Upload folder using huggingface_hub

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 sha256:0f22eb8a8f9417ff85bb0fe221ef029cfb2a4beb1e82b6d612f4bb1bcb0eb169
+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 sha256:0f22eb8a8f9417ff85bb0fe221ef029cfb2a4beb1e82b6d612f4bb1bcb0eb169
+size 74018232

checkpoint-3694/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f099c1d3208c1bb8be126ba6bdbfc0baa74138230e37ea06e375ff8a9da0f28
+size 148262384

checkpoint-3694/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9791d36a3d9c16966378862ec8776809d4b31b0d86cd32408bef75e7458717ff
+size 14244

checkpoint-3694/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f0c4c047aa3abb1bc366c2fb0e6a4fd601029da0338c37929023a3ff5f04bef3
+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": 0.65234375,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.7066,
+      "step": 10
+    },
+    {
+      "epoch": 0.005414185165132647,
+      "grad_norm": 0.4765625,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.7067,
+      "step": 20
+    },
+    {
+      "epoch": 0.008121277747698972,
+      "grad_norm": 0.71484375,
+      "learning_rate": 3e-06,
+      "loss": 0.7044,
+      "step": 30
+    },
+    {
+      "epoch": 0.010828370330265295,
+      "grad_norm": 0.345703125,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.7024,
+      "step": 40
+    },
+    {
+      "epoch": 0.01353546291283162,
+      "grad_norm": 0.3671875,
+      "learning_rate": 5e-06,
+      "loss": 0.7002,
+      "step": 50
+    },
+    {
+      "epoch": 0.016242555495397944,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 6e-06,
+      "loss": 0.6987,
+      "step": 60
+    },
+    {
+      "epoch": 0.018949648077964266,
+      "grad_norm": 0.220703125,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 0.6983,
+      "step": 70
+    },
+    {
+      "epoch": 0.02165674066053059,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.6971,
+      "step": 80
+    },
+    {
+      "epoch": 0.024363833243096916,
+      "grad_norm": 0.208984375,
+      "learning_rate": 9e-06,
+      "loss": 0.6968,
+      "step": 90
+    },
+    {
+      "epoch": 0.02707092582566324,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 1e-05,
+      "loss": 0.6964,
+      "step": 100
+    },
+    {
+      "epoch": 0.02707092582566324,
+      "eval_loss": 0.6960552334785461,
+      "eval_runtime": 96.0637,
+      "eval_samples_per_second": 5.205,
+      "eval_steps_per_second": 0.167,
+      "step": 100
+    },
+    {
+      "epoch": 0.02977801840822956,
+      "grad_norm": 0.16796875,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.6959,
+      "step": 110
+    },
+    {
+      "epoch": 0.03248511099079589,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 1.2e-05,
+      "loss": 0.6958,
+      "step": 120
+    },
+    {
+      "epoch": 0.03519220357336221,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.6954,
+      "step": 130
+    },
+    {
+      "epoch": 0.03789929615592853,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 0.6952,
+      "step": 140
+    },
+    {
+      "epoch": 0.040606388738494856,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 1.5e-05,
+      "loss": 0.6951,
+      "step": 150
+    },
+    {
+      "epoch": 0.04331348132106118,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.6951,
+      "step": 160
+    },
+    {
+      "epoch": 0.0460205739036275,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 0.6949,
+      "step": 170
+    },
+    {
+      "epoch": 0.04872766648619383,
+      "grad_norm": 0.11328125,
+      "learning_rate": 1.8e-05,
+      "loss": 0.6947,
+      "step": 180
+    },
+    {
+      "epoch": 0.051434759068760154,
+      "grad_norm": 0.17578125,
+      "learning_rate": 1.9e-05,
+      "loss": 0.6947,
+      "step": 190
+    },
+    {
+      "epoch": 0.05414185165132648,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 2e-05,
+      "loss": 0.6944,
+      "step": 200
+    },
+    {
+      "epoch": 0.05414185165132648,
+      "eval_loss": 0.6944612860679626,
+      "eval_runtime": 70.079,
+      "eval_samples_per_second": 7.135,
+      "eval_steps_per_second": 0.228,
+      "step": 200
+    },
+    {
+      "epoch": 0.0568489442338928,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 2.1e-05,
+      "loss": 0.6943,
+      "step": 210
+    },
+    {
+      "epoch": 0.05955603681645912,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 0.6943,
+      "step": 220
+    },
+    {
+      "epoch": 0.062263129399025445,
+      "grad_norm": 0.06396484375,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 0.6943,
+      "step": 230
+    },
+    {
+      "epoch": 0.06497022198159177,
+      "grad_norm": 0.123046875,
+      "learning_rate": 2.4e-05,
+      "loss": 0.6944,
+      "step": 240
+    },
+    {
+      "epoch": 0.0676773145641581,
+      "grad_norm": 0.068359375,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6943,
+      "step": 250
+    },
+    {
+      "epoch": 0.07038440714672442,
+      "grad_norm": 0.08349609375,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 0.6942,
+      "step": 260
+    },
+    {
+      "epoch": 0.07309149972929074,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 0.6941,
+      "step": 270
+    },
+    {
+      "epoch": 0.07579859231185707,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.694,
+      "step": 280
+    },
+    {
+      "epoch": 0.07850568489442339,
+      "grad_norm": 0.055419921875,
+      "learning_rate": 2.9e-05,
+      "loss": 0.6939,
+      "step": 290
+    },
+    {
+      "epoch": 0.08121277747698971,
+      "grad_norm": 0.0654296875,
+      "learning_rate": 3e-05,
+      "loss": 0.6939,
+      "step": 300
+    },
+    {
+      "epoch": 0.08121277747698971,
+      "eval_loss": 0.69379723072052,
+      "eval_runtime": 71.3483,
+      "eval_samples_per_second": 7.008,
+      "eval_steps_per_second": 0.224,
+      "step": 300
+    },
+    {
+      "epoch": 0.08391987005955603,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 3.1e-05,
+      "loss": 0.6939,
+      "step": 310
+    },
+    {
+      "epoch": 0.08662696264212236,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.6936,
+      "step": 320
+    },
+    {
+      "epoch": 0.08933405522468868,
+      "grad_norm": 0.06396484375,
+      "learning_rate": 3.3e-05,
+      "loss": 0.6931,
+      "step": 330
+    },
+    {
+      "epoch": 0.092041147807255,
+      "grad_norm": 0.09765625,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 0.6916,
+      "step": 340
+    },
+    {
+      "epoch": 0.09474824038982133,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 3.5e-05,
+      "loss": 0.6835,
+      "step": 350
+    },
+    {
+      "epoch": 0.09745533297238766,
+      "grad_norm": 0.349609375,
+      "learning_rate": 3.6e-05,
+      "loss": 0.6623,
+      "step": 360
+    },
+    {
+      "epoch": 0.10016242555495398,
+      "grad_norm": 0.375,
+      "learning_rate": 3.7e-05,
+      "loss": 0.6366,
+      "step": 370
+    },
+    {
+      "epoch": 0.10286951813752031,
+      "grad_norm": 0.345703125,
+      "learning_rate": 3.8e-05,
+      "loss": 0.6162,
+      "step": 380
+    },
+    {
+      "epoch": 0.10557661072008663,
+      "grad_norm": 0.5390625,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 0.5946,
+      "step": 390
+    },
+    {
+      "epoch": 0.10828370330265295,
+      "grad_norm": 0.337890625,
+      "learning_rate": 4e-05,
+      "loss": 0.5911,
+      "step": 400
+    },
+    {
+      "epoch": 0.10828370330265295,
+      "eval_loss": 0.5952291488647461,
+      "eval_runtime": 69.8814,
+      "eval_samples_per_second": 7.155,
+      "eval_steps_per_second": 0.229,
+      "step": 400
+    },
+    {
+      "epoch": 0.11099079588521928,
+      "grad_norm": 0.349609375,
+      "learning_rate": 4.1e-05,
+      "loss": 0.5777,
+      "step": 410
+    },
+    {
+      "epoch": 0.1136978884677856,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.2e-05,
+      "loss": 0.5755,
+      "step": 420
+    },
+    {
+      "epoch": 0.11640498105035192,
+      "grad_norm": 0.314453125,
+      "learning_rate": 4.3e-05,
+      "loss": 0.5732,
+      "step": 430
+    },
+    {
+      "epoch": 0.11911207363291824,
+      "grad_norm": 0.31640625,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.5729,
+      "step": 440
+    },
+    {
+      "epoch": 0.12181916621548457,
+      "grad_norm": 0.31640625,
+      "learning_rate": 4.5e-05,
+      "loss": 0.569,
+      "step": 450
+    },
+    {
+      "epoch": 0.12452625879805089,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 0.5648,
+      "step": 460
+    },
+    {
+      "epoch": 0.12723335138061723,
+      "grad_norm": 0.29296875,
+      "learning_rate": 4.7e-05,
+      "loss": 0.5778,
+      "step": 470
+    },
+    {
+      "epoch": 0.12994044396318355,
+      "grad_norm": 0.373046875,
+      "learning_rate": 4.8e-05,
+      "loss": 0.5705,
+      "step": 480
+    },
+    {
+      "epoch": 0.13264753654574987,
+      "grad_norm": 0.3671875,
+      "learning_rate": 4.9e-05,
+      "loss": 0.5675,
+      "step": 490
+    },
+    {
+      "epoch": 0.1353546291283162,
+      "grad_norm": 0.328125,
+      "learning_rate": 5e-05,
+      "loss": 0.5752,
+      "step": 500
+    },
+    {
+      "epoch": 0.1353546291283162,
+      "eval_loss": 0.5701072812080383,
+      "eval_runtime": 70.1857,
+      "eval_samples_per_second": 7.124,
+      "eval_steps_per_second": 0.228,
+      "step": 500
+    },
+    {
+      "epoch": 0.13806172171088252,
+      "grad_norm": 0.33984375,
+      "learning_rate": 4.984345648090169e-05,
+      "loss": 0.5513,
+      "step": 510
+    },
+    {
+      "epoch": 0.14076881429344884,
+      "grad_norm": 0.431640625,
+      "learning_rate": 4.9686912961803384e-05,
+      "loss": 0.5644,
+      "step": 520
+    },
+    {
+      "epoch": 0.14347590687601516,
+      "grad_norm": 0.318359375,
+      "learning_rate": 4.9530369442705075e-05,
+      "loss": 0.567,
+      "step": 530
+    },
+    {
+      "epoch": 0.1461829994585815,
+      "grad_norm": 0.33984375,
+      "learning_rate": 4.9373825923606765e-05,
+      "loss": 0.5639,
+      "step": 540
+    },
+    {
+      "epoch": 0.1488900920411478,
+      "grad_norm": 0.333984375,
+      "learning_rate": 4.9217282404508456e-05,
+      "loss": 0.556,
+      "step": 550
+    },
+    {
+      "epoch": 0.15159718462371413,
+      "grad_norm": 0.35546875,
+      "learning_rate": 4.906073888541015e-05,
+      "loss": 0.5697,
+      "step": 560
+    },
+    {
+      "epoch": 0.15430427720628045,
+      "grad_norm": 0.3359375,
+      "learning_rate": 4.890419536631184e-05,
+      "loss": 0.5458,
+      "step": 570
+    },
+    {
+      "epoch": 0.15701136978884678,
+      "grad_norm": 0.345703125,
+      "learning_rate": 4.874765184721353e-05,
+      "loss": 0.5701,
+      "step": 580
+    },
+    {
+      "epoch": 0.1597184623714131,
+      "grad_norm": 0.3984375,
+      "learning_rate": 4.859110832811522e-05,
+      "loss": 0.5665,
+      "step": 590
+    },
+    {
+      "epoch": 0.16242555495397942,
+      "grad_norm": 0.390625,
+      "learning_rate": 4.843456480901691e-05,
+      "loss": 0.55,
+      "step": 600
+    },
+    {
+      "epoch": 0.16242555495397942,
+      "eval_loss": 0.5633333325386047,
+      "eval_runtime": 75.1182,
+      "eval_samples_per_second": 6.656,
+      "eval_steps_per_second": 0.213,
+      "step": 600
+    },
+    {
+      "epoch": 0.16513264753654575,
