoch": 0.7065511640498106,
+      "grad_norm": 0.2578125,
+      "learning_rate": 1.5737514518002326e-05,
+      "loss": 0.0966,
+      "step": 2610
+    },
+    {
+      "epoch": 0.7092582566323768,
+      "grad_norm": 1.3046875,
+      "learning_rate": 1.559233449477352e-05,
+      "loss": 0.0998,
+      "step": 2620
+    },
+    {
+      "epoch": 0.7119653492149431,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.5447154471544717e-05,
+      "loss": 0.1566,
+      "step": 2630
+    },
+    {
+      "epoch": 0.7146724417975094,
+      "grad_norm": 1.90625,
+      "learning_rate": 1.530197444831591e-05,
+      "loss": 0.1078,
+      "step": 2640
+    },
+    {
+      "epoch": 0.7173795343800758,
+      "grad_norm": 1.703125,
+      "learning_rate": 1.5156794425087109e-05,
+      "loss": 0.1152,
+      "step": 2650
+    },
+    {
+      "epoch": 0.7200866269626421,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.5011614401858304e-05,
+      "loss": 0.0969,
+      "step": 2660
+    },
+    {
+      "epoch": 0.7227937195452084,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.4866434378629501e-05,
+      "loss": 0.111,
+      "step": 2670
+    },
+    {
+      "epoch": 0.7255008121277747,
+      "grad_norm": 3.765625,
+      "learning_rate": 1.4721254355400698e-05,
+      "loss": 0.1255,
+      "step": 2680
+    },
+    {
+      "epoch": 0.7282079047103411,
+      "grad_norm": 2.515625,
+      "learning_rate": 1.4576074332171893e-05,
+      "loss": 0.1017,
+      "step": 2690
+    },
+    {
+      "epoch": 0.7309149972929074,
+      "grad_norm": 0.38671875,
+      "learning_rate": 1.443089430894309e-05,
+      "loss": 0.0767,
+      "step": 2700
+    },
+    {
+      "epoch": 0.7309149972929074,
+      "eval_loss": 0.0986585021018982,
+      "eval_runtime": 62.4165,
+      "eval_samples_per_second": 8.011,
+      "eval_steps_per_second": 0.513,
+      "step": 2700
+    },
+    {
+      "epoch": 0.7336220898754737,
+      "grad_norm": 1.609375,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 0.1086,
+      "step": 2710
+    },
+    {
+      "epoch": 0.73632918245804,
+      "grad_norm": 0.7734375,
+      "learning_rate": 1.4140534262485482e-05,
+      "loss": 0.1022,
+      "step": 2720
+    },
+    {
+      "epoch": 0.7390362750406064,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.3995354239256678e-05,
+      "loss": 0.0834,
+      "step": 2730
+    },
+    {
+      "epoch": 0.7417433676231727,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.3850174216027875e-05,
+      "loss": 0.1013,
+      "step": 2740
+    },
+    {
+      "epoch": 0.744450460205739,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.3704994192799073e-05,
+      "loss": 0.0849,
+      "step": 2750
+    },
+    {
+      "epoch": 0.7471575527883053,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.3559814169570267e-05,
+      "loss": 0.1335,
+      "step": 2760
+    },
+    {
+      "epoch": 0.7498646453708717,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.3414634146341466e-05,
+      "loss": 0.1181,
+      "step": 2770
+    },
+    {
+      "epoch": 0.752571737953438,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.3269454123112659e-05,
+      "loss": 0.0826,
+      "step": 2780
+    },
+    {
+      "epoch": 0.7552788305360043,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.3124274099883858e-05,
+      "loss": 0.0965,
+      "step": 2790
+    },
+    {
+      "epoch": 0.7579859231185706,
+      "grad_norm": 3.171875,
+      "learning_rate": 1.2979094076655051e-05,
+      "loss": 0.1554,
+      "step": 2800
+    },
+    {
+      "epoch": 0.7579859231185706,
+      "eval_loss": 0.09837577491998672,
+      "eval_runtime": 62.557,
+      "eval_samples_per_second": 7.993,
+      "eval_steps_per_second": 0.512,
+      "step": 2800
+    },
+    {
+      "epoch": 0.760693015701137,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.283391405342625e-05,
+      "loss": 0.0893,
+      "step": 2810
+    },
+    {
+      "epoch": 0.7634001082837033,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.2688734030197447e-05,
+      "loss": 0.081,
+      "step": 2820
+    },
+    {
+      "epoch": 0.7661072008662696,
+      "grad_norm": 1.25,
+      "learning_rate": 1.2543554006968642e-05,
+      "loss": 0.0792,
+      "step": 2830
+    },
+    {
+      "epoch": 0.7688142934488359,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.2398373983739837e-05,
+      "loss": 0.107,
+      "step": 2840
+    },
+    {
+      "epoch": 0.7715213860314023,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.2253193960511034e-05,
+      "loss": 0.1189,
+      "step": 2850
+    },
+    {
+      "epoch": 0.7742284786139686,
+      "grad_norm": 1.3046875,
+      "learning_rate": 1.2108013937282231e-05,
+      "loss": 0.0842,
+      "step": 2860
+    },
+    {
+      "epoch": 0.7769355711965349,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.1962833914053427e-05,
+      "loss": 0.0787,
+      "step": 2870
+    },
+    {
+      "epoch": 0.7796426637791013,
+      "grad_norm": 1.0234375,
+      "learning_rate": 1.1817653890824624e-05,
+      "loss": 0.1177,
+      "step": 2880
+    },
+    {
+      "epoch": 0.7823497563616676,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.1672473867595819e-05,
+      "loss": 0.1093,
+      "step": 2890
+    },
+    {
