initial upload

README.md

@@ -0,0 +1,202 @@
+---
+base_model: liuhaotian/llava-v1.5-7b
+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 0.12.0

adapter_config.json

@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "liuhaotian/llava-v1.5-7b",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 256,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 128,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "down_proj",
+    "q_proj",
+    "k_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}

adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:afcc9dd6219dd4e03990d41d7adf04c1781ae9ccb04b7004512c76e33819e33b
+size 639692768

checkpoint-500/README.md

@@ -0,0 +1,202 @@
+---
+base_model: liuhaotian/llava-v1.5-7b
+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 0.12.0

checkpoint-500/adapter_config.json

@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "liuhaotian/llava-v1.5-7b",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 256,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 128,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "down_proj",
+    "q_proj",
+    "k_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}

checkpoint-500/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e0f1587acc7bfcf4ce7ff933de4aaa66c4bad91af226f117c13dcc937ec011b
+size 639692768

checkpoint-500/global_step500/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:915ae5376fe2f30965de47a0a4357122963d95eaac06e9a28ac33ad5d5d564ea
+size 1022392370

checkpoint-500/global_step500/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ae4b29d004bcea9a4cae3bd78dcd1fd469c6c1800ca999b1e0f1260208f40b28
+size 1022392370

checkpoint-500/global_step500/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9f5cbb69a0e15c47612833b62897d83b28a5ebbb5c17b3a2543eac80a97f0796
+size 1022392370

checkpoint-500/global_step500/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0cd82821cf0d2f5f37662845f6ddfbfde7927c2b9dc34e520b4e7260481f6f60
+size 1022392370

checkpoint-500/global_step500/zero_pp_rank_0_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ccefde6b62dd0262d1b0232e8e65d3a9a8a42d330a1a9a84b3d63c05c401d48f
+size 481762

checkpoint-500/global_step500/zero_pp_rank_1_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3aec879d2d8fe9a2bd9f008044539f4ba863ae4e333a94fe056387f929035505
+size 481762

checkpoint-500/global_step500/zero_pp_rank_2_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e48d1f4c2ce62bfa80a19f2973d3677921afcf3f3632691691e8c32047c18025
+size 481762

checkpoint-500/global_step500/zero_pp_rank_3_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac4d544af741690cac7330bbf778ff222a3a2190b42dab492970e4e55997213c
+size 481762

checkpoint-500/latest

@@ -0,0 +1 @@
+global_step500

checkpoint-500/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:82370032e137f3f55b0507a17d8f9a6ba3e128250291084d7134592c694d011f
+size 14960

checkpoint-500/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9831c46ec60268b43debb309dcdc1690fbf44f50726e5605cfd939ee9265e66d
+size 14960

checkpoint-500/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0bc8ec13e18a1ddde9e54fcc390912f7f549ad1ee5c0bc639e96249c36512207
+size 14960

checkpoint-500/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:363553c1cb5470a4810e404d999c9b3a2e91a32a825d15d41194739e610933d5
+size 14960

checkpoint-500/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:65938ca8bb140e91f88093d917db45c24de2ba39a5e01fc70571e5da0db42f8d
+size 1064

checkpoint-500/special_tokens_map.json

@@ -0,0 +1,24 @@
+{
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "<unk>",
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

checkpoint-500/tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
+size 499723

checkpoint-500/tokenizer_config.json

@@ -0,0 +1,43 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "add_prefix_space": true,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "</s>",
+  "legacy": false,
+  "model_max_length": 2048,
+  "pad_token": "<unk>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}