+      "grad_norm": 0.345703125,
+      "learning_rate": 4.82780212899186e-05,
+      "loss": 0.5584,
+      "step": 610
+    },
+    {
+      "epoch": 0.16783974011911207,
+      "grad_norm": 0.306640625,
+      "learning_rate": 4.812147777082029e-05,
+      "loss": 0.5505,
+      "step": 620
+    },
+    {
+      "epoch": 0.1705468327016784,
+      "grad_norm": 0.38671875,
+      "learning_rate": 4.796493425172198e-05,
+      "loss": 0.5526,
+      "step": 630
+    },
+    {
+      "epoch": 0.17325392528424471,
+      "grad_norm": 0.40625,
+      "learning_rate": 4.780839073262367e-05,
+      "loss": 0.5579,
+      "step": 640
+    },
+    {
+      "epoch": 0.17596101786681104,
+      "grad_norm": 0.373046875,
+      "learning_rate": 4.765184721352536e-05,
+      "loss": 0.5546,
+      "step": 650
+    },
+    {
+      "epoch": 0.17866811044937736,
+      "grad_norm": 0.30078125,
+      "learning_rate": 4.7495303694427054e-05,
+      "loss": 0.5467,
+      "step": 660
+    },
+    {
+      "epoch": 0.18137520303194368,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.7338760175328744e-05,
+      "loss": 0.5496,
+      "step": 670
+    },
+    {
+      "epoch": 0.18408229561451,
+      "grad_norm": 0.404296875,
+      "learning_rate": 4.7182216656230435e-05,
+      "loss": 0.563,
+      "step": 680
+    },
+    {
+      "epoch": 0.18678938819707633,
+      "grad_norm": 0.341796875,
+      "learning_rate": 4.7025673137132126e-05,
+      "loss": 0.5599,
+      "step": 690
+    },
+    {
+      "epoch": 0.18949648077964265,
+      "grad_norm": 0.310546875,
+      "learning_rate": 4.6869129618033816e-05,
+      "loss": 0.5542,
+      "step": 700
+    },
+    {
+      "epoch": 0.18949648077964265,
+      "eval_loss": 0.5565248727798462,
+      "eval_runtime": 70.5824,
+      "eval_samples_per_second": 7.084,
+      "eval_steps_per_second": 0.227,
+      "step": 700
+    },
+    {
+      "epoch": 0.19220357336220897,
+      "grad_norm": 0.33203125,
+      "learning_rate": 4.671258609893551e-05,
+      "loss": 0.5479,
+      "step": 710
+    },
+    {
+      "epoch": 0.19491066594477532,
+      "grad_norm": 0.333984375,
+      "learning_rate": 4.65560425798372e-05,
+      "loss": 0.5443,
+      "step": 720
+    },
+    {
+      "epoch": 0.19761775852734165,
+      "grad_norm": 0.3359375,
+      "learning_rate": 4.639949906073889e-05,
+      "loss": 0.549,
+      "step": 730
+    },
+    {
+      "epoch": 0.20032485110990797,
+      "grad_norm": 0.37890625,
+      "learning_rate": 4.624295554164057e-05,
+      "loss": 0.5605,
+      "step": 740
+    },
+    {
+      "epoch": 0.2030319436924743,
+      "grad_norm": 0.369140625,
+      "learning_rate": 4.608641202254227e-05,
+      "loss": 0.5498,
+      "step": 750
+    },
+    {
+      "epoch": 0.20573903627504062,
+      "grad_norm": 0.318359375,
+      "learning_rate": 4.5929868503443954e-05,
+      "loss": 0.5541,
+      "step": 760
+    },
+    {
+      "epoch": 0.20844612885760694,
+      "grad_norm": 0.392578125,
+      "learning_rate": 4.577332498434565e-05,
+      "loss": 0.5444,
+      "step": 770
+    },
+    {
+      "epoch": 0.21115322144017326,
+      "grad_norm": 0.3671875,
+      "learning_rate": 4.561678146524734e-05,
+      "loss": 0.541,
+      "step": 780
+    },
+    {
+      "epoch": 0.21386031402273958,
+      "grad_norm": 0.353515625,
+      "learning_rate": 4.546023794614903e-05,
+      "loss": 0.5646,
+      "step": 790
+    },
+    {
+      "epoch": 0.2165674066053059,
+      "grad_norm": 0.349609375,
+      "learning_rate": 4.5303694427050724e-05,
+      "loss": 0.543,
+      "step": 800
+    },
+    {
+      "epoch": 0.2165674066053059,
+      "eval_loss": 0.5533172488212585,
+      "eval_runtime": 70.7826,
+      "eval_samples_per_second": 7.064,
+      "eval_steps_per_second": 0.226,
+      "step": 800
+    },
+    {
+      "epoch": 0.21927449918787223,
+      "grad_norm": 0.302734375,
+      "learning_rate": 4.5147150907952414e-05,
+      "loss": 0.5387,
+      "step": 810
+    },
+    {
+      "epoch": 0.22198159177043855,
+      "grad_norm": 0.392578125,
+      "learning_rate": 4.4990607388854105e-05,
+      "loss": 0.5447,
+      "step": 820
+    },
+    {
+      "epoch": 0.22468868435300487,
+      "grad_norm": 0.37109375,
+      "learning_rate": 4.4834063869755796e-05,
+      "loss": 0.5397,
+      "step": 830
+    },
+    {
+      "epoch": 0.2273957769355712,
+      "grad_norm": 0.326171875,
+      "learning_rate": 4.4677520350657486e-05,
+      "loss": 0.5482,
+      "step": 840
+    },
+    {
+      "epoch": 0.23010286951813752,
+      "grad_norm": 0.322265625,
+      "learning_rate": 4.452097683155918e-05,
+      "loss": 0.5553,
+      "step": 850
+    },
+    {
+      "epoch": 0.23280996210070384,
+      "grad_norm": 0.388671875,
+      "learning_rate": 4.436443331246087e-05,
+      "loss": 0.5456,
+      "step": 860
+    },
+    {
+      "epoch": 0.23551705468327017,
+      "grad_norm": 0.431640625,
+      "learning_rate": 4.420788979336256e-05,
+      "loss": 0.5541,
+      "step": 870
+    },
+    {
+      "epoch": 0.2382241472658365,
+      "grad_norm": 0.330078125,
+      "learning_rate": 4.405134627426425e-05,
+      "loss": 0.5377,
+      "step": 880
+    },
+    {
+      "epoch": 0.2409312398484028,
+      "grad_norm": 0.361328125,
+      "learning_rate": 4.389480275516594e-05,
+      "loss": 0.5376,
+      "step": 890
+    },
+    {
+      "epoch": 0.24363833243096913,
+      "grad_norm": 0.35546875,
+      "learning_rate": 4.373825923606763e-05,
+      "loss": 0.5324,
+      "step": 900
+    },
+    {
+      "epoch": 0.24363833243096913,
+      "eval_loss": 0.5525312423706055,
+      "eval_runtime": 71.8319,
+      "eval_samples_per_second": 6.961,
+      "eval_steps_per_second": 0.223,
+      "step": 900
+    },
+    {
+      "epoch": 0.24634542501353546,
+      "grad_norm": 0.333984375,
+      "learning_rate": 4.358171571696932e-05,
+      "loss": 0.5521,
+      "step": 910
+    },
+    {
+      "epoch": 0.24905251759610178,
+      "grad_norm": 0.369140625,
+      "learning_rate": 4.342517219787101e-05,
+      "loss": 0.543,
+      "step": 920
+    },
+    {
+      "epoch": 0.2517596101786681,
+      "grad_norm": 0.322265625,
+      "learning_rate": 4.3268628678772696e-05,
+      "loss": 0.5553,
+      "step": 930
+    },
+    {
+      "epoch": 0.25446670276123445,
+      "grad_norm": 0.392578125,
+      "learning_rate": 4.3112085159674393e-05,
+      "loss": 0.5289,
+      "step": 940
+    },
+    {
+      "epoch": 0.25717379534380075,
+      "grad_norm": 0.294921875,
+      "learning_rate": 4.295554164057608e-05,
+      "loss": 0.5355,
+      "step": 950
+    },
+    {
+      "epoch": 0.2598808879263671,
+      "grad_norm": 0.36328125,
+      "learning_rate": 4.2798998121477775e-05,
+      "loss": 0.5488,
+      "step": 960
+    },
+    {
+      "epoch": 0.2625879805089334,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.264245460237946e-05,
+      "loss": 0.549,
+      "step": 970
+    },
+    {
+      "epoch": 0.26529507309149974,
+      "grad_norm": 0.3359375,
+      "learning_rate": 4.2485911083281156e-05,
+      "loss": 0.5395,
+      "step": 980
+    },
+    {
+      "epoch": 0.26800216567406604,
+      "grad_norm": 0.373046875,
+      "learning_rate": 4.232936756418284e-05,
+      "loss": 0.5408,
+      "step": 990
+    },
+    {
+      "epoch": 0.2707092582566324,
+      "grad_norm": 0.298828125,
+      "learning_rate": 4.217282404508454e-05,
+      "loss": 0.5399,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2707092582566324,
+      "eval_loss": 0.5489118695259094,
+      "eval_runtime": 70.8351,
+      "eval_samples_per_second": 7.059,
+      "eval_steps_per_second": 0.226,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2734163508391987,
+      "grad_norm": 0.306640625,
+      "learning_rate": 4.201628052598622e-05,
+      "loss": 0.5416,
+      "step": 1010
+    },
+    {
+      "epoch": 0.27612344342176504,
+      "grad_norm": 0.306640625,
+      "learning_rate": 4.185973700688792e-05,
+      "loss": 0.5422,
+      "step": 1020
+    },
+    {
+      "epoch": 0.27883053600433133,
+      "grad_norm": 0.359375,
+      "learning_rate": 4.170319348778961e-05,
+      "loss": 0.5417,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2815376285868977,
+      "grad_norm": 0.3125,
+      "learning_rate": 4.15466499686913e-05,
+      "loss": 0.5441,
+      "step": 1040
+    },
+    {
+      "epoch": 0.284244721169464,
+      "grad_norm": 0.369140625,
+      "learning_rate": 4.139010644959299e-05,
+      "loss": 0.5457,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2869518137520303,
+      "grad_norm": 0.326171875,
+      "learning_rate": 4.123356293049468e-05,
+      "loss": 0.5341,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2896589063345966,
+      "grad_norm": 0.3515625,
+      "learning_rate": 4.107701941139637e-05,
+      "loss": 0.5428,
+      "step": 1070
+    },
+    {
+      "epoch": 0.292365998917163,
+      "grad_norm": 0.345703125,
+      "learning_rate": 4.092047589229806e-05,
+      "loss": 0.5404,
+      "step": 1080
+    },
+    {
+      "epoch": 0.29507309149972927,
+      "grad_norm": 0.369140625,
+      "learning_rate": 4.0763932373199754e-05,
+      "loss": 0.5378,
+      "step": 1090
+    },
+    {
+      "epoch": 0.2977801840822956,
+      "grad_norm": 0.345703125,
+      "learning_rate": 4.0607388854101445e-05,
+      "loss": 0.549,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2977801840822956,
+      "eval_loss": 0.5469087362289429,
+      "eval_runtime": 69.8693,
+      "eval_samples_per_second": 7.156,
+      "eval_steps_per_second": 0.229,
+      "step": 1100
+    },
+    {
+      "epoch": 0.3004872766648619,
+      "grad_norm": 0.390625,
+      "learning_rate": 4.0450845335003135e-05,
+      "loss": 0.5505,
+      "step": 1110
+    },
+    {
+      "epoch": 0.30319436924742826,
+      "grad_norm": 0.353515625,
+      "learning_rate": 4.029430181590482e-05,
+      "loss": 0.5404,
+      "step": 1120
+    },
+    {
+      "epoch": 0.30590146182999456,
+      "grad_norm": 0.4375,
+      "learning_rate": 4.013775829680652e-05,
+      "loss": 0.5411,
+      "step": 1130
+    },
+    {
+      "epoch": 0.3086085544125609,
+      "grad_norm": 0.337890625,
+      "learning_rate": 3.99812147777082e-05,
+      "loss": 0.5404,
+      "step": 1140
+    },
+    {
+      "epoch": 0.31131564699512726,
+      "grad_norm": 0.34375,
+      "learning_rate": 3.98246712586099e-05,
+      "loss": 0.5309,
+      "step": 1150
+    },
+    {
+      "epoch": 0.31402273957769355,
+      "grad_norm": 0.369140625,
+      "learning_rate": 3.966812773951158e-05,
+      "loss": 0.5448,
+      "step": 1160
+    },
+    {
+      "epoch": 0.3167298321602599,
+      "grad_norm": 0.375,
+      "learning_rate": 3.951158422041328e-05,
+      "loss": 0.5374,
+      "step": 1170
+    },
+    {
+      "epoch": 0.3194369247428262,
+      "grad_norm": 0.3515625,
+      "learning_rate": 3.9355040701314964e-05,
+      "loss": 0.5482,
+      "step": 1180
+    },
+    {
+      "epoch": 0.32214401732539255,
+      "grad_norm": 0.330078125,
+      "learning_rate": 3.919849718221666e-05,
+      "loss": 0.5411,
+      "step": 1190
+    },
+    {
+      "epoch": 0.32485110990795885,
+      "grad_norm": 0.35546875,
+      "learning_rate": 3.9041953663118345e-05,
+      "loss": 0.5409,
+      "step": 1200
+    },
+    {
+      "epoch": 0.32485110990795885,