+      "epoch": 0.7850568489442339,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.1527293844367016e-05,
+      "loss": 0.0763,
+      "step": 2900
+    },
+    {
+      "epoch": 0.7850568489442339,
+      "eval_loss": 0.0983569398522377,
+      "eval_runtime": 61.958,
+      "eval_samples_per_second": 8.07,
+      "eval_steps_per_second": 0.516,
+      "step": 2900
+    },
+    {
+      "epoch": 0.7877639415268002,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.1382113821138211e-05,
+      "loss": 0.0737,
+      "step": 2910
+    },
+    {
+      "epoch": 0.7904710341093666,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.1236933797909408e-05,
+      "loss": 0.0957,
+      "step": 2920
+    },
+    {
+      "epoch": 0.7931781266919329,
+      "grad_norm": 1.296875,
+      "learning_rate": 1.1091753774680605e-05,
+      "loss": 0.0986,
+      "step": 2930
+    },
+    {
+      "epoch": 0.7958852192744992,
+      "grad_norm": 1.015625,
+      "learning_rate": 1.09465737514518e-05,
+      "loss": 0.099,
+      "step": 2940
+    },
+    {
+      "epoch": 0.7985923118570655,
+      "grad_norm": 1.859375,
+      "learning_rate": 1.0801393728222997e-05,
+      "loss": 0.1174,
+      "step": 2950
+    },
+    {
+      "epoch": 0.8012994044396319,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.0656213704994193e-05,
+      "loss": 0.1235,
+      "step": 2960
+    },
+    {
+      "epoch": 0.8040064970221982,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.051103368176539e-05,
+      "loss": 0.0913,
+      "step": 2970
+    },
+    {
+      "epoch": 0.8067135896047645,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.0365853658536585e-05,
+      "loss": 0.1006,
+      "step": 2980
+    },
+    {
+      "epoch": 0.8094206821873308,
+      "grad_norm": 0.59765625,
+      "learning_rate": 1.0220673635307783e-05,
+      "loss": 0.0819,
+      "step": 2990
+    },
+    {
+      "epoch": 0.8121277747698972,
+      "grad_norm": 1.6796875,
+      "learning_rate": 1.0075493612078979e-05,
+      "loss": 0.0886,
+      "step": 3000
+    },
+    {
+      "epoch": 0.8121277747698972,
+      "eval_loss": 0.09685203433036804,
+      "eval_runtime": 62.2767,
+      "eval_samples_per_second": 8.029,
+      "eval_steps_per_second": 0.514,
+      "step": 3000
+    },
+    {
+      "epoch": 0.8148348673524635,
+      "grad_norm": 0.546875,
+      "learning_rate": 9.930313588850176e-06,
+      "loss": 0.1091,
+      "step": 3010
+    },
+    {
+      "epoch": 0.8175419599350298,
+      "grad_norm": 1.9296875,
+      "learning_rate": 9.785133565621371e-06,
+      "loss": 0.0931,
+      "step": 3020
+    },
+    {
+      "epoch": 0.8202490525175961,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 9.639953542392568e-06,
+      "loss": 0.1179,
+      "step": 3030
+    },
+    {
+      "epoch": 0.8229561451001625,
+      "grad_norm": 2.046875,
+      "learning_rate": 9.494773519163763e-06,
+      "loss": 0.0984,
+      "step": 3040
+    },
+    {
+      "epoch": 0.8256632376827288,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 9.34959349593496e-06,
+      "loss": 0.0745,
+      "step": 3050
+    },
+    {
+      "epoch": 0.828370330265295,
+      "grad_norm": 1.6640625,
+      "learning_rate": 9.204413472706155e-06,
+      "loss": 0.1009,
+      "step": 3060
+    },
+    {
+      "epoch": 0.8310774228478613,
+      "grad_norm": 1.5859375,
+      "learning_rate": 9.059233449477352e-06,
+      "loss": 0.0984,
+      "step": 3070
+    },
+    {
+      "epoch": 0.8337845154304278,
+      "grad_norm": 1.09375,
+      "learning_rate": 8.91405342624855e-06,
+      "loss": 0.0975,
+      "step": 3080
+    },
+    {
+      "epoch": 0.836491608012994,
+      "grad_norm": 0.640625,
+      "learning_rate": 8.768873403019745e-06,
+      "loss": 0.0778,
+      "step": 3090
+    },
+    {
+      "epoch": 0.8391987005955603,
+      "grad_norm": 1.6484375,
+      "learning_rate": 8.623693379790942e-06,
+      "loss": 0.1036,
+      "step": 3100
+    },
+    {
+      "epoch": 0.8391987005955603,
+      "eval_loss": 0.0955013856291771,
+      "eval_runtime": 63.1472,
+      "eval_samples_per_second": 7.918,
+      "eval_steps_per_second": 0.507,
+      "step": 3100
+    },
+    {
+      "epoch": 0.8419057931781266,
+      "grad_norm": 1.453125,
+      "learning_rate": 8.478513356562137e-06,
+      "loss": 0.1405,
+      "step": 3110
+    },
+    {
+      "epoch": 0.844612885760693,
+      "grad_norm": 3.28125,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.1595,
+      "step": 3120
+    },
+    {
+      "epoch": 0.8473199783432593,
+      "grad_norm": 0.8359375,
+      "learning_rate": 8.188153310104529e-06,
+      "loss": 0.1146,
+      "step": 3130
+    },
+    {
+      "epoch": 0.8500270709258256,
+      "grad_norm": 1.71875,
+      "learning_rate": 8.042973286875728e-06,
+      "loss": 0.1328,
+      "step": 3140
+    },
+    {
+      "epoch": 0.852734163508392,
+      "grad_norm": 2.453125,
+      "learning_rate": 7.897793263646923e-06,
+      "loss": 0.119,
+      "step": 3150
+    },
+    {
+      "epoch": 0.8554412560909583,
+      "grad_norm": 1.9921875,
+      "learning_rate": 7.752613240418118e-06,
+      "loss": 0.1161,
+      "step": 3160
+    },
+    {
+      "epoch": 0.8581483486735246,
+      "grad_norm": 4.9375,
+      "learning_rate": 7.607433217189315e-06,
+      "loss": 0.1501,
+      "step": 3170
+    },
+    {
+      "epoch": 0.8608554412560909,
+      "grad_norm": 2.84375,
+      "learning_rate": 7.462253193960511e-06,
+      "loss": 0.1128,
+      "step": 3180
+    },
+    {
+      "epoch": 0.8635625338386573,
+      "grad_norm": 2.015625,
+      "learning_rate": 7.317073170731707e-06,
+      "loss": 0.1266,
+      "step": 3190
+    },
+    {
+      "epoch": 0.8662696264212236,
+      "grad_norm": 1.1328125,
+      "learning_rate": 7.1718931475029035e-06,
+      "loss": 0.092,
+      "step": 3200
+    },
+    {
+      "epoch": 0.8662696264212236,
+      "eval_loss": 0.09543344378471375,
+      "eval_runtime": 63.0583,
+      "eval_samples_per_second": 7.929,
+      "eval_steps_per_second": 0.507,
+      "step": 3200
+    },
+    {
+      "epoch": 0.8689767190037899,
+      "grad_norm": 1.7890625,
+      "learning_rate": 7.0267131242741005e-06,
+      "loss": 0.0913,
+      "step": 3210
+    },
+    {
+      "epoch": 0.8716838115863562,
+      "grad_norm": 1.0078125,
+      "learning_rate": 6.8815331010452966e-06,
+      "loss": 0.1009,
+      "step": 3220
+    },
+    {
+      "epoch": 0.8743909041689226,
+      "grad_norm": 0.953125,
+      "learning_rate": 6.736353077816493e-06,
+      "loss": 0.0858,
+      "step": 3230
+    },
+    {
+      "epoch": 0.8770979967514889,
+      "grad_norm": 0.98828125,
+      "learning_rate": 6.591173054587689e-06,
+      "loss": 0.1458,
+      "step": 3240
+    },
+    {
+      "epoch": 0.8798050893340552,
+      "grad_norm": 1.265625,
+      "learning_rate": 6.445993031358885e-06,
+      "loss": 0.1029,
+      "step": 3250
+    },
+    {
+      "epoch": 0.8825121819166215,
+      "grad_norm": 1.2890625,
+      "learning_rate": 6.300813008130081e-06,
+      "loss": 0.1067,
+      "step": 3260
+    },
+    {
+      "epoch": 0.8852192744991879,
+      "grad_norm": 0.77734375,
+      "learning_rate": 6.155632984901278e-06,
+      "loss": 0.0908,
+      "step": 3270
+    },
+    {
+      "epoch": 0.8879263670817542,
+      "grad_norm": 1.5859375,
+      "learning_rate": 6.010452961672474e-06,
+      "loss": 0.1325,
+      "step": 3280
+    },
+    {
+      "epoch": 0.8906334596643205,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.86527293844367e-06,
+      "loss": 0.0917,
+      "step": 3290
+    },
+    {
+      "epoch": 0.8933405522468868,
+      "grad_norm": 1.625,
+      "learning_rate": 5.720092915214867e-06,
+      "loss": 0.1192,
+      "step": 3300
+    },
+    {
+      "epoch": 0.8933405522468868,
+      "eval_loss": 0.09418918937444687,
+      "eval_runtime": 62.0629,
+      "eval_samples_per_second": 8.056,
+      "eval_steps_per_second": 0.516,
+      "step": 3300
+    },
+    {
+      "epoch": 0.8960476448294532,
+      "grad_norm": 3.375,
+      "learning_rate": 5.574912891986063e-06,
+      "loss": 0.0943,
+      "step": 3310
+    },
+    {
+      "epoch": 0.8987547374120195,
+      "grad_norm": 2.15625,
+      "learning_rate": 5.429732868757259e-06,
+      "loss": 0.061,
+      "step": 3320
+    },
+    {
+      "epoch": 0.9014618299945858,
+      "grad_norm": 2.1875,
+      "learning_rate": 5.2845528455284555e-06,
+      "loss": 0.0946,
+      "step": 3330
+    },
+    {
+      "epoch": 0.9041689225771521,
+      "grad_norm": 0.412109375,
+      "learning_rate": 5.139372822299652e-06,
+      "loss": 0.0628,
+      "step": 3340
+    },
+    {
+      "epoch": 0.9068760151597185,
+      "grad_norm": 0.6796875,
+      "learning_rate": 4.994192799070848e-06,
+      "loss": 0.1023,
+      "step": 3350
+    },
+    {
+      "epoch": 0.9095831077422848,
+      "grad_norm": 1.140625,
+      "learning_rate": 4.849012775842044e-06,
+      "loss": 0.106,
+      "step": 3360
+    },
+    {
+      "epoch": 0.9122902003248511,
+      "grad_norm": 1.671875,
+      "learning_rate": 4.703832752613241e-06,
+      "loss": 0.1219,
+      "step": 3370
+    },
+    {
+      "epoch": 0.9149972929074174,
+      "grad_norm": 0.208984375,
+      "learning_rate": 4.558652729384437e-06,
+      "loss": 0.1005,
+      "step": 3380
+    },
+    {
+      "epoch": 0.9177043854899838,
+      "grad_norm": 0.6328125,
+      "learning_rate": 4.413472706155633e-06,
+      "loss": 0.068,
+      "step": 3390
+    },
+    {
+      "epoch": 0.9204114780725501,
+      "grad_norm": 1.6796875,
+      "learning_rate": 4.26829268292683e-06,
+      "loss": 0.0998,
+      "step": 3400
+    },
+    {
+      "epoch": 0.9204114780725501,
+      "eval_loss": 0.09559421986341476,
+      "eval_runtime": 62.484,
+      "eval_samples_per_second": 8.002,
+      "eval_steps_per_second": 0.512,
+      "step": 3400
+    },
+    {
+      "epoch": 0.9231185706551164,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.123112659698026e-06,
+      "loss": 0.1139,
+      "step": 3410
+    },
+    {
+      "epoch": 0.9258256632376828,
+      "grad_norm": 0.640625,
+      "learning_rate": 3.977932636469222e-06,
+      "loss": 0.1073,
+      "step": 3420
+    },
+    {
+      "epoch": 0.9285327558202491,
+      "grad_norm": 1.625,
+      "learning_rate": 3.832752613240418e-06,
+      "loss": 0.1067,
+      "step": 3430
+    },
+    {
+      "epoch": 0.9312398484028154,
+      "grad_norm": 1.5703125,
+      "learning_rate": 3.687572590011615e-06,
+      "loss": 0.0804,
+      "step": 3440
+    },
+    {
+      "epoch": 0.9339469409853817,
+      "grad_norm": 2.046875,
+      "learning_rate": 3.542392566782811e-06,