+      "eval_loss": 0.546714186668396,
+      "eval_runtime": 71.2431,
+      "eval_samples_per_second": 7.018,
+      "eval_steps_per_second": 0.225,
+      "step": 1200
+    },
+    {
+      "epoch": 0.3275582024905252,
+      "grad_norm": 0.431640625,
+      "learning_rate": 3.888541014402004e-05,
+      "loss": 0.5334,
+      "step": 1210
+    },
+    {
+      "epoch": 0.3302652950730915,
+      "grad_norm": 0.359375,
+      "learning_rate": 3.8728866624921726e-05,
+      "loss": 0.5338,
+      "step": 1220
+    },
+    {
+      "epoch": 0.33297238765565784,
+      "grad_norm": 0.326171875,
+      "learning_rate": 3.8572323105823424e-05,
+      "loss": 0.5433,
+      "step": 1230
+    },
+    {
+      "epoch": 0.33567948023822414,
+      "grad_norm": 0.3515625,
+      "learning_rate": 3.841577958672511e-05,
+      "loss": 0.5372,
+      "step": 1240
+    },
+    {
+      "epoch": 0.3383865728207905,
+      "grad_norm": 0.380859375,
+      "learning_rate": 3.8259236067626805e-05,
+      "loss": 0.5421,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3410936654033568,
+      "grad_norm": 0.31640625,
+      "learning_rate": 3.810269254852849e-05,
+      "loss": 0.5436,
+      "step": 1260
+    },
+    {
+      "epoch": 0.34380075798592313,
+      "grad_norm": 0.361328125,
+      "learning_rate": 3.794614902943019e-05,
+      "loss": 0.5408,
+      "step": 1270
+    },
+    {
+      "epoch": 0.34650785056848943,
+      "grad_norm": 0.365234375,
+      "learning_rate": 3.778960551033187e-05,
+      "loss": 0.5433,
+      "step": 1280
+    },
+    {
+      "epoch": 0.3492149431510558,
+      "grad_norm": 0.3515625,
+      "learning_rate": 3.763306199123356e-05,
+      "loss": 0.5396,
+      "step": 1290
+    },
+    {
+      "epoch": 0.3519220357336221,
+      "grad_norm": 0.421875,
+      "learning_rate": 3.747651847213526e-05,
+      "loss": 0.5438,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3519220357336221,
+      "eval_loss": 0.5430516004562378,
+      "eval_runtime": 70.6664,
+      "eval_samples_per_second": 7.075,
+      "eval_steps_per_second": 0.226,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3546291283161884,
+      "grad_norm": 0.310546875,
+      "learning_rate": 3.731997495303694e-05,
+      "loss": 0.5328,
+      "step": 1310
+    },
+    {
+      "epoch": 0.3573362208987547,
+      "grad_norm": 0.390625,
+      "learning_rate": 3.716343143393864e-05,
+      "loss": 0.5399,
+      "step": 1320
+    },
+    {
+      "epoch": 0.36004331348132107,
+      "grad_norm": 0.408203125,
+      "learning_rate": 3.7006887914840324e-05,
+      "loss": 0.5301,
+      "step": 1330
+    },
+    {
+      "epoch": 0.36275040606388737,
+      "grad_norm": 0.359375,
+      "learning_rate": 3.685034439574202e-05,
+      "loss": 0.5356,
+      "step": 1340
+    },
+    {
+      "epoch": 0.3654574986464537,
+      "grad_norm": 0.341796875,
+      "learning_rate": 3.6693800876643706e-05,
+      "loss": 0.5431,
+      "step": 1350
+    },
+    {
+      "epoch": 0.36816459122902,
+      "grad_norm": 0.353515625,
+      "learning_rate": 3.65372573575454e-05,
+      "loss": 0.5278,
+      "step": 1360
+    },
+    {
+      "epoch": 0.37087168381158636,
+      "grad_norm": 0.376953125,
+      "learning_rate": 3.638071383844709e-05,
+      "loss": 0.5382,
+      "step": 1370
+    },
+    {
+      "epoch": 0.37357877639415266,
+      "grad_norm": 0.341796875,
+      "learning_rate": 3.6224170319348784e-05,
+      "loss": 0.5361,
+      "step": 1380
+    },
+    {
+      "epoch": 0.376285868976719,
+      "grad_norm": 0.35546875,
+      "learning_rate": 3.606762680025047e-05,
+      "loss": 0.5397,
+      "step": 1390
+    },
+    {
+      "epoch": 0.3789929615592853,
+      "grad_norm": 0.390625,
+      "learning_rate": 3.5911083281152166e-05,
+      "loss": 0.5467,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3789929615592853,
+      "eval_loss": 0.5431599020957947,
+      "eval_runtime": 70.1255,
+      "eval_samples_per_second": 7.13,
+      "eval_steps_per_second": 0.228,
+      "step": 1400
+    },
+    {
+      "epoch": 0.38170005414185165,
+      "grad_norm": 0.369140625,
+      "learning_rate": 3.575453976205385e-05,
+      "loss": 0.5332,
+      "step": 1410
+    },
+    {
+      "epoch": 0.38440714672441795,
+      "grad_norm": 0.37890625,
+      "learning_rate": 3.559799624295555e-05,
+      "loss": 0.542,
+      "step": 1420
+    },
+    {
+      "epoch": 0.3871142393069843,
+      "grad_norm": 0.365234375,
+      "learning_rate": 3.544145272385723e-05,
+      "loss": 0.5312,
+      "step": 1430
+    },
+    {
+      "epoch": 0.38982133188955065,
+      "grad_norm": 0.3671875,
+      "learning_rate": 3.528490920475893e-05,
+      "loss": 0.5375,
+      "step": 1440
+    },
+    {
+      "epoch": 0.39252842447211694,
+      "grad_norm": 0.306640625,
+      "learning_rate": 3.512836568566061e-05,
+      "loss": 0.5287,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3952355170546833,
+      "grad_norm": 0.353515625,
+      "learning_rate": 3.497182216656231e-05,
+      "loss": 0.5203,
+      "step": 1460
+    },
+    {
+      "epoch": 0.3979426096372496,
+      "grad_norm": 0.42578125,
+      "learning_rate": 3.4815278647463994e-05,
+      "loss": 0.5135,
+      "step": 1470
+    },
+    {
+      "epoch": 0.40064970221981594,
+      "grad_norm": 0.412109375,
+      "learning_rate": 3.4658735128365685e-05,
+      "loss": 0.549,
+      "step": 1480
+    },
+    {
+      "epoch": 0.40335679480238223,
+      "grad_norm": 0.330078125,
+      "learning_rate": 3.4502191609267375e-05,
+      "loss": 0.5543,
+      "step": 1490
+    },
+    {
+      "epoch": 0.4060638873849486,
+      "grad_norm": 0.369140625,
+      "learning_rate": 3.4345648090169066e-05,
+      "loss": 0.5466,
+      "step": 1500
+    },
+    {
+      "epoch": 0.4060638873849486,
+      "eval_loss": 0.5431920289993286,
+      "eval_runtime": 71.0212,
+      "eval_samples_per_second": 7.04,
+      "eval_steps_per_second": 0.225,
+      "step": 1500
+    },
+    {
+      "epoch": 0.4087709799675149,
+      "grad_norm": 0.353515625,
+      "learning_rate": 3.418910457107076e-05,
+      "loss": 0.5372,
+      "step": 1510
+    },
+    {
+      "epoch": 0.41147807255008123,
+      "grad_norm": 0.369140625,
+      "learning_rate": 3.403256105197245e-05,
+      "loss": 0.5361,
+      "step": 1520
+    },
+    {
+      "epoch": 0.4141851651326475,
+      "grad_norm": 0.357421875,
+      "learning_rate": 3.387601753287414e-05,
+      "loss": 0.5377,
+      "step": 1530
+    },
+    {
+      "epoch": 0.4168922577152139,
+      "grad_norm": 0.38671875,
+      "learning_rate": 3.371947401377583e-05,
+      "loss": 0.5281,
+      "step": 1540
+    },
+    {
+      "epoch": 0.41959935029778017,
+      "grad_norm": 0.380859375,
+      "learning_rate": 3.3562930494677526e-05,
+      "loss": 0.5284,
+      "step": 1550
+    },
+    {
+      "epoch": 0.4223064428803465,
+      "grad_norm": 0.34765625,
+      "learning_rate": 3.340638697557921e-05,
+      "loss": 0.5338,
+      "step": 1560
+    },
+    {
+      "epoch": 0.4250135354629128,
+      "grad_norm": 0.34765625,
+      "learning_rate": 3.324984345648091e-05,
+      "loss": 0.5275,
+      "step": 1570
+    },
+    {
+      "epoch": 0.42772062804547917,
+      "grad_norm": 0.3359375,
+      "learning_rate": 3.309329993738259e-05,
+      "loss": 0.5402,
+      "step": 1580
+    },
+    {
+      "epoch": 0.43042772062804546,
+      "grad_norm": 0.34765625,
+      "learning_rate": 3.293675641828429e-05,
+      "loss": 0.5393,
+      "step": 1590
+    },
+    {
+      "epoch": 0.4331348132106118,
+      "grad_norm": 0.392578125,
+      "learning_rate": 3.278021289918597e-05,
+      "loss": 0.5309,
+      "step": 1600
+    },
+    {
+      "epoch": 0.4331348132106118,
+      "eval_loss": 0.5397661328315735,
+      "eval_runtime": 70.9009,
+      "eval_samples_per_second": 7.052,
+      "eval_steps_per_second": 0.226,
+      "step": 1600
+    },
+    {
+      "epoch": 0.4358419057931781,
+      "grad_norm": 0.4296875,
+      "learning_rate": 3.262366938008767e-05,
+      "loss": 0.537,
+      "step": 1610
+    },
+    {
+      "epoch": 0.43854899837574446,
+      "grad_norm": 0.40625,
+      "learning_rate": 3.2467125860989355e-05,
+      "loss": 0.5405,
+      "step": 1620
+    },
+    {
+      "epoch": 0.44125609095831075,
+      "grad_norm": 0.40234375,
+      "learning_rate": 3.231058234189105e-05,
+      "loss": 0.5228,
+      "step": 1630
+    },
+    {
+      "epoch": 0.4439631835408771,
+      "grad_norm": 0.390625,
+      "learning_rate": 3.2154038822792736e-05,
+      "loss": 0.5315,
+      "step": 1640
+    },
+    {
+      "epoch": 0.4466702761234434,
+      "grad_norm": 0.412109375,
+      "learning_rate": 3.1997495303694433e-05,
+      "loss": 0.5531,
+      "step": 1650
+    },
+    {
+      "epoch": 0.44937736870600975,
+      "grad_norm": 0.3984375,
+      "learning_rate": 3.184095178459612e-05,
+      "loss": 0.5467,
+      "step": 1660
+    },
+    {
+      "epoch": 0.45208446128857604,
+      "grad_norm": 0.38671875,
+      "learning_rate": 3.168440826549781e-05,
+      "loss": 0.5302,
+      "step": 1670
+    },
+    {
+      "epoch": 0.4547915538711424,
+      "grad_norm": 0.482421875,
+      "learning_rate": 3.15278647463995e-05,
+      "loss": 0.5465,
+      "step": 1680
+    },
+    {
+      "epoch": 0.4574986464537087,
+      "grad_norm": 0.380859375,
+      "learning_rate": 3.137132122730119e-05,
+      "loss": 0.5241,
+      "step": 1690
+    },
+    {
+      "epoch": 0.46020573903627504,
+      "grad_norm": 0.3515625,
+      "learning_rate": 3.121477770820288e-05,
+      "loss": 0.5241,
+      "step": 1700
+    },
+    {
+      "epoch": 0.46020573903627504,
+      "eval_loss": 0.540625810623169,
+      "eval_runtime": 70.5131,
+      "eval_samples_per_second": 7.091,
+      "eval_steps_per_second": 0.227,
+      "step": 1700
+    },
+    {
+      "epoch": 0.4629128316188414,
+      "grad_norm": 0.43359375,
+      "learning_rate": 3.105823418910457e-05,
+      "loss": 0.5276,
+      "step": 1710
+    },
+    {
+      "epoch": 0.4656199242014077,
+      "grad_norm": 0.392578125,
+      "learning_rate": 3.090169067000626e-05,
+      "loss": 0.5396,
+      "step": 1720
+    },
+    {
+      "epoch": 0.46832701678397404,
+      "grad_norm": 0.439453125,
+      "learning_rate": 3.074514715090795e-05,
+      "loss": 0.5386,
+      "step": 1730
+    },
+    {
+      "epoch": 0.47103410936654033,
+      "grad_norm": 0.423828125,
+      "learning_rate": 3.058860363180964e-05,
+      "loss": 0.5324,
+      "step": 1740
+    },
+    {
+      "epoch": 0.4737412019491067,
+      "grad_norm": 0.435546875,
+      "learning_rate": 3.0432060112711337e-05,
+      "loss": 0.5328,
+      "step": 1750
+    },
+    {
+      "epoch": 0.476448294531673,
+      "grad_norm": 0.375,
+      "learning_rate": 3.0275516593613024e-05,
+      "loss": 0.5314,
+      "step": 1760
+    },
+    {
+      "epoch": 0.47915538711423933,
+      "grad_norm": 0.396484375,
+      "learning_rate": 3.011897307451472e-05,
+      "loss": 0.5327,
+      "step": 1770
+    },
+    {
+      "epoch": 0.4818624796968056,
+      "grad_norm": 0.365234375,
+      "learning_rate": 2.9962429555416406e-05,
+      "loss": 0.5435,
+      "step": 1780
+    },
+    {
+      "epoch": 0.484569572279372,
+      "grad_norm": 0.353515625,
+      "learning_rate": 2.9805886036318097e-05,
+      "loss": 0.5261,
+      "step": 1790
+    },
+    {
+      "epoch": 0.48727666486193827,
+      "grad_norm": 0.421875,
+      "learning_rate": 2.9649342517219787e-05,
+      "loss": 0.523,
+      "step": 1800
+    },
+    {
+      "epoch": 0.48727666486193827,
+      "eval_loss": 0.5400844812393188,
+      "eval_runtime": 70.1604,
+      "eval_samples_per_second": 7.127,
+      "eval_steps_per_second": 0.228,
+      "step": 1800
+    },
+    {
+      "epoch": 0.4899837574445046,
+      "grad_norm": 0.361328125,
+      "learning_rate": 2.9492798998121478e-05,
+      "loss": 0.5265,
+      "step": 1810
+    },
+    {
+      "epoch": 0.4926908500270709,
+      "grad_norm": 0.3984375,
+      "learning_rate": 2.9336255479023172e-05,
+      "loss": 0.5389,
+      "step": 1820
+    },
+    {
+      "epoch": 0.49539794260963727,
+      "grad_norm": 0.4375,
+      "learning_rate": 2.917971195992486e-05,
+      "loss": 0.5239,
+      "step": 1830
+    },
+    {
+      "epoch": 0.49810503519220356,
+      "grad_norm": 0.494140625,
+      "learning_rate": 2.9023168440826553e-05,
+      "loss": 0.5212,
+      "step": 1840
+    },
+    {
+      "epoch": 0.5008121277747699,
+      "grad_norm": 0.40625,
+      "learning_rate": 2.886662492172824e-05,
+      "loss": 0.5437,
+      "step": 1850
+    },
+    {
+      "epoch": 0.5035192203573362,
+      "grad_norm": 0.423828125,
+      "learning_rate": 2.8710081402629935e-05,
+      "loss": 0.5356,
+      "step": 1860
+    },
+    {
+      "epoch": 0.5062263129399025,
+      "grad_norm": 0.37109375,
+      "learning_rate": 2.8553537883531622e-05,
+      "loss": 0.5233,
+      "step": 1870
+    },
+    {
+      "epoch": 0.5089334055224689,
+      "grad_norm": 0.38671875,
+      "learning_rate": 2.8396994364433316e-05,
+      "loss": 0.5255,
+      "step": 1880
+    },
+    {
+      "epoch": 0.5116404981050352,
+      "grad_norm": 0.431640625,
+      "learning_rate": 2.8240450845335004e-05,
+      "loss": 0.5206,
+      "step": 1890
+    },
+    {
+      "epoch": 0.5143475906876015,
+      "grad_norm": 0.365234375,
+      "learning_rate": 2.8083907326236698e-05,
+      "loss": 0.5241,
+      "step": 1900
+    },
+    {
+      "epoch": 0.5143475906876015,
+      "eval_loss": 0.5394680500030518,
+      "eval_runtime": 70.7559,
+      "eval_samples_per_second": 7.067,
+      "eval_steps_per_second": 0.226,
+      "step": 1900
+    },
+    {
+      "epoch": 0.5170546832701678,
+      "grad_norm": 0.419921875,
+      "learning_rate": 2.7927363807138385e-05,
+      "loss": 0.5319,
+      "step": 1910
+    },
+    {
+      "epoch": 0.5197617758527342,
+      "grad_norm": 0.431640625,
+      "learning_rate": 2.777082028804008e-05,
+      "loss": 0.5351,
+      "step": 1920
+    },
+    {
+      "epoch": 0.5224688684353005,
+      "grad_norm": 0.43359375,
+      "learning_rate": 2.7614276768941766e-05,
+      "loss": 0.5284,
+      "step": 1930
+    },
+    {
+      "epoch": 0.5251759610178668,
+      "grad_norm": 0.447265625,
+      "learning_rate": 2.745773324984346e-05,
+      "loss": 0.531,
+      "step": 1940
+    },
+    {
+      "epoch": 0.5278830536004331,
+      "grad_norm": 0.451171875,
+      "learning_rate": 2.7301189730745148e-05,
+      "loss": 0.533,
+      "step": 1950
+    },
+    {
+      "epoch": 0.5305901461829995,
+      "grad_norm": 0.421875,
+      "learning_rate": 2.7144646211646842e-05,
+      "loss": 0.5226,
+      "step": 1960
+    },
+    {
+      "epoch": 0.5332972387655658,
+      "grad_norm": 0.486328125,
+      "learning_rate": 2.698810269254853e-05,
+      "loss": 0.5311,
+      "step": 1970
+    },
+    {
+      "epoch": 0.5360043313481321,
+      "grad_norm": 0.466796875,
+      "learning_rate": 2.683155917345022e-05,
+      "loss": 0.5244,
+      "step": 1980
+    },
+    {
+      "epoch": 0.5387114239306985,
+      "grad_norm": 0.388671875,
+      "learning_rate": 2.667501565435191e-05,
+      "loss": 0.5393,
+      "step": 1990
+    },
+    {
+      "epoch": 0.5414185165132648,
+      "grad_norm": 0.453125,
+      "learning_rate": 2.65184721352536e-05,
+      "loss": 0.5365,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5414185165132648,
+      "eval_loss": 0.5388503670692444,
+      "eval_runtime": 70.6892,
+      "eval_samples_per_second": 7.073,
+      "eval_steps_per_second": 0.226,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5441256090958311,
+      "grad_norm": 0.404296875,
+      "learning_rate": 2.636192861615529e-05,
+      "loss": 0.5295,
+      "step": 2010
+    },
+    {
+      "epoch": 0.5468327016783974,
+      "grad_norm": 0.345703125,
+      "learning_rate": 2.6205385097056983e-05,
+      "loss": 0.5302,
+      "step": 2020
+    },
+    {
+      "epoch": 0.5495397942609638,
+      "grad_norm": 0.40234375,
+      "learning_rate": 2.604884157795867e-05,
+      "loss": 0.5305,
+      "step": 2030
+    },
+    {
+      "epoch": 0.5522468868435301,
+      "grad_norm": 0.423828125,
+      "learning_rate": 2.5892298058860364e-05,
+      "loss": 0.5196,
+      "step": 2040
+    },
+    {
+      "epoch": 0.5549539794260964,
+      "grad_norm": 0.384765625,
+      "learning_rate": 2.573575453976205e-05,
+      "loss": 0.5223,
+      "step": 2050
+    },
+    {
+      "epoch": 0.5576610720086627,
+      "grad_norm": 0.34765625,
+      "learning_rate": 2.5579211020663746e-05,
+      "loss": 0.5312,
+      "step": 2060
+    },
+    {
+      "epoch": 0.5603681645912291,
+      "grad_norm": 0.431640625,
+      "learning_rate": 2.5422667501565433e-05,
+      "loss": 0.532,
+      "step": 2070
+    },
+    {
+      "epoch": 0.5630752571737954,
+      "grad_norm": 0.396484375,
+      "learning_rate": 2.5266123982467127e-05,
+      "loss": 0.5216,
+      "step": 2080
+    },
+    {
+      "epoch": 0.5657823497563617,
+      "grad_norm": 0.39453125,
+      "learning_rate": 2.510958046336882e-05,
+      "loss": 0.5334,
+      "step": 2090
+    },
+    {
+      "epoch": 0.568489442338928,
+      "grad_norm": 0.435546875,
+      "learning_rate": 2.495303694427051e-05,
+      "loss": 0.5296,
+      "step": 2100
+    },
+    {
+      "epoch": 0.568489442338928,
+      "eval_loss": 0.5370610952377319,
+      "eval_runtime": 71.1301,
+      "eval_samples_per_second": 7.029,
+      "eval_steps_per_second": 0.225,
+      "step": 2100
+    },
+    {
+      "epoch": 0.5711965349214944,
+      "grad_norm": 0.3359375,
+      "learning_rate": 2.47964934251722e-05,
+      "loss": 0.5315,
+      "step": 2110
+    },
+    {
+      "epoch": 0.5739036275040607,
+      "grad_norm": 0.3671875,
+      "learning_rate": 2.463994990607389e-05,
+      "loss": 0.51,
+      "step": 2120
+    },
+    {
+      "epoch": 0.576610720086627,
+      "grad_norm": 0.423828125,
+      "learning_rate": 2.448340638697558e-05,
+      "loss": 0.5339,
+      "step": 2130
+    },
+    {
+      "epoch": 0.5793178126691932,
+      "grad_norm": 0.375,
+      "learning_rate": 2.432686286787727e-05,
+      "loss": 0.5215,
+      "step": 2140
+    },
+    {
+      "epoch": 0.5820249052517596,
+      "grad_norm": 0.392578125,
+      "learning_rate": 2.4170319348778962e-05,
+      "loss": 0.5235,
+      "step": 2150
+    },
+    {
+      "epoch": 0.584731997834326,
+      "grad_norm": 0.40625,
+      "learning_rate": 2.4013775829680653e-05,
+      "loss": 0.5246,
+      "step": 2160
+    },
+    {
+      "epoch": 0.5874390904168922,
+      "grad_norm": 0.4453125,
+      "learning_rate": 2.3857232310582343e-05,
+      "loss": 0.528,
+      "step": 2170
+    },
+    {
+      "epoch": 0.5901461829994585,
+      "grad_norm": 0.447265625,
+      "learning_rate": 2.3700688791484034e-05,
+      "loss": 0.532,
+      "step": 2180
+    },
+    {
+      "epoch": 0.5928532755820249,
+      "grad_norm": 0.38671875,
+      "learning_rate": 2.3544145272385725e-05,
+      "loss": 0.5283,
+      "step": 2190
+    },
+    {
+      "epoch": 0.5955603681645912,
+      "grad_norm": 0.396484375,
+      "learning_rate": 2.3387601753287412e-05,
+      "loss": 0.5256,
+      "step": 2200
+    },
+    {
+      "epoch": 0.5955603681645912,
+      "eval_loss": 0.5376315712928772,
+      "eval_runtime": 70.6924,
+      "eval_samples_per_second": 7.073,
+      "eval_steps_per_second": 0.226,
+      "step": 2200
+    },
+    {
+      "epoch": 0.5982674607471575,
+      "grad_norm": 0.373046875,
+      "learning_rate": 2.3231058234189106e-05,
+      "loss": 0.5264,
+      "step": 2210
+    },
+    {
+      "epoch": 0.6009745533297238,
+      "grad_norm": 0.40625,
+      "learning_rate": 2.3074514715090797e-05,
+      "loss": 0.5271,
+      "step": 2220
+    },
+    {
+      "epoch": 0.6036816459122902,
+      "grad_norm": 0.39453125,
+      "learning_rate": 2.2917971195992488e-05,
+      "loss": 0.5483,
+      "step": 2230
+    },
+    {
+      "epoch": 0.6063887384948565,
+      "grad_norm": 0.37109375,
+      "learning_rate": 2.2761427676894178e-05,
+      "loss": 0.5243,
+      "step": 2240
+    },
+    {
+      "epoch": 0.6090958310774228,
+      "grad_norm": 0.37109375,
+      "learning_rate": 2.260488415779587e-05,
+      "loss": 0.5223,
+      "step": 2250
+    },
+    {
+      "epoch": 0.6118029236599891,
+      "grad_norm": 0.44140625,
+      "learning_rate": 2.244834063869756e-05,
+      "loss": 0.5245,
+      "step": 2260
+    },
+    {
+      "epoch": 0.6145100162425555,
+      "grad_norm": 0.53125,
+      "learning_rate": 2.229179711959925e-05,
+      "loss": 0.5253,
+      "step": 2270
+    },
+    {
+      "epoch": 0.6172171088251218,
+      "grad_norm": 0.5390625,
+      "learning_rate": 2.213525360050094e-05,
+      "loss": 0.5339,
+      "step": 2280
+    },
+    {
+      "epoch": 0.6199242014076881,
+      "grad_norm": 0.431640625,
+      "learning_rate": 2.1978710081402632e-05,
+      "loss": 0.5269,
+      "step": 2290
+    },
+    {
+      "epoch": 0.6226312939902545,
+      "grad_norm": 0.404296875,
+      "learning_rate": 2.1822166562304323e-05,
+      "loss": 0.5319,
+      "step": 2300
+    },
+    {
+      "epoch": 0.6226312939902545,
+      "eval_loss": 0.5361697673797607,
+      "eval_runtime": 70.5195,
+      "eval_samples_per_second": 7.09,
+      "eval_steps_per_second": 0.227,
+      "step": 2300
+    },
+    {
+      "epoch": 0.6253383865728208,
+      "grad_norm": 0.486328125,
+      "learning_rate": 2.1665623043206013e-05,
+      "loss": 0.5139,
+      "step": 2310
+    },
+    {
+      "epoch": 0.6280454791553871,
+      "grad_norm": 0.51953125,
+      "learning_rate": 2.1509079524107704e-05,
+      "loss": 0.5373,
+      "step": 2320
+    },
+    {
+      "epoch": 0.6307525717379534,
+      "grad_norm": 0.458984375,
+      "learning_rate": 2.1352536005009395e-05,
+      "loss": 0.53,
+      "step": 2330
+    },
+    {
+      "epoch": 0.6334596643205198,
+      "grad_norm": 0.37890625,
+      "learning_rate": 2.1195992485911085e-05,
+      "loss": 0.5282,
+      "step": 2340
+    },
+    {
+      "epoch": 0.6361667569030861,
+      "grad_norm": 0.4296875,
+      "learning_rate": 2.1039448966812776e-05,
+      "loss": 0.5211,
+      "step": 2350
+    },
+    {
+      "epoch": 0.6388738494856524,
+      "grad_norm": 0.388671875,
+      "learning_rate": 2.0882905447714467e-05,
+      "loss": 0.5268,
+      "step": 2360
+    },
+    {
+      "epoch": 0.6415809420682187,
+      "grad_norm": 0.369140625,
+      "learning_rate": 2.0726361928616157e-05,
+      "loss": 0.5222,
+      "step": 2370
+    },
+    {
+      "epoch": 0.6442880346507851,
+      "grad_norm": 0.52734375,