+      "loss": 0.1129,
+      "step": 3450
+    },
+    {
+      "epoch": 0.9366540335679481,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.397212543554007e-06,
+      "loss": 0.1391,
+      "step": 3460
+    },
+    {
+      "epoch": 0.9393611261505144,
+      "grad_norm": 1.375,
+      "learning_rate": 3.2520325203252037e-06,
+      "loss": 0.0884,
+      "step": 3470
+    },
+    {
+      "epoch": 0.9420682187330807,
+      "grad_norm": 2.328125,
+      "learning_rate": 3.1068524970963998e-06,
+      "loss": 0.1047,
+      "step": 3480
+    },
+    {
+      "epoch": 0.944775311315647,
+      "grad_norm": 1.171875,
+      "learning_rate": 2.961672473867596e-06,
+      "loss": 0.1278,
+      "step": 3490
+    },
+    {
+      "epoch": 0.9474824038982134,
+      "grad_norm": 0.453125,
+      "learning_rate": 2.8164924506387924e-06,
+      "loss": 0.0854,
+      "step": 3500
+    },
+    {
+      "epoch": 0.9474824038982134,
+      "eval_loss": 0.0960383489727974,
+      "eval_runtime": 62.7188,
+      "eval_samples_per_second": 7.972,
+      "eval_steps_per_second": 0.51,
+      "step": 3500
+    },
+    {
+      "epoch": 0.9501894964807797,
+      "grad_norm": 1.25,
+      "learning_rate": 2.6713124274099885e-06,
+      "loss": 0.071,
+      "step": 3510
+    },
+    {
+      "epoch": 0.952896589063346,
+      "grad_norm": 0.7421875,
+      "learning_rate": 2.5261324041811846e-06,
+      "loss": 0.1106,
+      "step": 3520
+    },
+    {
+      "epoch": 0.9556036816459123,
+      "grad_norm": 1.1484375,
+      "learning_rate": 2.3809523809523808e-06,
+      "loss": 0.0926,
+      "step": 3530
+    },
+    {
+      "epoch": 0.9583107742284787,
+      "grad_norm": 0.53125,
+      "learning_rate": 2.2357723577235773e-06,
+      "loss": 0.093,
+      "step": 3540
+    },
+    {
+      "epoch": 0.961017866811045,
+      "grad_norm": 0.52734375,
+      "learning_rate": 2.090592334494774e-06,
+      "loss": 0.0976,
+      "step": 3550
+    },
+    {
+      "epoch": 0.9637249593936112,
+      "grad_norm": 0.74609375,
+      "learning_rate": 1.94541231126597e-06,
+      "loss": 0.1135,
+      "step": 3560
+    },
+    {
+      "epoch": 0.9664320519761775,
+      "grad_norm": 2.25,
+      "learning_rate": 1.8002322880371663e-06,
+      "loss": 0.1142,
+      "step": 3570
+    },
+    {
+      "epoch": 0.969139144558744,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.6550522648083624e-06,
+      "loss": 0.0899,
+      "step": 3580
+    },
+    {
+      "epoch": 0.9718462371413102,
+      "grad_norm": 0.5,
+      "learning_rate": 1.5098722415795587e-06,
+      "loss": 0.1226,
+      "step": 3590
+    },
+    {
+      "epoch": 0.9745533297238765,
+      "grad_norm": 0.74609375,
+      "learning_rate": 1.364692218350755e-06,
+      "loss": 0.1007,
+      "step": 3600
+    },
+    {
+      "epoch": 0.9745533297238765,
+      "eval_loss": 0.0948868989944458,
+      "eval_runtime": 62.1229,
+      "eval_samples_per_second": 8.049,
+      "eval_steps_per_second": 0.515,
+      "step": 3600
+    },
+    {
+      "epoch": 0.9772604223064428,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.2195121951219514e-06,
+      "loss": 0.1054,
+      "step": 3610
+    },
+    {
+      "epoch": 0.9799675148890092,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.0743321718931475e-06,
+      "loss": 0.0944,
+      "step": 3620
+    },
+    {
+      "epoch": 0.9826746074715755,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.291521486643438e-07,
+      "loss": 0.1047,
+      "step": 3630
+    },
+    {
+      "epoch": 0.9853817000541418,
+      "grad_norm": 0.76953125,
+      "learning_rate": 7.839721254355401e-07,
+      "loss": 0.1071,
+      "step": 3640
+    },
+    {
+      "epoch": 0.9880887926367081,
+      "grad_norm": 0.421875,
+      "learning_rate": 6.387921022067365e-07,
+      "loss": 0.1229,
+      "step": 3650
+    },
+    {
+      "epoch": 0.9907958852192745,
+      "grad_norm": 2.53125,
+      "learning_rate": 4.936120789779327e-07,
+      "loss": 0.1092,
+      "step": 3660
+    },
+    {
+      "epoch": 0.9935029778018408,
+      "grad_norm": 1.1796875,
+      "learning_rate": 3.4843205574912896e-07,
+      "loss": 0.0988,
+      "step": 3670
+    },
+    {
+      "epoch": 0.9962100703844071,
+      "grad_norm": 2.703125,
+      "learning_rate": 2.0325203252032523e-07,
+      "loss": 0.1089,
+      "step": 3680
+    },
+    {
+      "epoch": 0.9989171629669734,
+      "grad_norm": 0.5859375,
+      "learning_rate": 5.807200929152149e-08,
+      "loss": 0.0807,
+      "step": 3690
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 3694,
+  "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": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.5935579955478095e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-3694/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:325c2c05de97b3fd6f337bae3f7de80e88004be032aa944334587b15bf5338b9
+size 5176

git_hash.txt

@@ -0,0 +1 @@
+fac8c95b525c8a0f77268b5488c8d1af31c6784b

preprocessor_config.json

@@ -0,0 +1,29 @@
+{
+  "do_convert_rgb": true,
+  "do_normalize": true,
+  "do_rescale": true,
+  "do_resize": true,
+  "image_mean": [
+    0.48145466,
+    0.4578275,
+    0.40821073
+  ],
+  "image_processor_type": "Qwen2VLImageProcessor",
+  "image_std": [
+    0.26862954,
+    0.26130258,
+    0.27577711
+  ],
+  "max_pixels": 12845056,