+      "learning_rate": 2.0569818409517845e-05,
+      "loss": 0.5249,
+      "step": 2380
+    },
+    {
+      "epoch": 0.6469951272333514,
+      "grad_norm": 0.41015625,
+      "learning_rate": 2.0413274890419535e-05,
+      "loss": 0.542,
+      "step": 2390
+    },
+    {
+      "epoch": 0.6497022198159177,
+      "grad_norm": 0.396484375,
+      "learning_rate": 2.0256731371321226e-05,
+      "loss": 0.5172,
+      "step": 2400
+    },
+    {
+      "epoch": 0.6497022198159177,
+      "eval_loss": 0.5371243953704834,
+      "eval_runtime": 70.8703,
+      "eval_samples_per_second": 7.055,
+      "eval_steps_per_second": 0.226,
+      "step": 2400
+    },
+    {
+      "epoch": 0.652409312398484,
+      "grad_norm": 0.494140625,
+      "learning_rate": 2.0100187852222917e-05,
+      "loss": 0.5347,
+      "step": 2410
+    },
+    {
+      "epoch": 0.6551164049810504,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.9943644333124608e-05,
+      "loss": 0.5288,
+      "step": 2420
+    },
+    {
+      "epoch": 0.6578234975636167,
+      "grad_norm": 0.4453125,
+      "learning_rate": 1.9787100814026298e-05,
+      "loss": 0.5282,
+      "step": 2430
+    },
+    {
+      "epoch": 0.660530590146183,
+      "grad_norm": 0.4453125,
+      "learning_rate": 1.963055729492799e-05,
+      "loss": 0.5151,
+      "step": 2440
+    },
+    {
+      "epoch": 0.6632376827287493,
+      "grad_norm": 0.470703125,
+      "learning_rate": 1.947401377582968e-05,
+      "loss": 0.5339,
+      "step": 2450
+    },
+    {
+      "epoch": 0.6659447753113157,
+      "grad_norm": 0.412109375,
+      "learning_rate": 1.931747025673137e-05,
+      "loss": 0.5322,
+      "step": 2460
+    },
+    {
+      "epoch": 0.668651867893882,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.9160926737633064e-05,
+      "loss": 0.512,
+      "step": 2470
+    },
+    {
+      "epoch": 0.6713589604764483,
+      "grad_norm": 0.421875,
+      "learning_rate": 1.9004383218534755e-05,
+      "loss": 0.5334,
+      "step": 2480
+    },
+    {
+      "epoch": 0.6740660530590146,
+      "grad_norm": 0.3828125,
+      "learning_rate": 1.8847839699436446e-05,
+      "loss": 0.5265,
+      "step": 2490
+    },
+    {
+      "epoch": 0.676773145641581,
+      "grad_norm": 0.39453125,
+      "learning_rate": 1.8691296180338137e-05,
+      "loss": 0.5355,
+      "step": 2500
+    },
+    {
+      "epoch": 0.676773145641581,
+      "eval_loss": 0.5358985066413879,
+      "eval_runtime": 72.2037,
+      "eval_samples_per_second": 6.925,
+      "eval_steps_per_second": 0.222,
+      "step": 2500
+    },
+    {
+      "epoch": 0.6794802382241473,
+      "grad_norm": 0.412109375,
+      "learning_rate": 1.8534752661239827e-05,
+      "loss": 0.534,
+      "step": 2510
+    },
+    {
+      "epoch": 0.6821873308067136,
+      "grad_norm": 0.388671875,
+      "learning_rate": 1.8378209142141518e-05,
+      "loss": 0.5373,
+      "step": 2520
+    },
+    {
+      "epoch": 0.6848944233892799,
+      "grad_norm": 0.447265625,
+      "learning_rate": 1.822166562304321e-05,
+      "loss": 0.5381,
+      "step": 2530
+    },
+    {
+      "epoch": 0.6876015159718463,
+      "grad_norm": 0.427734375,
+      "learning_rate": 1.80651221039449e-05,
+      "loss": 0.5087,
+      "step": 2540
+    },
+    {
+      "epoch": 0.6903086085544126,
+      "grad_norm": 0.427734375,
+      "learning_rate": 1.790857858484659e-05,
+      "loss": 0.5328,
+      "step": 2550
+    },
+    {
+      "epoch": 0.6930157011369789,
+      "grad_norm": 0.39453125,
+      "learning_rate": 1.775203506574828e-05,
+      "loss": 0.515,
+      "step": 2560
+    },
+    {
+      "epoch": 0.6957227937195453,
+      "grad_norm": 0.33984375,
+      "learning_rate": 1.7595491546649968e-05,
+      "loss": 0.5161,
+      "step": 2570
+    },
+    {
+      "epoch": 0.6984298863021116,
+      "grad_norm": 0.388671875,
+      "learning_rate": 1.743894802755166e-05,
+      "loss": 0.5295,
+      "step": 2580
+    },
+    {
+      "epoch": 0.7011369788846779,
+      "grad_norm": 0.396484375,
+      "learning_rate": 1.728240450845335e-05,
+      "loss": 0.5188,
+      "step": 2590
+    },
+    {
+      "epoch": 0.7038440714672441,
+      "grad_norm": 0.408203125,
+      "learning_rate": 1.712586098935504e-05,
+      "loss": 0.5081,
+      "step": 2600
+    },
+    {
+      "epoch": 0.7038440714672441,
+      "eval_loss": 0.5356553792953491,
+      "eval_runtime": 70.9086,
+      "eval_samples_per_second": 7.051,
+      "eval_steps_per_second": 0.226,
+      "step": 2600
+    },
+    {
+      "epoch": 0.7065511640498106,
+      "grad_norm": 0.404296875,
+      "learning_rate": 1.696931747025673e-05,
+      "loss": 0.5228,
+      "step": 2610
+    },
+    {
+      "epoch": 0.7092582566323768,
+      "grad_norm": 0.41015625,
+      "learning_rate": 1.681277395115842e-05,
+      "loss": 0.5256,
+      "step": 2620
+    },
+    {
+      "epoch": 0.7119653492149431,
+      "grad_norm": 0.515625,
+      "learning_rate": 1.6656230432060112e-05,
+      "loss": 0.5379,
+      "step": 2630
+    },
+    {
+      "epoch": 0.7146724417975094,
+      "grad_norm": 0.421875,
+      "learning_rate": 1.6499686912961803e-05,
+      "loss": 0.5289,
+      "step": 2640
+    },
+    {
+      "epoch": 0.7173795343800758,
+      "grad_norm": 0.40625,
+      "learning_rate": 1.6343143393863494e-05,
+      "loss": 0.5217,
+      "step": 2650
+    },
+    {
+      "epoch": 0.7200866269626421,
+      "grad_norm": 0.44921875,
+      "learning_rate": 1.6186599874765184e-05,
+      "loss": 0.5284,
+      "step": 2660
+    },
+    {
+      "epoch": 0.7227937195452084,
+      "grad_norm": 0.376953125,
+      "learning_rate": 1.6030056355666875e-05,
+      "loss": 0.5143,
+      "step": 2670
+    },
+    {
+      "epoch": 0.7255008121277747,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.5873512836568566e-05,
+      "loss": 0.5426,
+      "step": 2680
+    },
+    {
+      "epoch": 0.7282079047103411,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.5716969317470257e-05,
+      "loss": 0.5211,
+      "step": 2690
+    },
+    {
+      "epoch": 0.7309149972929074,
+      "grad_norm": 0.4609375,
+      "learning_rate": 1.5560425798371947e-05,
+      "loss": 0.5318,
+      "step": 2700
+    },
+    {
+      "epoch": 0.7309149972929074,
+      "eval_loss": 0.5345706343650818,
+      "eval_runtime": 71.1706,
+      "eval_samples_per_second": 7.025,
+      "eval_steps_per_second": 0.225,
+      "step": 2700
+    },
+    {
+      "epoch": 0.7336220898754737,
+      "grad_norm": 0.466796875,
+      "learning_rate": 1.5403882279273638e-05,
+      "loss": 0.5184,
+      "step": 2710
+    },
+    {
+      "epoch": 0.73632918245804,
+      "grad_norm": 0.47265625,
+      "learning_rate": 1.5247338760175329e-05,
+      "loss": 0.5234,
+      "step": 2720
+    },
+    {
+      "epoch": 0.7390362750406064,
+      "grad_norm": 0.427734375,
+      "learning_rate": 1.5090795241077021e-05,
+      "loss": 0.523,
+      "step": 2730
+    },
+    {
+      "epoch": 0.7417433676231727,
+      "grad_norm": 0.376953125,
+      "learning_rate": 1.4934251721978712e-05,
+      "loss": 0.5274,
+      "step": 2740
+    },
+    {
+      "epoch": 0.744450460205739,
+      "grad_norm": 0.482421875,
+      "learning_rate": 1.4777708202880403e-05,
+      "loss": 0.5179,
+      "step": 2750
+    },
+    {
+      "epoch": 0.7471575527883053,
+      "grad_norm": 0.421875,
+      "learning_rate": 1.4621164683782093e-05,
+      "loss": 0.5255,
+      "step": 2760
+    },
+    {
+      "epoch": 0.7498646453708717,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.4464621164683784e-05,
+      "loss": 0.5413,
+      "step": 2770
+    },
+    {
+      "epoch": 0.752571737953438,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.4308077645585475e-05,
+      "loss": 0.5164,
+      "step": 2780
+    },
+    {
+      "epoch": 0.7552788305360043,
+      "grad_norm": 0.349609375,
+      "learning_rate": 1.4151534126487165e-05,
+      "loss": 0.5266,
+      "step": 2790
+    },
+    {
+      "epoch": 0.7579859231185706,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.3994990607388856e-05,
+      "loss": 0.5436,
+      "step": 2800
+    },
+    {
+      "epoch": 0.7579859231185706,
+      "eval_loss": 0.5348847508430481,
+      "eval_runtime": 70.2342,
+      "eval_samples_per_second": 7.119,
+      "eval_steps_per_second": 0.228,
+      "step": 2800
+    },
+    {
+      "epoch": 0.760693015701137,
+      "grad_norm": 0.4140625,
+      "learning_rate": 1.3838447088290547e-05,
+      "loss": 0.5208,
+      "step": 2810
+    },
+    {
+      "epoch": 0.7634001082837033,
+      "grad_norm": 0.40234375,
+      "learning_rate": 1.3681903569192236e-05,
+      "loss": 0.5145,
+      "step": 2820
+    },
+    {
+      "epoch": 0.7661072008662696,
+      "grad_norm": 0.375,
+      "learning_rate": 1.3525360050093926e-05,
+      "loss": 0.5099,
+      "step": 2830
+    },
+    {
+      "epoch": 0.7688142934488359,
+      "grad_norm": 0.380859375,
+      "learning_rate": 1.3368816530995617e-05,
+      "loss": 0.5377,
+      "step": 2840
+    },
+    {
+      "epoch": 0.7715213860314023,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.3212273011897308e-05,
+      "loss": 0.5156,
+      "step": 2850
+    },
+    {
+      "epoch": 0.7742284786139686,
+      "grad_norm": 0.466796875,
+      "learning_rate": 1.3055729492798999e-05,
+      "loss": 0.5171,
+      "step": 2860
+    },
+    {
+      "epoch": 0.7769355711965349,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.289918597370069e-05,
+      "loss": 0.5226,
+      "step": 2870
+    },
+    {
+      "epoch": 0.7796426637791013,
+      "grad_norm": 0.333984375,
+      "learning_rate": 1.274264245460238e-05,
+      "loss": 0.5347,
+      "step": 2880
+    },
+    {
+      "epoch": 0.7823497563616676,
+      "grad_norm": 0.458984375,
+      "learning_rate": 1.258609893550407e-05,
+      "loss": 0.5331,
+      "step": 2890
+    },
+    {
+      "epoch": 0.7850568489442339,
+      "grad_norm": 0.470703125,
+      "learning_rate": 1.2429555416405761e-05,
+      "loss": 0.5124,
+      "step": 2900
+    },
+    {
+      "epoch": 0.7850568489442339,
+      "eval_loss": 0.5337508916854858,
+      "eval_runtime": 70.6912,
+      "eval_samples_per_second": 7.073,
+      "eval_steps_per_second": 0.226,
+      "step": 2900
+    },
+    {
+      "epoch": 0.7877639415268002,
+      "grad_norm": 0.427734375,
+      "learning_rate": 1.2273011897307452e-05,
+      "loss": 0.5211,
+      "step": 2910
+    },
+    {
+      "epoch": 0.7904710341093666,
+      "grad_norm": 0.45703125,
+      "learning_rate": 1.2116468378209143e-05,
+      "loss": 0.5151,
+      "step": 2920
+    },
+    {
+      "epoch": 0.7931781266919329,
+      "grad_norm": 0.400390625,
+      "learning_rate": 1.1959924859110834e-05,
+      "loss": 0.5144,
+      "step": 2930
+    },
+    {
+      "epoch": 0.7958852192744992,
+      "grad_norm": 0.455078125,
+      "learning_rate": 1.1803381340012524e-05,
+      "loss": 0.5349,
+      "step": 2940
+    },
+    {
+      "epoch": 0.7985923118570655,
+      "grad_norm": 0.458984375,
+      "learning_rate": 1.1646837820914215e-05,
+      "loss": 0.5272,
+      "step": 2950
+    },
+    {
+      "epoch": 0.8012994044396319,
+      "grad_norm": 0.40234375,
+      "learning_rate": 1.1490294301815906e-05,
+      "loss": 0.5262,
+      "step": 2960
+    },
+    {
+      "epoch": 0.8040064970221982,
+      "grad_norm": 0.40234375,
+      "learning_rate": 1.1333750782717596e-05,
+      "loss": 0.5352,
+      "step": 2970
+    },
+    {
+      "epoch": 0.8067135896047645,
+      "grad_norm": 0.359375,
+      "learning_rate": 1.1177207263619287e-05,
+      "loss": 0.528,
+      "step": 2980
+    },
+    {
+      "epoch": 0.8094206821873308,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.1020663744520978e-05,
+      "loss": 0.5194,
+      "step": 2990
+    },
+    {
+      "epoch": 0.8121277747698972,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.0864120225422668e-05,
+      "loss": 0.5224,
+      "step": 3000
+    },
+    {
+      "epoch": 0.8121277747698972,
+      "eval_loss": 0.532988429069519,
+      "eval_runtime": 70.1018,
+      "eval_samples_per_second": 7.132,
+      "eval_steps_per_second": 0.228,
+      "step": 3000
+    },
+    {
+      "epoch": 0.8148348673524635,
+      "grad_norm": 0.447265625,
+      "learning_rate": 1.070757670632436e-05,
+      "loss": 0.5239,
+      "step": 3010
+    },
+    {
+      "epoch": 0.8175419599350298,
+      "grad_norm": 0.37890625,
+      "learning_rate": 1.0551033187226048e-05,
+      "loss": 0.5323,
+      "step": 3020
+    },
+    {
+      "epoch": 0.8202490525175961,
+      "grad_norm": 0.388671875,
+      "learning_rate": 1.0394489668127739e-05,
+      "loss": 0.5334,
+      "step": 3030
+    },
+    {
+      "epoch": 0.8229561451001625,
+      "grad_norm": 0.390625,
+      "learning_rate": 1.023794614902943e-05,
+      "loss": 0.5318,
+      "step": 3040
+    },
+    {
+      "epoch": 0.8256632376827288,
+      "grad_norm": 0.45703125,
+      "learning_rate": 1.008140262993112e-05,
+      "loss": 0.5202,
+      "step": 3050
+    },
+    {
+      "epoch": 0.828370330265295,
+      "grad_norm": 0.41796875,
+      "learning_rate": 9.924859110832813e-06,
+      "loss": 0.5239,
+      "step": 3060
+    },
+    {
+      "epoch": 0.8310774228478613,
+      "grad_norm": 0.462890625,
+      "learning_rate": 9.768315591734503e-06,
+      "loss": 0.5261,
+      "step": 3070
+    },
+    {
+      "epoch": 0.8337845154304278,
+      "grad_norm": 0.384765625,
+      "learning_rate": 9.611772072636194e-06,
+      "loss": 0.5242,
+      "step": 3080
+    },
+    {
+      "epoch": 0.836491608012994,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.455228553537885e-06,
+      "loss": 0.5327,
+      "step": 3090
+    },
+    {
+      "epoch": 0.8391987005955603,
+      "grad_norm": 0.4296875,
+      "learning_rate": 9.298685034439576e-06,
+      "loss": 0.5204,
+      "step": 3100
+    },
+    {
+      "epoch": 0.8391987005955603,
+      "eval_loss": 0.5333003997802734,
+      "eval_runtime": 71.1268,
+      "eval_samples_per_second": 7.03,
+      "eval_steps_per_second": 0.225,
+      "step": 3100
+    },
+    {
+      "epoch": 0.8419057931781266,
+      "grad_norm": 0.4765625,
+      "learning_rate": 9.142141515341266e-06,
+      "loss": 0.5437,
+      "step": 3110
+    },
+    {
+      "epoch": 0.844612885760693,
+      "grad_norm": 0.462890625,
+      "learning_rate": 8.985597996242955e-06,
+      "loss": 0.5297,
+      "step": 3120
+    },
+    {
+      "epoch": 0.8473199783432593,
+      "grad_norm": 0.4296875,
+      "learning_rate": 8.829054477144646e-06,
+      "loss": 0.5305,
+      "step": 3130
+    },
+    {
+      "epoch": 0.8500270709258256,
+      "grad_norm": 0.400390625,
+      "learning_rate": 8.672510958046337e-06,
+      "loss": 0.5486,
+      "step": 3140
+    },
+    {
+      "epoch": 0.852734163508392,
+      "grad_norm": 0.40234375,
+      "learning_rate": 8.515967438948027e-06,
+      "loss": 0.5296,
+      "step": 3150
+    },
+    {
+      "epoch": 0.8554412560909583,
+      "grad_norm": 0.376953125,
+      "learning_rate": 8.359423919849718e-06,
+      "loss": 0.5323,
+      "step": 3160
+    },
+    {
+      "epoch": 0.8581483486735246,
+      "grad_norm": 0.427734375,
+      "learning_rate": 8.202880400751409e-06,
+      "loss": 0.5381,
+      "step": 3170
+    },
+    {
+      "epoch": 0.8608554412560909,
+      "grad_norm": 0.44140625,
+      "learning_rate": 8.0463368816531e-06,
+      "loss": 0.5278,
+      "step": 3180
+    },
+    {
+      "epoch": 0.8635625338386573,
+      "grad_norm": 0.51953125,
+      "learning_rate": 7.889793362554792e-06,
+      "loss": 0.5252,
+      "step": 3190
+    },
+    {
+      "epoch": 0.8662696264212236,
+      "grad_norm": 0.447265625,
+      "learning_rate": 7.733249843456483e-06,
+      "loss": 0.5215,
+      "step": 3200
+    },
+    {
+      "epoch": 0.8662696264212236,
+      "eval_loss": 0.5333092212677002,
+      "eval_runtime": 70.1888,
+      "eval_samples_per_second": 7.124,
+      "eval_steps_per_second": 0.228,
+      "step": 3200
+    },
+    {
+      "epoch": 0.8689767190037899,
+      "grad_norm": 0.453125,
+      "learning_rate": 7.576706324358172e-06,
+      "loss": 0.5226,
+      "step": 3210
+    },
+    {
+      "epoch": 0.8716838115863562,
+      "grad_norm": 0.416015625,
+      "learning_rate": 7.420162805259863e-06,
+      "loss": 0.5325,
+      "step": 3220
+    },
+    {
+      "epoch": 0.8743909041689226,
+      "grad_norm": 0.4375,
+      "learning_rate": 7.263619286161554e-06,
+      "loss": 0.5318,
+      "step": 3230
+    },
+    {
+      "epoch": 0.8770979967514889,
+      "grad_norm": 0.451171875,
+      "learning_rate": 7.107075767063244e-06,
+      "loss": 0.5224,
+      "step": 3240
+    },
+    {
+      "epoch": 0.8798050893340552,
+      "grad_norm": 0.42578125,
+      "learning_rate": 6.950532247964934e-06,
+      "loss": 0.5196,
+      "step": 3250
+    },
+    {
+      "epoch": 0.8825121819166215,
+      "grad_norm": 0.5546875,
+      "learning_rate": 6.793988728866625e-06,
+      "loss": 0.5205,
+      "step": 3260
+    },
+    {
+      "epoch": 0.8852192744991879,
+      "grad_norm": 0.408203125,
+      "learning_rate": 6.637445209768316e-06,
+      "loss": 0.5291,
+      "step": 3270
+    },
+    {
+      "epoch": 0.8879263670817542,
+      "grad_norm": 0.435546875,
+      "learning_rate": 6.4809016906700065e-06,
+      "loss": 0.5229,
+      "step": 3280
+    },
+    {
+      "epoch": 0.8906334596643205,
+      "grad_norm": 0.439453125,
+      "learning_rate": 6.324358171571697e-06,
+      "loss": 0.5286,
+      "step": 3290
+    },
+    {
+      "epoch": 0.8933405522468868,
+      "grad_norm": 0.47265625,
+      "learning_rate": 6.167814652473388e-06,
+      "loss": 0.5373,
+      "step": 3300
+    },
+    {
+      "epoch": 0.8933405522468868,
+      "eval_loss": 0.5331239104270935,
+      "eval_runtime": 70.3149,
+      "eval_samples_per_second": 7.111,
+      "eval_steps_per_second": 0.228,
+      "step": 3300
+    },
+    {
+      "epoch": 0.8960476448294532,
+      "grad_norm": 0.408203125,
+      "learning_rate": 6.011271133375079e-06,
+      "loss": 0.5325,
+      "step": 3310
+    },
+    {
+      "epoch": 0.8987547374120195,
+      "grad_norm": 0.455078125,
+      "learning_rate": 5.854727614276769e-06,
+      "loss": 0.5176,
+      "step": 3320
+    },
+    {
+      "epoch": 0.9014618299945858,
+      "grad_norm": 0.435546875,
+      "learning_rate": 5.69818409517846e-06,
+      "loss": 0.5185,
+      "step": 3330
+    },
+    {
+      "epoch": 0.9041689225771521,
+      "grad_norm": 0.48046875,
+      "learning_rate": 5.54164057608015e-06,
+      "loss": 0.5155,
+      "step": 3340
+    },
+    {
+      "epoch": 0.9068760151597185,
+      "grad_norm": 0.36328125,
+      "learning_rate": 5.3850970569818414e-06,
+      "loss": 0.5176,
+      "step": 3350
+    },
+    {
+      "epoch": 0.9095831077422848,
+      "grad_norm": 0.45703125,
+      "learning_rate": 5.228553537883532e-06,
+      "loss": 0.5201,
+      "step": 3360
+    },
+    {
+      "epoch": 0.9122902003248511,
+      "grad_norm": 0.40625,
+      "learning_rate": 5.072010018785223e-06,
+      "loss": 0.5267,
+      "step": 3370
+    },
+    {
+      "epoch": 0.9149972929074174,
+      "grad_norm": 0.421875,
+      "learning_rate": 4.9154664996869136e-06,
+      "loss": 0.5171,
+      "step": 3380
+    },
+    {
+      "epoch": 0.9177043854899838,
+      "grad_norm": 0.4140625,
+      "learning_rate": 4.758922980588603e-06,
+      "loss": 0.517,
+      "step": 3390
+    },
+    {
+      "epoch": 0.9204114780725501,
+      "grad_norm": 0.44140625,
+      "learning_rate": 4.602379461490294e-06,
+      "loss": 0.5138,
+      "step": 3400
+    },
+    {
+      "epoch": 0.9204114780725501,
+      "eval_loss": 0.5331882238388062,
+      "eval_runtime": 70.6838,
+      "eval_samples_per_second": 7.074,
+      "eval_steps_per_second": 0.226,
+      "step": 3400
+    },
+    {
+      "epoch": 0.9231185706551164,
+      "grad_norm": 0.384765625,
+      "learning_rate": 4.445835942391985e-06,
+      "loss": 0.5122,
+      "step": 3410
+    },
+    {
+      "epoch": 0.9258256632376828,
+      "grad_norm": 0.49609375,
+      "learning_rate": 4.289292423293676e-06,
+      "loss": 0.5288,
+      "step": 3420
+    },
+    {
+      "epoch": 0.9285327558202491,
+      "grad_norm": 0.3984375,
+      "learning_rate": 4.132748904195367e-06,
+      "loss": 0.5262,
+      "step": 3430
+    },
+    {
+      "epoch": 0.9312398484028154,
+      "grad_norm": 0.462890625,
+      "learning_rate": 3.976205385097057e-06,
+      "loss": 0.5329,
+      "step": 3440
+    },
+    {
+      "epoch": 0.9339469409853817,
+      "grad_norm": 0.37890625,
+      "learning_rate": 3.819661865998748e-06,
+      "loss": 0.5339,
+      "step": 3450
+    },
+    {
+      "epoch": 0.9366540335679481,
+      "grad_norm": 0.431640625,
+      "learning_rate": 3.6631183469004384e-06,
+      "loss": 0.5403,
+      "step": 3460
+    },
+    {
+      "epoch": 0.9393611261505144,
+      "grad_norm": 0.39453125,
+      "learning_rate": 3.506574827802129e-06,
+      "loss": 0.5145,
+      "step": 3470
+    },
+    {
+      "epoch": 0.9420682187330807,
+      "grad_norm": 0.474609375,
+      "learning_rate": 3.35003130870382e-06,
+      "loss": 0.5253,
+      "step": 3480
+    },
+    {
+      "epoch": 0.944775311315647,
+      "grad_norm": 0.4140625,
+      "learning_rate": 3.193487789605511e-06,
+      "loss": 0.5183,
+      "step": 3490
+    },
+    {
+      "epoch": 0.9474824038982134,
+      "grad_norm": 0.408203125,
+      "learning_rate": 3.036944270507201e-06,
+      "loss": 0.5153,
+      "step": 3500
+    },
+    {
+      "epoch": 0.9474824038982134,
+      "eval_loss": 0.5332643985748291,
+      "eval_runtime": 71.6079,
+      "eval_samples_per_second": 6.982,
+      "eval_steps_per_second": 0.223,
+      "step": 3500
+    },
+    {
+      "epoch": 0.9501894964807797,
+      "grad_norm": 0.5234375,
+      "learning_rate": 2.880400751408892e-06,
+      "loss": 0.5083,
+      "step": 3510
+    },
+    {
+      "epoch": 0.952896589063346,
+      "grad_norm": 0.57421875,
+      "learning_rate": 2.7238572323105826e-06,
+      "loss": 0.5247,
+      "step": 3520
+    },
+    {
+      "epoch": 0.9556036816459123,
+      "grad_norm": 0.5234375,
+      "learning_rate": 2.5673137132122733e-06,
+      "loss": 0.5215,
+      "step": 3530
+    },
+    {
+      "epoch": 0.9583107742284787,
+      "grad_norm": 0.453125,
+      "learning_rate": 2.410770194113964e-06,
+      "loss": 0.5254,
+      "step": 3540
+    },
+    {
+      "epoch": 0.961017866811045,
+      "grad_norm": 0.42578125,
+      "learning_rate": 2.2542266750156543e-06,
+      "loss": 0.5253,
+      "step": 3550
+    },
+    {
+      "epoch": 0.9637249593936112,