+  "merge_size": 2,
+  "min_pixels": 3136,
+  "patch_size": 14,
+  "processor_class": "ColQwen2Processor",
+  "resample": 3,
+  "rescale_factor": 0.00392156862745098,
+  "size": {
+    "max_pixels": 12845056,
+    "min_pixels": 3136
+  },
+  "temporal_patch_size": 2
+}

results.json

@@ -0,0 +1 @@
+{"validation_set": {"ndcg_at_1": 0.812, "ndcg_at_3": 0.86786, "ndcg_at_5": 0.87526, "ndcg_at_10": 0.88112, "ndcg_at_20": 0.88502, "ndcg_at_50": 0.88796, "ndcg_at_100": 0.89119, "map_at_1": 0.812, "map_at_3": 0.854, "map_at_5": 0.8581, "map_at_10": 0.86054, "map_at_20": 0.86153, "map_at_50": 0.86207, "map_at_100": 0.86235, "recall_at_1": 0.812, "recall_at_3": 0.908, "recall_at_5": 0.926, "recall_at_10": 0.944, "recall_at_20": 0.96, "recall_at_50": 0.974, "recall_at_100": 0.994, "precision_at_1": 0.812, "precision_at_3": 0.30267, "precision_at_5": 0.1852, "precision_at_10": 0.0944, "precision_at_20": 0.048, "precision_at_50": 0.01948, "precision_at_100": 0.00994, "mrr_at_1": 0.814, "mrr_at_3": 0.856, "mrr_at_5": 0.8601, "mrr_at_10": 0.862818253968254, "mrr_at_20": 0.8641760273610118, "mrr_at_50": 0.8646866894307063, "mrr_at_100": 0.8648347594798413, "naucs_at_1_max": 0.24277355735658093, "naucs_at_1_std": 0.12502909621854932, "naucs_at_1_diff1": 0.923508686543792, "naucs_at_3_max": 0.2787520805423625, "naucs_at_3_std": 0.4184427394146067, "naucs_at_3_diff1": 0.9243393009377648, "naucs_at_5_max": 0.3077825724884538, "naucs_at_5_std": 0.5717313952608055, "naucs_at_5_diff1": 0.9271330153683086, "naucs_at_10_max": 0.39259036948112686, "naucs_at_10_std": 0.8404861944777965, "naucs_at_10_diff1": 0.9521475256769385, "naucs_at_20_max": 0.36094771241829526, "naucs_at_20_std": 0.8856442577030798, "naucs_at_20_diff1": 0.9730392156862733, "naucs_at_50_max": 0.3941679235796853, "naucs_at_50_std": 0.9786324786324722, "naucs_at_50_diff1": 0.9685771744595257, "naucs_at_100_max": 0.14285714285712342, "naucs_at_100_std": 1.0, "naucs_at_100_diff1": 1.0}, "syntheticDocQA_energy": {"ndcg_at_1": 0.91, "ndcg_at_3": 0.94393, "ndcg_at_5": 0.94823, "ndcg_at_10": 0.95124, "ndcg_at_20": 0.95124, "ndcg_at_50": 0.95328, "ndcg_at_100": 0.95328, "map_at_1": 0.91, "map_at_3": 0.935, "map_at_5": 0.9375, "map_at_10": 0.93861, "map_at_20": 0.93861, "map_at_50": 0.93896, "map_at_100": 0.93896, "recall_at_1": 0.91, "recall_at_3": 0.97, "recall_at_5": 0.98, "recall_at_10": 0.99, "recall_at_20": 0.99, "recall_at_50": 1.0, "recall_at_100": 1.0, "precision_at_1": 0.91, "precision_at_3": 0.32333, "precision_at_5": 0.196, "precision_at_10": 0.099, "precision_at_20": 0.0495, "precision_at_50": 0.02, "precision_at_100": 0.01, "mrr_at_1": 0.91, "mrr_at_3": 0.9383333333333332, "mrr_at_5": 0.9383333333333332, "mrr_at_10": 0.94, "mrr_at_20": 0.94, "mrr_at_50": 0.9403448275862069, "mrr_at_100": 0.9403448275862069, "naucs_at_1_max": 0.6073762838468708, "naucs_at_1_std": -0.36616868969810257, "naucs_at_1_diff1": 0.954611474219316, "naucs_at_3_max": 0.6498599439775861, "naucs_at_3_std": -0.5961718020541529, "naucs_at_3_diff1": 0.9564270152505466, "naucs_at_5_max": 0.7957516339869297, "naucs_at_5_std": -0.024276377217554025, "naucs_at_5_diff1": 0.9346405228758136, "naucs_at_10_max": 0.8692810457516413, "naucs_at_10_std": 0.12278244631185926, "naucs_at_10_diff1": 0.8692810457516413, "naucs_at_20_max": 0.8692810457516413, "naucs_at_20_std": 0.12278244631185926, "naucs_at_20_diff1": 0.8692810457516413, "naucs_at_50_max": NaN, "naucs_at_50_std": NaN, "naucs_at_50_diff1": NaN, "naucs_at_100_max": NaN, "naucs_at_100_std": NaN, "naucs_at_100_diff1": NaN}, "syntheticDocQA_healthcare_industry": {"ndcg_at_1": 0.95, "ndcg_at_3": 0.97262, "ndcg_at_5": 0.97693, "ndcg_at_10": 0.97693, "ndcg_at_20": 0.97693, "ndcg_at_50": 0.97693, "ndcg_at_100": 0.97693, "map_at_1": 0.95, "map_at_3": 0.96667, "map_at_5": 0.96917, "map_at_10": 0.96917, "map_at_20": 0.96917, "map_at_50": 0.96917, "map_at_100": 0.96917, "recall_at_1": 0.95, "recall_at_3": 0.99, "recall_at_5": 1.0, "recall_at_10": 1.0, "recall_at_20": 1.0, "recall_at_50": 1.0, "recall_at_100": 1.0, "precision_at_1": 0.95, "precision_at_3": 0.33, "precision_at_5": 0.2, "precision_at_10": 0.1, "precision_at_20": 0.05, "precision_at_50": 0.02, "precision_at_100": 0.01, "mrr_at_1": 0.95, "mrr_at_3": 0.97, "mrr_at_5": 0.97, "mrr_at_10": 0.97, "mrr_at_20": 0.97, "mrr_at_50": 0.97, "mrr_at_100": 0.97, "naucs_at_1_max": 0.6321195144724556, "naucs_at_1_std": 0.24295051353874755, "naucs_at_1_diff1": 0.9738562091503253, "naucs_at_3_max": 1.0, "naucs_at_3_std": 0.7222222222222157, "naucs_at_3_diff1": 1.0, "naucs_at_5_max": 1.0, "naucs_at_5_std": 1.0, "naucs_at_5_diff1": 1.0, "naucs_at_10_max": 1.0, "naucs_at_10_std": 1.0, "naucs_at_10_diff1": 1.0, "naucs_at_20_max": 1.0, "naucs_at_20_std": 1.0, "naucs_at_20_diff1": 1.0, "naucs_at_50_max": NaN, "naucs_at_50_std": NaN, "naucs_at_50_diff1": NaN, "naucs_at_100_max": NaN, "naucs_at_100_std": NaN, "naucs_at_100_diff1": NaN}, "syntheticDocQA_artificial_intelligence_test": {"ndcg_at_1": 0.95, "ndcg_at_3": 0.97524, "ndcg_at_5": 0.97911, "ndcg_at_10": 0.97911, "ndcg_at_20": 0.97911, "ndcg_at_50": 0.97911, "ndcg_at_100": 0.97911, "map_at_1": 0.95, "map_at_3": 0.97, "map_at_5": 0.972, "map_at_10": 0.972, "map_at_20": 0.972, "map_at_50": 0.972, "map_at_100": 0.972, "recall_at_1": 0.95, "recall_at_3": 0.99, "recall_at_5": 1.0, "recall_at_10": 1.0, "recall_at_20": 1.0, "recall_at_50": 1.0, "recall_at_100": 