+      "grad_norm": 0.4609375,
+      "learning_rate": 2.0976831559173454e-06,
+      "loss": 0.5354,
+      "step": 3560
+    },
+    {
+      "epoch": 0.9664320519761775,
+      "grad_norm": 0.458984375,
+      "learning_rate": 1.9411396368190357e-06,
+      "loss": 0.5227,
+      "step": 3570
+    },
+    {
+      "epoch": 0.969139144558744,
+      "grad_norm": 0.44140625,
+      "learning_rate": 1.7845961177207264e-06,
+      "loss": 0.5226,
+      "step": 3580
+    },
+    {
+      "epoch": 0.9718462371413102,
+      "grad_norm": 0.46875,
+      "learning_rate": 1.6280525986224169e-06,
+      "loss": 0.5331,
+      "step": 3590
+    },
+    {
+      "epoch": 0.9745533297238765,
+      "grad_norm": 0.43359375,
+      "learning_rate": 1.4715090795241078e-06,
+      "loss": 0.5319,
+      "step": 3600
+    },
+    {
+      "epoch": 0.9745533297238765,
+      "eval_loss": 0.5332371592521667,
+      "eval_runtime": 70.7142,
+      "eval_samples_per_second": 7.071,
+      "eval_steps_per_second": 0.226,
+      "step": 3600
+    },
+    {
+      "epoch": 0.9772604223064428,
+      "grad_norm": 0.40234375,
+      "learning_rate": 1.3149655604257985e-06,
+      "loss": 0.5263,
+      "step": 3610
+    },
+    {
+      "epoch": 0.9799675148890092,
+      "grad_norm": 0.439453125,
+      "learning_rate": 1.1584220413274892e-06,
+      "loss": 0.5123,
+      "step": 3620
+    },
+    {
+      "epoch": 0.9826746074715755,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.0018785222291797e-06,
+      "loss": 0.5197,
+      "step": 3630
+    },
+    {
+      "epoch": 0.9853817000541418,
+      "grad_norm": 0.51953125,
+      "learning_rate": 8.453350031308704e-07,
+      "loss": 0.5253,
+      "step": 3640
+    },
+    {
+      "epoch": 0.9880887926367081,
+      "grad_norm": 0.37890625,
+      "learning_rate": 6.887914840325611e-07,
+      "loss": 0.5301,
+      "step": 3650
+    },
+    {
+      "epoch": 0.9907958852192745,
+      "grad_norm": 0.46875,
+      "learning_rate": 5.322479649342517e-07,
+      "loss": 0.532,
+      "step": 3660
+    },
+    {
+      "epoch": 0.9935029778018408,
+      "grad_norm": 0.419921875,
+      "learning_rate": 3.757044458359424e-07,
+      "loss": 0.5349,
+      "step": 3670
+    },
+    {
+      "epoch": 0.9962100703844071,
+      "grad_norm": 0.478515625,
+      "learning_rate": 2.1916092673763307e-07,
+      "loss": 0.5278,
+      "step": 3680
+    },
+    {
+      "epoch": 0.9989171629669734,
+      "grad_norm": 0.392578125,
+      "learning_rate": 6.261740763932373e-08,
+      "loss": 0.5235,
+      "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": 1.0993863336142579e+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:d26ff0bd7f07448258fc45a5b47b5248d2f44416c5ab44b40b3aade04e73c9c7
+size 5176

git_hash.txt

@@ -0,0 +1 @@
+ad1211bf04a805374df2dab1868655c381c11c6d

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": "BiQwen2Processor",
+  "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": 0.654, "ndcg_at_3": 0.74488, "ndcg_at_5": 0.76485, "ndcg_at_10": 0.77602, "ndcg_at_20": 0.78474, "ndcg_at_100": 0.79733, "ndcg_at_1000": 0.79875, "map_at_1": 0.654, "map_at_3": 0.723, "map_at_5": 0.7342, "map_at_10": 0.7389, "map_at_20": 0.74137, "map_at_100": 0.74325, "map_at_1000": 0.74333, "recall_at_1": 0.654, "recall_at_3": 0.808, "recall_at_5": 0.856, "recall_at_10": 0.89, "recall_at_20": 0.924, "recall_at_100": 0.99, "recall_at_1000": 1.0, "precision_at_1": 0.654, "precision_at_3": 0.26933, "precision_at_5": 0.1712, "precision_at_10": 0.089, "precision_at_20": 0.0462, "precision_at_100": 0.0099, "precision_at_1000": 0.001, "mrr_at_1": 0.656, "mrr_at_3": 0.7233333333333332, "mrr_at_5": 0.735033333333333, "mrr_at_10": 0.7397373015873013, "mrr_at_20": 0.7422049430102057, "mrr_at_100": 0.7440758589391191, "mrr_at_1000": 0.7441544798215474, "naucs_at_1_max": 0.5424486060403283, "naucs_at_1_std": -0.2286461254842529, "naucs_at_1_diff1": 0.7430088797478107, "naucs_at_3_max": 0.659422470324563, "naucs_at_3_std": -0.15940690628372156, "naucs_at_3_diff1": 0.6914999584958922, "naucs_at_5_max": 0.673485789928679, "naucs_at_5_std": -0.11610519775232316, "naucs_at_5_diff1": 0.6829816835962839, "naucs_at_10_max": 0.700601322910404, "naucs_at_10_std": -0.09349712224035832, "naucs_at_10_diff1": 0.6729920109956204, "naucs_at_20_max": 0.7408103592314129, "naucs_at_20_std": -0.06297606762003415, "naucs_at_20_diff1": 0.6291709666322639, "naucs_at_100_max": 0.9444444444444274, "naucs_at_100_std": -0.11746031746034293, "naucs_at_100_diff1": 0.9183006535947666, "naucs_at_1000_max": 1.0, "naucs_at_1000_std": 1.0, "naucs_at_1000_diff1": 1.0}, "syntheticDocQA_energy": {"ndcg_at_1": 0.61, "ndcg_at_3": 0.71309, "ndcg_at_5": 0.73419, "ndcg_at_10": 0.75159, "ndcg_at_20": 0.76467, "ndcg_at_100": 0.7723, "ndcg_at_1000": 0.77503, "map_at_1": 0.61, "map_at_3": 0.68667, "map_at_5": 0.69867, "map_at_10": 0.70658, "map_at_20": 0.7104, "map_at_100": 0.71151, "map_at_1000": 0.71165, "recall_at_1": 0.61, "recall_at_3": 0.79, "recall_at_5": 0.84, "recall_at_10": 0.89, "recall_at_20": 0.94, "recall_at_100": 0.98, "recall_at_1000": 1.0, "precision_at_1": 0.61, "precision_at_3": 0.26333, "precision_at_5": 0.168, "precision_at_10": 0.089, "precision_at_20": 0.047, "precision_at_100": 0.0098, "precision_at_1000": 0.001, "mrr_at_1": 0.63, "mrr_at_3": 0.7, "mrr_at_5": 0.7095000000000001, "mrr_at_10": 0.7174166666666669, "mrr_at_20": 0.7213066002624826, "mrr_at_100": 0.7224397983693116, "mrr_at_1000": 0.7225753046206667, "naucs_at_1_max": 0.23386158657259817, "naucs_at_1_std": -0.2906349614007525, "naucs_at_1_diff1": 0.6505219585152507, "naucs_at_3_max": 0.06316877870488354, "naucs_at_3_std": -0.451799841296559, "naucs_at_3_diff1": 0.6346935340370786, "naucs_at_5_max": 0.1687223858615598, "naucs_at_5_std": -0.5421882670594004, "naucs_at_5_diff1": 0.7002638684339698, "naucs_at_10_max": 0.41061764453225646, "naucs_at_10_std": -0.18409071385619744, "naucs_at_10_diff1": 0.7034619018984622, "naucs_at_20_max": 0.23646125116713304, "naucs_at_20_std": -0.17460317460317504, "naucs_at_20_diff1": 0.5091036414565855, "naucs_at_100_max": 0.795751633986937, "naucs_at_100_std": 0.4561157796451911, "naucs_at_100_diff1": 0.4561157796451911, "naucs_at_1000_max": NaN, "naucs_at_1000_std": NaN, "naucs_at_1000_diff1": NaN}, "syntheticDocQA_healthcare_industry": {"ndcg_at_1": 0.69, "ndcg_at_3": 0.78178, "ndcg_at_5": 0.81105, "ndcg_at_10": 0.82061, "ndcg_at_20": 0.82603, "ndcg_at_100": 0.83174, "ndcg_at_1000": 0.83174, "map_at_1": 0.69, "map_at_3": 0.75833, "map_at_5": 0.77483, "map_at_10": 0.77869, "map_at_20": 0.78037, "map_at_100": 0.7812, "map_at_1000": 0.7812, "recall_at_1": 0.69, "recall_at_3": 0.85, "recall_at_5": 0.92, "recall_at_10": 0.95, "recall_at_20": 0.97, "recall_at_100": 1.0, "recall_at_1000": 1.0, "precision_at_1": 0.69, "precision_at_3": 0.28333, "precision_at_5": 0.184, "precision_at_10": 0.095, "precision_at_20": 0.0485, "precision_at_100": 0.01, "precision_at_1000": 0.001, "mrr_at_1": 0.69, "mrr_at_3": 0.76, "mrr_at_5": 0.7765000000000001, "mrr_at_10": 0.7805952380952381, "mrr_at_20": 0.78227355977356, "mrr_at_100": 0.7831095674757421, "mrr_at_1000": 0.7831095674757421, "naucs_at_1_max": 0.22142265022840735, "naucs_at_1_std": -0.22965941760586991, "naucs_at_1_diff1": 0.6699825387586643, "naucs_at_3_max": 0.4295151317930353, "naucs_at_3_std": -0.31630328669052893, "naucs_at_3_diff1": 0.5726651480637807, "naucs_at_5_max": 0.37511671335200797, "naucs_at_5_std": -0.8036881419234269, "naucs_at_5_diff1": 0.7068160597572369, "naucs_at_10_max": 0.6909430438842239, "naucs_at_10_std": -0.5556489262371483, "naucs_at_10_diff1": 0.7063492063492085, "naucs_at_20_max": 0.5720510426392758, "naucs_at_20_std": -0.7791783380018631, "naucs_at_20_diff1": 0.6591970121381856, "naucs_at_100_max": NaN, "naucs_at_100_std": NaN, "naucs_at_100_diff1": NaN, "naucs_at_1000_max": NaN, "naucs_at_1000_std": NaN, "naucs_at_1000_diff1": NaN}, "syntheticDocQA_artificial_intelligence_test": {"ndcg_at_1": 0.67, "ndcg_at_3": 0.73917, "ndcg_at_5": 0.77274, "ndcg_at_10": 0.78964, "ndcg_at_20": 0.79482, "ndcg_at_100": 0.80581, "ndcg_at_1000": 0.80581, "map_at_1": 0.67, "map_at_3": 0.72167, "map_at_5": 0.74067, "map_at_10": 0.74805, "map_at_20": 0.74953, "map_at_100": 0.75095, "map_at_1000": 0.75095, "recall_at_1": 0.67, "recall_at_3": 0.79, "recall_at_5": 0.87, "recall_at_10": 0.92, "recall_at_20": 0.94, "recall_at_100": 1.0, "recall_at_1000": 1.0, "precision_at_1": 0.67, "precision_at_3": 0.26333, "precision_at_5": 0.174, "precision_at_10": 0.092, "precision_at_20": 0.047, "precision_at_100": 0.01, "precision_at_1000": 0.001, "mrr_at_1": 0.67, "mrr_at_3": 0.725, "mrr_at_5": 0.7415, "mrr_at_10": 0.7488809523809524, "mrr_at_20": 0.7503644688644688, "mrr_at_100": 0.7518057124116964, "mrr_at_1000": 0.7518057124116964, "naucs_at_1_max": 0.34800494320394065, "naucs_at_1_std": -0.16266865297692681, "naucs_at_1_diff1": 0.6552792402363256, "naucs_at_3_max": 0.6477985909058118, "naucs_at_3_std": -0.02707576887007951, "naucs_at_3_diff1": 0.5577945030899083, "naucs_at_5_max": 0.7339850403613998, "naucs_at_5_std": 0.13378508479597254, "naucs_at_5_diff1": 0.5720580611715914, "naucs_at_10_max": 0.7024393090569564, "naucs_at_10_std": -0.03699813258636463, "naucs_at_10_diff1": 0.4370331465919713, "naucs_at_20_max": 0.6468253968253975, "naucs_at_20_std": -0.041238717709307156, "naucs_at_20_diff1": 0.46988795518207455, "naucs_at_100_max": NaN, "naucs_at_100_std": NaN, "naucs_at_100_diff1": NaN, "naucs_at_1000_max": NaN, "naucs_at_1000_std": NaN, "naucs_at_1000_diff1": NaN}, "syntheticDocQA_government_reports": {"ndcg_at_1": 0.64, "ndcg_at_3": 0.72678, "ndcg_at_5": 0.75992, "ndcg_at_10": 0.783, "ndcg_at_20": 0.78858, "ndcg_at_100": 0.79643, "ndcg_at_1000": 0.79643, "map_at_1": 0.64, "map_at_3": 0.705, "map_at_5": 0.7235, "map_at_10": 0.73329, "map_at_20": 0.73511, "map_at_100": 0.73633, "map_at_1000": 0.73633, "recall_at_1": 0.64, "recall_at_3": 0.79, "recall_at_5": 0.87, "recall_at_10": 0.94, "recall_at_20": 0.96, "recall_at_100": 1.0, "recall_at_1000": 1.0, "precision_at_1": 0.64, "precision_at_3": 0.26333, "precision_at_5": 0.174, "precision_at_10": 0.094, "precision_at_20": 0.048, "precision_at_100": 0.01, "precision_at_1000": 0.001, "mrr_at_1": 0.64, "mrr_at_3": 0.7050000000000002, "mrr_at_5": 0.7240000000000002, "mrr_at_10": 0.7337896825396828, "mrr_at_20": 0.7356078643578647, "mrr_at_100": 0.73683114867159, "mrr_at_1000": 0.73683114867159, "naucs_at_1_max": -0.07770788095542226, "naucs_at_1_std": -0.2624057989526122, "naucs_at_1_diff1": 0.7647368757184824, "naucs_at_3_max": -0.14247240724264795, "naucs_at_3_std": -0.44718301392261994, "naucs_at_3_diff1": 0.6850698535600062, "naucs_at_5_max": 0.2072502406872557, "naucs_at_5_std": -0.35751314522698247, "naucs_at_5_diff1": 0.5192920091831437, "naucs_at_10_max": 0.11399004046062691, "naucs_at_10_std": -0.7413632119514489, "naucs_at_10_diff1": 0.5308901338313118, "naucs_at_20_max": 0.3203781512605042, "naucs_at_20_std": -0.8302987861811303, "naucs_at_20_diff1": 0.5262605042016808, "naucs_at_100_max": NaN, "naucs_at_100_std": NaN, "naucs_at_100_diff1": NaN, "naucs_at_1000_max": NaN, "naucs_at_1000_std": NaN, "naucs_at_1000_diff1": NaN}, "infovqa_subsampled": {"ndcg_at_1": 0.63, "ndcg_at_3": 0.71335, "ndcg_at_5": 0.73273, "ndcg_at_10": 0.75089, "ndcg_at_20": 0.76004, "ndcg_at_100": 0.76788, "ndcg_at_1000": 0.7744, "map_at_1": 0.63, "map_at_3": 0.69367, "map_at_5": 0.70467, "map_at_10": 0.7122, "map_at_20": 