1.0, "precision_at_1": 0.95, "precision_at_3": 0.33, "precision_at_5": 0.2, "precision_at_10": 0.1, "precision_at_20": 0.05, "precision_at_50": 0.02, "precision_at_100": 0.01, "mrr_at_1": 0.96, "mrr_at_3": 0.975, "mrr_at_5": 0.977, "mrr_at_10": 0.977, "mrr_at_20": 0.977, "mrr_at_50": 0.977, "mrr_at_100": 0.977, "naucs_at_1_max": 0.6671335200746964, "naucs_at_1_std": 0.4161531279178331, "naucs_at_1_diff1": 1.0, "naucs_at_3_max": 0.8692810457516356, "naucs_at_3_std": -0.1713352007469878, "naucs_at_3_diff1": 1.0, "naucs_at_5_max": 1.0, "naucs_at_5_std": 1.0, "naucs_at_5_diff1": 1.0, "naucs_at_10_max": 1.0, "naucs_at_10_std": 1.0, "naucs_at_10_diff1": 1.0, "naucs_at_20_max": 1.0, "naucs_at_20_std": 1.0, "naucs_at_20_diff1": 1.0, "naucs_at_50_max": NaN, "naucs_at_50_std": NaN, "naucs_at_50_diff1": NaN, "naucs_at_100_max": NaN, "naucs_at_100_std": NaN, "naucs_at_100_diff1": NaN}, "syntheticDocQA_government_reports": {"ndcg_at_1": 0.84, "ndcg_at_3": 0.9144, "ndcg_at_5": 0.92302, "ndcg_at_10": 0.92302, "ndcg_at_20": 0.92851, "ndcg_at_50": 0.92851, "ndcg_at_100": 0.92851, "map_at_1": 0.84, "map_at_3": 0.89833, "map_at_5": 0.90333, "map_at_10": 0.90333, "map_at_20": 0.90508, "map_at_50": 0.90508, "map_at_100": 0.90508, "recall_at_1": 0.84, "recall_at_3": 0.96, "recall_at_5": 0.98, "recall_at_10": 0.98, "recall_at_20": 1.0, "recall_at_50": 1.0, "recall_at_100": 1.0, "precision_at_1": 0.84, "precision_at_3": 0.32, "precision_at_5": 0.196, "precision_at_10": 0.098, "precision_at_20": 0.05, "precision_at_50": 0.02, "precision_at_100": 0.01, "mrr_at_1": 0.87, "mrr_at_3": 0.9133333333333333, "mrr_at_5": 0.9183333333333333, "mrr_at_10": 0.9203333333333333, "mrr_at_20": 0.9203333333333333, "mrr_at_50": 0.9203333333333333, "mrr_at_100": 0.9203333333333333, "naucs_at_1_max": 0.43387947962690165, "naucs_at_1_std": 0.19615243004418212, "naucs_at_1_diff1": 0.913352970054001, "naucs_at_3_max": 0.8068394024276336, "naucs_at_3_std": 0.052404295051358386, "naucs_at_3_diff1": 1.0, "naucs_at_5_max": 0.6790382819794609, "naucs_at_5_std": -0.5088702147525628, "naucs_at_5_diff1": 1.0, "naucs_at_10_max": 0.6790382819794609, "naucs_at_10_std": -0.5088702147525628, "naucs_at_10_diff1": 1.0, "naucs_at_20_max": 1.0, "naucs_at_20_std": 1.0, "naucs_at_20_diff1": 1.0, "naucs_at_50_max": NaN, "naucs_at_50_std": NaN, "naucs_at_50_diff1": NaN, "naucs_at_100_max": NaN, "naucs_at_100_std": NaN, "naucs_at_100_diff1": NaN}, "infovqa_subsampled": {"ndcg_at_1": 0.844, "ndcg_at_3": 0.88081, "ndcg_at_5": 0.89036, "ndcg_at_10": 0.89818, "ndcg_at_20": 0.90126, "ndcg_at_50": 0.90426, "ndcg_at_100": 0.90521, "map_at_1": 0.844, "map_at_3": 0.87267, "map_at_5": 0.87777, "map_at_10": 0.88104, "map_at_20": 0.88191, "map_at_50": 0.88248, "map_at_100": 0.88256, "recall_at_1": 0.844, "recall_at_3": 0.904, "recall_at_5": 0.928, "recall_at_10": 0.952, "recall_at_20": 0.964, "recall_at_50": 0.978, "recall_at_100": 0.984, "precision_at_1": 0.844, "precision_at_3": 0.30133, "precision_at_5": 0.1856, "precision_at_10": 0.0952, "precision_at_20": 0.0482, "precision_at_50": 0.01956, "precision_at_100": 0.00984, "mrr_at_1": 0.842, "mrr_at_3": 0.8713333333333334, "mrr_at_5": 0.8765333333333333, "mrr_at_10": 0.8800960317460317, "mrr_at_20": 0.8806633541295306, "mrr_at_50": 0.8813297802013275, "mrr_at_100": 0.8814079680628487, "naucs_at_1_max": 0.46847446557311256, "naucs_at_1_std": -0.08684593936044609, "naucs_at_1_diff1": 0.9173553719008263, "naucs_at_3_max": 0.533068783068784, "naucs_at_3_std": -0.13130252100840056, "naucs_at_3_diff1": 0.8704968098350448, "naucs_at_5_max": 0.6500804025313851, "naucs_at_5_std": 0.03800964830376213, "naucs_at_5_diff1": 0.8660390081958717, "naucs_at_10_max": 0.8607415188297549, "naucs_at_10_std": 0.30569172113289783, "naucs_at_10_diff1": 0.9255952380952372, "naucs_at_20_max": 0.9105197634609353, "naucs_at_20_std": 0.516651104886399, "naucs_at_20_diff1": 0.9153179790434682, "naucs_at_50_max": 0.9643493761140892, "naucs_at_50_std": 0.6295306001188329, "naucs_at_50_diff1": 0.9881164587046902, "naucs_at_100_max": 0.9509803921568729, "naucs_at_100_std": 0.6823062558356714, "naucs_at_100_diff1": 0.9836601307189619}, "docvqa_subsampled": {"ndcg_at_1": 0.458, "ndcg_at_3": 0.5229, "ndcg_at_5": 0.5526, "ndcg_at_10": 0.57454, "ndcg_at_20": 0.58849, "ndcg_at_50": 0.60199, "ndcg_at_100": 0.61333, "map_at_1": 0.458, "map_at_3": 0.50667, "map_at_5": 0.52317, "map_at_10": 0.53219, "map_at_20": 0.5359, "map_at_50": 0.53807, "map_at_100": 0.53907, "recall_at_1": 0.458, "recall_at_3": 0.57, "recall_at_5": 0.642, "recall_at_10": 0.71, "recall_at_20": 0.766, "recall_at_50": 0.834, "recall_at_100": 0.904, "precision_at_1": 0.458, "precision_at_3": 0.19, "precision_at_5": 0.1284, "precision_at_10": 0.071, "precision_at_20": 0.0383, "precision_at_50": 0.01668, "precision_at_100": 0.00904, "mrr_at_1": 0.46, "mrr_at_3": 0.5093333333333333, "mrr_at_5": 0.5229333333333333, "mrr_at_10": 0.532697619047619, "mrr_at_20": 0.5365797975674137, "mrr_at_50": 0.5389656939675365, "mrr_at_100": 0.539727584246403, "naucs_at_1_max": 0.3528060025264396, "naucs_at_1_std": 0.3861355715872713, "naucs_at_1_diff1": 0.8678241366903641, "naucs_at_3_max": 0.2883454600249708, "naucs_at_3_std": 0.4173677186033496, "naucs_at_3_diff1": 0.7927268552403092, "naucs_at_5_max": 0.23390485715109952, "naucs_at_5_std": 0.4980142512332795, "naucs_at_5_diff1": 0.7531549204578601, "naucs_at_10_max": 0.1449585827326127, "naucs_at_10_std": 0.5381835702240428, "naucs_at_10_diff1": 0.6989678112780986, "naucs_at_20_max": 0.05591636166163949, "naucs_at_20_std": 0.6568903173965446, "naucs_at_20_diff1": 0.6609344853619661, "naucs_at_50_max": 0.07566480367377483, "naucs_at_50_std": 0.7492553961529781, "naucs_at_50_diff1": 0.6152623022794955, "naucs_at_100_max": 0.1628540305010893, "naucs_at_100_std": 0.8644763460939922, "naucs_at_100_diff1": 0.5679952536570193}, "arxivqa_subsampled": {"ndcg_at_1": 0.798, "ndcg_at_3": 0.85012, "ndcg_at_5": 0.86406, "ndcg_at_10": 0.87306, "ndcg_at_20": 0.87665, "ndcg_at_50": 0.88058, "ndcg_at_100": 0.88319, "map_at_1": 0.798, "map_at_3": 0.83767, "map_at_5": 0.84537, "map_at_10": 0.84904, "map_at_20": 0.85005, "map_at_50": 0.85066, "map_at_100": 0.8509, "recall_at_1": 0.798, "recall_at_3": 0.886, "recall_at_5": 0.92, "recall_at_10": 0.948, "recall_at_20": 0.962, "recall_at_50": 0.982, "recall_at_100": 0.998, "precision_at_1": 0.798, "precision_at_3": 0.29533, "precision_at_5": 0.184, "precision_at_10": 0.0948, "precision_at_20": 0.0481, "precision_at_50": 0.01964, "precision_at_100": 0.00998, "mrr_at_1": 0.8, "mrr_at_3": 0.8396666666666667, "mrr_at_5": 0.8475666666666667, "mrr_at_10": 0.8502825396825396, "mrr_at_20": 0.8514597252420782, "mrr_at_50": 0.8520988813643489, "mrr_at_100": 0.8523317553428907, "naucs_at_1_max": 0.6973486594596197, "naucs_at_1_std": 0.08025893817504043, "naucs_at_1_diff1": 0.9113989303434495, "naucs_at_3_max": 0.7709141458704454, "naucs_at_3_std": 0.09967203289659907, "naucs_at_3_diff1": 0.9058551284399088, "naucs_at_5_max": 0.8029411764705887, "naucs_at_5_std": 0.12350606909430441, "naucs_at_5_diff1": 0.9101890756302522, "naucs_at_10_max": 0.8147130647130642, "naucs_at_10_std": 0.1442038353803018, "naucs_at_10_diff1": 0.9097716009480712, "naucs_at_20_max": 0.8086146739397527, "naucs_at_20_std": 0.06732517568429877, "naucs_at_20_diff1": 0.9352548036758574, "naucs_at_50_max": 0.9709513435003653, "naucs_at_50_std": 0.006328457308844334, "naucs_at_50_diff1": 0.9709513435003653, "naucs_at_100_max": 0.86928104575168, "naucs_at_100_std": -1.739962651727529, "naucs_at_100_diff1": 1.0}, "tabfquad_subsampled": {"ndcg_at_1": 0.79286, "ndcg_at_3": 0.84507, "ndcg_at_5": 0.86136, "ndcg_at_10": 0.86829, "ndcg_at_20": 0.8756, "ndcg_at_50": 0.88137, "ndcg_at_100": 0.88199, "map_at_1": 0.79286, "map_at_3": 0.83333, "map_at_5": 0.84244, "map_at_10": 0.8453, "map_at_20": 0.84735, "map_at_50": 0.84833, "map_at_100": 0.84839, "recall_at_1": 0.79286, "recall_at_3": 0.87857, "recall_at_5": 0.91786, "recall_at_10": 0.93929, "recall_at_20": 0.96786, "recall_at_50": 0.99643, "recall_at_100": 1.0, "precision_at_1": 0.79286, "precision_at_3": 0.29286, "precision_at_5": 0.18357, "precision_at_10": 0.09393, "precision_at_20": 0.04839, "precision_at_50": 0.01993, "precision_at_100": 0.01, "mrr_at_1": 0.7964285714285714, "mrr_at_3": 0.8345238095238094, "mrr_at_5": 0.8436309523809524, "mrr_at_10": 0.8464923469387755, "mrr_at_20": 0.8485418927876635, "mrr_at_50": 0.849505299576952, "mrr_at_100": 0.849570234641887, "naucs_at_1_max": 0.3751019912157351, "naucs_at_1_std": 0.2023332118118846, "naucs_at_1_diff1": 0.8272572609065495, "naucs_at_3_max": 0.5210547028728848, "naucs_at_3_std": 0.30988924495418024, "naucs_at_3_diff1": 0.7500702760443004, "naucs_at_5_max": 0.4590995818617313, "naucs_at_5_std": 0.18519871716802636, "naucs_at_5_diff1": 0.6934599926927276, "naucs_at_10_max": 0.449524908002418, "naucs_at_10_std": 0.1558466523864459, "naucs_at_10_diff1": 0.6794364804745436, "naucs_at_20_max": 0.6411453470276975, "naucs_at_20_std": 0.4541964934121846, "naucs_at_20_diff1": 0.7170349621330007, "naucs_at_50_max": 0.8692810457515607, "naucs_at_50_std": 0.7222222222221809, "naucs_at_50_diff1": 0.7222222222221809, "naucs_at_100_max": 1.0, "naucs_at_100_std": 1.0, "naucs_at_100_diff1": 1.0}, "tatdqa": {"ndcg_at_1": 0.60974, "ndcg_at_3": 0.71586, "ndcg_at_5": 0.7459, "ndcg_at_10": 0.76445, "ndcg_at_20": 0.77266, "ndcg_at_50": 0.77705, "ndcg_at_100": 0.77841, "map_at_1": 0.60974, "map_at_3": 0.69032, "map_at_5": 0.70689, "map_at_10": 0.71458, "map_at_20": 0.71684, "map_at_50": 0.71759, "map_at_100": 0.71771, "recall_at_1": 0.60974, "recall_at_3": 0.78954, "recall_at_5": 0.8629, "recall_at_10": 0.92002, "recall_at_20": 0.9525, "recall_at_50": 0.97414, "recall_at_100": 0.98256, "precision_at_1": 0.60974, "precision_at_3": 0.26318, "precision_at_5": 0.17258, "precision_at_10": 0.092, "precision_at_20": 0.04762, "precision_at_50": 0.01948, "precision_at_100": 0.00983, "mrr_at_1": 0.6103427540589297, "mrr_at_3": 0.6915213469633193, "mrr_at_5": 0.7078773301262778, "mrr_at_10": 0.7151915165745593, "mrr_at_20": 0.7174885033422889, "mrr_at_50": 0.718276905274417, "mrr_at_100": 0.7184007462918804, "naucs_at_1_max": 0.1677616463547121, "naucs_at_1_std": -0.2994746722385698, "naucs_at_1_diff1": 0.8001707275183607, "naucs_at_3_max": 0.13932493130162668, "naucs_at_3_std": -0.3030066980462948, "naucs_at_3_diff1": 0.69560870623967, "naucs_at_5_max": 0.2722755429808133, "naucs_at_5_std": -0.15825137064450098, "naucs_at_5_diff1": 0.6825585257611015, "naucs_at_10_max": 0.34997449212156656, "naucs_at_10_std": 0.0543166540386977, "naucs_at_10_diff1": 0.6657264160921791, "naucs_at_20_max": 0.46471371978487297, "naucs_at_20_std": 0.327828544707501, "naucs_at_20_diff1": 0.624379378196635, "naucs_at_50_max": 