0.71475, "map_at_100": 0.71585, "map_at_1000": 0.71612, "recall_at_1": 0.63, "recall_at_3": 0.77, "recall_at_5": 0.816, "recall_at_10": 0.872, "recall_at_20": 0.908, "recall_at_100": 0.95, "recall_at_1000": 1.0, "precision_at_1": 0.63, "precision_at_3": 0.25667, "precision_at_5": 0.1632, "precision_at_10": 0.0872, "precision_at_20": 0.0454, "precision_at_100": 0.0095, "precision_at_1000": 0.001, "mrr_at_1": 0.628, "mrr_at_3": 0.6929999999999997, "mrr_at_5": 0.7043999999999996, "mrr_at_10": 0.7110206349206348, "mrr_at_20": 0.7137322081427344, "mrr_at_100": 0.7148334815579046, "mrr_at_1000": 0.7150973312254364, "naucs_at_1_max": 0.2879612654538615, "naucs_at_1_std": -0.17261317162600479, "naucs_at_1_diff1": 0.6919538068600259, "naucs_at_3_max": 0.3511022582305684, "naucs_at_3_std": -0.1458158948993979, "naucs_at_3_diff1": 0.5988451467066533, "naucs_at_5_max": 0.4033426153779729, "naucs_at_5_std": -0.08899969779389545, "naucs_at_5_diff1": 0.6016707680352285, "naucs_at_10_max": 0.5376892346509666, "naucs_at_10_std": 0.07437973086627317, "naucs_at_10_diff1": 0.5457317073170729, "naucs_at_20_max": 0.6503876913084073, "naucs_at_20_std": 0.3150042625745933, "naucs_at_20_diff1": 0.5015527950310549, "naucs_at_100_max": 0.8512044817927161, "naucs_at_100_std": 0.6533893557423003, "naucs_at_100_diff1": 0.5630252100840363, "naucs_at_1000_max": 1.0, "naucs_at_1000_std": 1.0, "naucs_at_1000_diff1": 1.0}, "docvqa_subsampled": {"ndcg_at_1": 0.238, "ndcg_at_3": 0.31847, "ndcg_at_5": 0.34507, "ndcg_at_10": 0.3759, "ndcg_at_20": 0.398, "ndcg_at_100": 0.4315, "ndcg_at_1000": 0.45687, "map_at_1": 0.238, "map_at_3": 0.298, "map_at_5": 0.3129, "map_at_10": 0.32549, "map_at_20": 0.33174, "map_at_100": 0.33641, "map_at_1000": 0.3374, "recall_at_1": 0.238, "recall_at_3": 0.378, "recall_at_5": 0.442, "recall_at_10": 0.538, "recall_at_20": 0.624, "recall_at_100": 0.804, "recall_at_1000": 1.0, "precision_at_1": 0.238, "precision_at_3": 0.126, "precision_at_5": 0.0884, "precision_at_10": 0.0538, "precision_at_20": 0.0312, "precision_at_100": 0.00804, "precision_at_1000": 0.001, "mrr_at_1": 0.238, "mrr_at_3": 0.2986666666666666, "mrr_at_5": 0.31356666666666666, "mrr_at_10": 0.32620238095238085, "mrr_at_20": 0.3326839395039085, "mrr_at_100": 0.337163189635003, "mrr_at_1000": 0.33814966621974824, "naucs_at_1_max": 0.37426232514512997, "naucs_at_1_std": 0.03749778284175053, "naucs_at_1_diff1": 0.5223624835779973, "naucs_at_3_max": 0.4079919243007262, "naucs_at_3_std": 0.11157681670712347, "naucs_at_3_diff1": 0.3746354357729545, "naucs_at_5_max": 0.47320941189990573, "naucs_at_5_std": 0.18466504908887219, "naucs_at_5_diff1": 0.3441601349780741, "naucs_at_10_max": 0.5009135597778619, "naucs_at_10_std": 0.23186396692885855, "naucs_at_10_diff1": 0.2878087596381246, "naucs_at_20_max": 0.5574729743132002, "naucs_at_20_std": 0.332202097590003, "naucs_at_20_diff1": 0.24505665476544666, "naucs_at_100_max": 0.7063661719497548, "naucs_at_100_std": 0.5383303813188383, "naucs_at_100_diff1": 0.27940308231030814, "naucs_at_1000_max": 1.0, "naucs_at_1000_std": 1.0, "naucs_at_1000_diff1": 1.0}, "arxivqa_subsampled": {"ndcg_at_1": 0.616, "ndcg_at_3": 0.69709, "ndcg_at_5": 0.70785, "ndcg_at_10": 0.72543, "ndcg_at_20": 0.73755, "ndcg_at_100": 0.75612, "ndcg_at_1000": 0.7591, "map_at_1": 0.616, "map_at_3": 0.678, "map_at_5": 0.684, "map_at_10": 0.69132, "map_at_20": 0.69464, "map_at_100": 0.69735, "map_at_1000": 0.69749, "recall_at_1": 0.616, "recall_at_3": 0.752, "recall_at_5": 0.778, "recall_at_10": 0.832, "recall_at_20": 0.88, "recall_at_100": 0.978, "recall_at_1000": 1.0, "precision_at_1": 0.616, "precision_at_3": 0.25067, "precision_at_5": 0.1556, "precision_at_10": 0.0832, "precision_at_20": 0.044, "precision_at_100": 0.00978, "precision_at_1000": 0.001, "mrr_at_1": 0.614, "mrr_at_3": 0.6773333333333332, "mrr_at_5": 0.6839333333333332, "mrr_at_10": 0.6908960317460314, "mrr_at_20": 0.6943154074253918, "mrr_at_100": 0.6969429379176545, "mrr_at_1000": 0.6970832194369305, "naucs_at_1_max": 0.2696576706454412, "naucs_at_1_std": -0.4221115197126487, "naucs_at_1_diff1": 0.7556566199557733, "naucs_at_3_max": 0.22414969999325837, "naucs_at_3_std": -0.5203220858895704, "naucs_at_3_diff1": 0.6703001752848374, "naucs_at_5_max": 0.2149923808909975, "naucs_at_5_std": -0.4618423051602793, "naucs_at_5_diff1": 0.6562746009750616, "naucs_at_10_max": 0.21354044411060547, "naucs_at_10_std": -0.5508250786348065, "naucs_at_10_diff1": 0.6116496878080835, "naucs_at_20_max": 0.20875755647470573, "naucs_at_20_std": -0.5492284441616272, "naucs_at_20_diff1": 0.5875357938275549, "naucs_at_100_max": 0.45390883626177436, "naucs_at_100_std": -0.28486546133605595, "naucs_at_100_diff1": 0.6827943298531558, "naucs_at_1000_max": 1.0, "naucs_at_1000_std": 1.0, "naucs_at_1000_diff1": 1.0}, "tabfquad_subsampled": {"ndcg_at_1": 0.72143, "ndcg_at_3": 0.80689, "ndcg_at_5": 0.82764, "ndcg_at_10": 0.84264, "ndcg_at_20": 0.85001, "ndcg_at_100": 0.85202, "ndcg_at_1000": 0.85202, "map_at_1": 0.72143, "map_at_3": 0.7869, "map_at_5": 0.79851, "map_at_10": 0.8047, "map_at_20": 0.8068, "map_at_100": 0.80707, "map_at_1000": 0.80707, "recall_at_1": 0.72143, "recall_at_3": 0.86429, "recall_at_5": 0.91429, "recall_at_10": 0.96071, "recall_at_20": 0.98929, "recall_at_100": 1.0, "recall_at_1000": 1.0, "precision_at_1": 0.72143, "precision_at_3": 0.2881, "precision_at_5": 0.18286, "precision_at_10": 0.09607, "precision_at_20": 0.04946, "precision_at_100": 0.01, "precision_at_1000": 0.001, "mrr_at_1": 0.7142857142857143, "mrr_at_3": 0.7809523809523811, "mrr_at_5": 0.7943452380952382, "mrr_at_10": 0.8005328798185941, "mrr_at_20": 0.8026187697505697, "mrr_at_100": 0.8028969807868453, "mrr_at_1000": 0.8028969807868453, "naucs_at_1_max": 0.6117649473813859, "naucs_at_1_std": -0.08210282867817133, "naucs_at_1_diff1": 0.7093865039070517, "naucs_at_3_max": 0.6382385872054547, "naucs_at_3_std": 0.03996386584559059, "naucs_at_3_diff1": 0.5676879230495184, "naucs_at_5_max": 0.6357181761593519, "naucs_at_5_std": 0.10261048863990033, "naucs_at_5_diff1": 0.4911492374727658, "naucs_at_10_max": 0.7294372294372314, "naucs_at_10_std": 0.3711909006026658, "naucs_at_10_diff1": 0.4168152109328591, "naucs_at_20_max": 0.820261437908502, "naucs_at_20_std": 0.6640211640211727, "naucs_at_20_diff1": 0.24696545284782312, "naucs_at_100_max": 1.0, "naucs_at_100_std": 1.0, "naucs_at_100_diff1": 1.0, "naucs_at_1000_max": NaN, "naucs_at_1000_std": NaN, "naucs_at_1000_diff1": NaN}, "tatdqa": {"ndcg_at_1": 0.25797, "ndcg_at_3": 0.36742, "ndcg_at_5": 0.411, "ndcg_at_10": 0.4552, "ndcg_at_20": 0.48287, "ndcg_at_100": 0.51206, "ndcg_at_1000": 0.51804, "map_at_1": 0.25797, "map_at_3": 0.33915, "map_at_5": 0.36332, "map_at_10": 0.38172, "map_at_20": 0.38955, "map_at_100": 0.39361, "map_at_1000": 0.39388, "recall_at_1": 0.25797, "recall_at_3": 0.44979, "recall_at_5": 0.55562, "recall_at_10": 0.69152, "recall_at_20": 0.79916, "recall_at_100": 0.9561, "recall_at_1000": 1.0, "precision_at_1": 0.25797, "precision_at_3": 0.14993, "precision_at_5": 0.11112, "precision_at_10": 0.06915, "precision_at_20": 0.03996, "precision_at_100": 0.00956, "precision_at_1000": 0.001, "mrr_at_1": 0.25917017438364404, "mrr_at_3": 0.3386450190418916, "mrr_at_5": 0.36353978753257143, "mrr_at_10": 0.3816897173782327, "mrr_at_20": 0.38959489532784913, "mrr_at_100": 0.393673179772931, "mrr_at_1000": 0.39395809866826603, "naucs_at_1_max": 0.07651717209332923, "naucs_at_1_std": -0.170613975030892, "naucs_at_1_diff1": 0.4069030045158366, "naucs_at_3_max": 0.10186478252621102, "naucs_at_3_std": -0.15025760798995128, "naucs_at_3_diff1": 0.25624884527344044, "naucs_at_5_max": 0.10392462082206755, "naucs_at_5_std": -0.11065497042619936, "naucs_at_5_diff1": 0.21117051594147637, "naucs_at_10_max": 0.22807420896702585, "naucs_at_10_std": -0.005265186873844066, "naucs_at_10_diff1": 0.1958616429018192, "naucs_at_20_max": 0.24932293803177794, "naucs_at_20_std": 0.10619499344456142, "naucs_at_20_diff1": 0.1956965423270599, "naucs_at_100_max": 0.44195985099952334, "naucs_at_100_std": 0.3418484734766106, "naucs_at_100_diff1": 0.19878169612163848, "naucs_at_1000_max": NaN, "naucs_at_1000_std": NaN, "naucs_at_1000_diff1": NaN}, "shift_project": {"ndcg_at_1": 0.35, "ndcg_at_3": 0.50857, "ndcg_at_5": 0.54127, "ndcg_at_10": 0.58302, "ndcg_at_20": 0.59655, "ndcg_at_100": 0.61464, "ndcg_at_1000": 0.61729, "map_at_1": 0.35, "map_at_3": 0.47, "map_at_5": 0.488, "map_at_10": 0.50505, "map_at_20": 0.50924, "map_at_100": 0.51159, "map_at_1000": 0.5117, "recall_at_1": 0.35, "recall_at_3": 0.62, "recall_at_5": 0.7, "recall_at_10": 0.83, "recall_at_20": 0.88, "recall_at_100": 0.98, "recall_at_1000": 1.0, "precision_at_1": 0.35, "precision_at_3": 0.20667, "precision_at_5": 0.14, "precision_at_10": 0.083, "precision_at_20": 0.044, "precision_at_100": 0.0098, "precision_at_1000": 0.001, "mrr_at_1": 0.38, "mrr_at_3": 0.49000000000000005, "mrr_at_5": 0.5115, "mrr_at_10": 0.5285515873015874, "mrr_at_20": 0.5318965756465757, "mrr_at_100": 0.534266246185204, "mrr_at_1000": 0.5343760163001465, "naucs_at_1_max": -0.06241677104621251, "naucs_at_1_std": -0.1998154130133824, "naucs_at_1_diff1": 0.4911727866042588, "naucs_at_3_max": -0.03931944743562056, "naucs_at_3_std": -0.23156324031627729, "naucs_at_3_diff1": 0.26706765529335796, "naucs_at_5_max": 0.0717528373266085, "naucs_at_5_std": -0.15932264456854553, "naucs_at_5_diff1": 0.24467663484056926, "naucs_at_10_max": 0.3753990828350857, "naucs_at_10_std": 0.17855691646833521, "naucs_at_10_diff1": 0.06742323097463344, "naucs_at_20_max": 0.47772828507795045, "naucs_at_20_std": 0.26324371619471887, "naucs_at_20_diff1": -0.07986000636334661, "naucs_at_100_max": 0.561391223155939, "naucs_at_100_std": 0.5401493930905731, "naucs_at_100_diff1": 0.41433239962652313, "naucs_at_1000_max": NaN, "naucs_at_1000_std": NaN, "naucs_at_1000_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_config.json

@@ -0,0 +1,144 @@
+{
+  "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,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "left",
+  "processor_class": "BiQwen2Processor",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

training_config.yml

@@ -0,0 +1,61 @@
+config:
+  (): colpali_engine.trainer.colmodel_training.ColModelTrainingConfig
+  output_dir: !path ../../../models/biqwen2-last
+  processor:
+    (): colpali_engine.utils.transformers_wrappers.AllPurposeWrapper
+    class_to_instanciate: !ext colpali_engine.models.BiQwen2Processor
+    pretrained_model_name_or_path:  "./models/colqwen2_base" # "./models/paligemma-3b-mix-448"
+    # max_length: 50
+
+  model:
+    (): colpali_engine.utils.transformers_wrappers.AllPurposeWrapper
+    class_to_instanciate: !ext colpali_engine.models.BiQwen2
+    pretrained_model_name_or_path: "./models/colqwen2_base"
+    torch_dtype:  !ext torch.bfloat16
+    use_cache: false
+#    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_detailed
+  eval_dataset_loader: !import ../data/test_data.yaml
+
+  # max_length: 50
+  run_eval: true
+  add_suffix: true
+  loss_func:
+    (): colpali_engine.loss.bi_encoder_losses.BiPairwiseCELoss
+  tr_args:
+    (): transformers.training_args.TrainingArguments
+    output_dir: null
+    overwrite_output_dir: true
+    num_train_epochs: 1
+    per_device_train_batch_size: 4
+    # 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: 500
+    learning_rate: 5e-5
+    save_total_limit: 1
+    # 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).*$|.*(custom_text_proj).*$)'
+