0.5249536127855756, "naucs_at_50_std": 0.6604230927181238, "naucs_at_50_diff1": 0.6028649549065449, "naucs_at_100_max": 0.6651098964642715, "naucs_at_100_std": 0.7610489179350859, "naucs_at_100_diff1": 0.625785333338424}, "shift_project": {"ndcg_at_1": 0.76, "ndcg_at_3": 0.8544, "ndcg_at_5": 0.86302, "ndcg_at_10": 0.87274, "ndcg_at_20": 0.8751, "ndcg_at_50": 0.87891, "ndcg_at_100": 0.87891, "map_at_1": 0.76, "map_at_3": 0.83167, "map_at_5": 0.83667, "map_at_10": 0.84069, "map_at_20": 0.84125, "map_at_50": 0.8418, "map_at_100": 0.8418, "recall_at_1": 0.76, "recall_at_3": 0.92, "recall_at_5": 0.94, "recall_at_10": 0.97, "recall_at_20": 0.98, "recall_at_50": 1.0, "recall_at_100": 1.0, "precision_at_1": 0.76, "precision_at_3": 0.30667, "precision_at_5": 0.188, "precision_at_10": 0.097, "precision_at_20": 0.049, "precision_at_50": 0.02, "precision_at_100": 0.01, "mrr_at_1": 0.76, "mrr_at_3": 0.8383333333333333, "mrr_at_5": 0.8408333333333333, "mrr_at_10": 0.8448611111111112, "mrr_at_20": 0.8454166666666667, "mrr_at_50": 0.8459742602103204, "mrr_at_100": 0.8459742602103204, "naucs_at_1_max": 0.05836289694557389, "naucs_at_1_std": -0.13786017132473838, "naucs_at_1_diff1": 0.6319330276023185, "naucs_at_3_max": -0.012488328664798848, "naucs_at_3_std": -0.37073996265172776, "naucs_at_3_diff1": 0.46358543417367004, "naucs_at_5_max": 0.15483971366324228, "naucs_at_5_std": -0.22478991596638775, "naucs_at_5_diff1": 0.5771864301276066, "naucs_at_10_max": -0.35434173669467856, "naucs_at_10_std": -0.3963585434173691, "naucs_at_10_diff1": 0.6934329287270459, "naucs_at_20_max": -0.5929038281979383, "naucs_at_20_std": -0.5088702147525547, "naucs_at_20_diff1": 0.6790382819794609, "naucs_at_50_max": NaN, "naucs_at_50_std": NaN, "naucs_at_50_diff1": NaN, "naucs_at_100_max": NaN, "naucs_at_100_std": NaN, "naucs_at_100_diff1": NaN}}

special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_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|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:88a3a6fcb80132f76da8aa40cdc3fccd7e5d8468ef15421f5b0c2715e85217d2
+size 11420538

tokenizer_config.json

@@ -0,0 +1,145 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_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|>"
+  ],
+  "bos_token": null,
+  "chat_template": "{% set image_count = namespace(value=0) %}{% set video_count = namespace(value=0) %}{% for message in messages %}{% if loop.first and message['role'] != 'system' %}<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n{% endif %}<|im_start|>{{ message['role'] }}\n{% if message['content'] is string %}{{ message['content'] }}<|im_end|>\n{% 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 %}<|im_end|>\n{% endif %}{% endfor %}{% if add_generation_prompt %}<|im_start|>assistant\n{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "model_max_length": 32768,
+  "num_image_tokens": 2048,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "left",
+  "processor_class": "ColQwen2Processor",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

training_config.yml

@@ -0,0 +1,65 @@
+config:
+  (): colpali_engine.trainer.colmodel_training.ColModelTrainingConfig
+  output_dir: !path ../../../models/colqwen2-ba16 # ba4-highres
+  processor:
+    (): colpali_engine.utils.transformers_wrappers.AllPurposeWrapper
+    class_to_instanciate: !ext colpali_engine.models.ColQwen2Processor
+    pretrained_model_name_or_path:  "./models/colqwen2_base" # "./models/paligemma-3b-mix-448"
+    # num_image_tokens: 2048
+    # max_length: 50
+
+  model:
+    (): colpali_engine.utils.transformers_wrappers.AllPurposeWrapper
+    class_to_instanciate: !ext colpali_engine.models.ColQwen2
+    pretrained_model_name_or_path: "./models/colqwen2_base"
+    torch_dtype:  !ext torch.bfloat16
+    use_cache: false
+    attn_implementation: "flash_attention_2"
+#    device_map: "auto"
+#    quantization_config:
+#      (): transformers.BitsAndBytesConfig
+#      load_in_4bit: true
+#      bnb_4bit_quant_type: "nf4"
+#      bnb_4bit_compute_dtype:  "bfloat16"
+#      bnb_4bit_use_double_quant: true
+
+  dataset_loading_func: !ext colpali_engine.utils.dataset_transformation.load_train_set
+  eval_dataset_loader: !import ../data/test_data.yaml
+
+  # max_length: 50
+  run_eval: true
+  add_suffix: true
+  loss_func:
+    (): colpali_engine.loss.late_interaction_losses.ColbertPairwiseCELoss
+  tr_args:
+    (): transformers.training_args.TrainingArguments
+    output_dir: null
+    overwrite_output_dir: true
+    num_train_epochs: 1
+    per_device_train_batch_size: 16
+    # 6 x 8 gpus = 48 batch size
+    # gradient_accumulation_steps: 4
+    per_device_eval_batch_size: 4
+    eval_strategy: "steps"
+    # dataloader_num_workers: 8
+    # bf16: true
+    save_steps: 500
+    logging_steps: 10
+    eval_steps: 100
+    warmup_steps: 250
+    learning_rate: 5e-5
+    save_total_limit: 1
+    # resume_from_checkpoint: true
+    # optim: "paged_adamw_8bit"
+
+  peft_config:
+    (): peft.LoraConfig
+    r: 32
+    lora_alpha: 32
+    lora_dropout: 0.1
+    init_lora_weights: "gaussian"
+    bias: "none"
+    task_type: "FEATURE_EXTRACTION"
+    target_modules: '(.*(model).*(down_proj|gate_proj|up_proj|k_proj|q_proj|v_proj|o_proj).*$|.*(custom_text_proj).*$)'
+    # target_modules: '(.*(language_model).*(down_proj|gate_proj|up_proj|k_proj|q_proj|v_proj|o_proj).*$|.*(custom_text_proj).*$)'
+