checkpoint-500/trainer_state.json

@@ -0,0 +1,3641 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.750593824228028,
+  "eval_steps": 100,
+  "global_step": 500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.008147055910771,
+      "learning_rate": 1.25e-05,
+      "loss": 4.2415,
+      "step": 1
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.04569203441769,
+      "learning_rate": 2.5e-05,
+      "loss": 4.3121,
+      "step": 2
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.865746651377984,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 4.3208,
+      "step": 3
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.6407193073379105,
+      "learning_rate": 5e-05,
+      "loss": 3.8848,
+      "step": 4
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.451159328560232,
+      "learning_rate": 6.25e-05,
+      "loss": 3.4391,
+      "step": 5
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.8259504797317525,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 3.0656,
+      "step": 6
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1881779175566867,
+      "learning_rate": 8.75e-05,
+      "loss": 2.8135,
+      "step": 7
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.614839668966139,
+      "learning_rate": 0.0001,
+      "loss": 2.7319,
+      "step": 8
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.5198673994210212,
+      "learning_rate": 0.00011250000000000001,
+      "loss": 2.6903,
+      "step": 9
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0044025931610727,
+      "learning_rate": 0.000125,
+      "loss": 2.584,
+      "step": 10
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.1531821793787296,
+      "learning_rate": 0.0001375,
+      "loss": 2.586,
+      "step": 11
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6210600474209341,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 2.5298,
+      "step": 12
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5025244204180619,
+      "learning_rate": 0.00016250000000000002,
+      "loss": 2.4665,
+      "step": 13
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5058788641352842,
+      "learning_rate": 0.000175,
+      "loss": 2.4194,
+      "step": 14
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.44571801666869537,
+      "learning_rate": 0.0001875,
+      "loss": 2.3531,
+      "step": 15
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.44028009268534757,
+      "learning_rate": 0.0002,
+      "loss": 2.2749,
+      "step": 16
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.42473118020142525,
+      "learning_rate": 0.00019999809527270051,
+      "loss": 2.2587,
+      "step": 17
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.465029302165452,
+      "learning_rate": 0.0001999923811633618,
+      "loss": 2.2196,
+      "step": 18
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.49040381415815754,
+      "learning_rate": 0.00019998285788966027,
+      "loss": 2.2061,
+      "step": 19
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4160855034634493,
+      "learning_rate": 0.00019996952581438068,
+      "loss": 2.1173,
+      "step": 20
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.45625369964232165,
+      "learning_rate": 0.00019995238544540241,
+      "loss": 2.1267,
+      "step": 21
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.42551849567803673,
+      "learning_rate": 0.00019993143743568,
+      "loss": 2.0976,
+      "step": 22
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5100052595965069,
+      "learning_rate": 0.0001999066825832184,
+      "loss": 2.0428,
+      "step": 23
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4717525078599394,
+      "learning_rate": 0.00019987812183104247,
+      "loss": 2.0068,
+      "step": 24
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.5596905853419681,
+      "learning_rate": 0.0001998457562671611,
+      "loss": 2.0303,
+      "step": 25
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4931645550169434,
+      "learning_rate": 0.00019980958712452577,
+      "loss": 1.9722,
+      "step": 26
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4433810930704678,
+      "learning_rate": 0.0001997696157809835,
+      "loss": 1.957,
+      "step": 27
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5522396650266582,
+      "learning_rate": 0.0001997258437592245,
+      "loss": 1.915,
+      "step": 28
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.49861222066728145,
+      "learning_rate": 0.00019967827272672408,
+      "loss": 1.8303,
+      "step": 29
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6169911964169147,
+      "learning_rate": 0.00019962690449567912,
+      "loss": 1.8454,
+      "step": 30
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5639780725078123,
+      "learning_rate": 0.000199571741022939,
+      "loss": 1.8068,
+      "step": 31
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6302805853808786,
+      "learning_rate": 0.0001995127844099313,
+      "loss": 1.7166,
+      "step": 32
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6494693483139545,
+      "learning_rate": 0.00019945003690258125,
+      "loss": 1.6433,
+      "step": 33
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.7598443409498918,
+      "learning_rate": 0.00019938350089122682,
+      "loss": 1.7081,
+      "step": 34
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6512764391881087,
+      "learning_rate": 0.00019931317891052708,
+      "loss": 1.6436,
+      "step": 35
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6953537359048508,
+      "learning_rate": 0.00019923907363936593,
+      "loss": 1.5862,
+      "step": 36
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6011387829084072,
+      "learning_rate": 0.00019916118790075008,
+      "loss": 1.5432,
+      "step": 37
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.659130437748028,
+      "learning_rate": 0.00019907952466170138,
+      "loss": 1.5132,
+      "step": 38
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7211467253555573,
+      "learning_rate": 0.00019899408703314385,
+      "loss": 1.506,
+      "step": 39
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7006890038987398,
+      "learning_rate": 0.0001989048782697851,
+      "loss": 1.4498,
+      "step": 40
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.64642158324997,
+      "learning_rate": 0.00019881190176999255,
+      "loss": 1.4478,
+      "step": 41
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6608085069521318,
+      "learning_rate": 0.00019871516107566366,
+      "loss": 1.3542,
+      "step": 42
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7707478188072372,
+      "learning_rate": 0.0001986146598720913,
+      "loss": 1.3309,
+      "step": 43
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8119298049916807,
+      "learning_rate": 0.00019851040198782326,
+      "loss": 1.345,
+      "step": 44
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7712308653234212,
+      "learning_rate": 0.0001984023913945162,
+      "loss": 1.3076,
+      "step": 45
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.682341709525683,
+      "learning_rate": 0.0001982906322067847,
+      "loss": 1.2565,
+      "step": 46
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7071991083514119,
+      "learning_rate": 0.00019817512868204425,
+      "loss": 1.1796,
+      "step": 47
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.745222014713615,
+      "learning_rate": 0.00019805588522034916,
+      "loss": 1.1649,
+      "step": 48
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7158459299510994,
+      "learning_rate": 0.00019793290636422505,
+      "loss": 1.2109,
+      "step": 49
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7335821144549012,
+      "learning_rate": 0.00019780619679849552,
+      "loss": 1.1475,
+      "step": 50
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7804306024320766,
+      "learning_rate": 0.000197675761350104,
+      "loss": 1.1068,
+      "step": 51
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8274924156959725,
+      "learning_rate": 0.00019754160498792965,
+      "loss": 1.1839,
+      "step": 52
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.8840482383868431,
+      "learning_rate": 0.0001974037328225982,
+      "loss": 1.0928,
+      "step": 53
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7224652999279871,
+      "learning_rate": 0.00019726215010628718,
+      "loss": 1.0299,
+      "step": 54
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7109288879933862,
+      "learning_rate": 0.0001971168622325259,
+      "loss": 1.0436,
+      "step": 55
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7650325966583326,
+      "learning_rate": 0.00019696787473598993,
+      "loss": 1.041,
+      "step": 56
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7307809391946058,
+      "learning_rate": 0.00019681519329229033,
+      "loss": 1.0195,
+      "step": 57
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6873943623441443,
+      "learning_rate": 0.00019665882371775733,
+      "loss": 0.972,
+      "step": 58
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8185924734616268,
+      "learning_rate": 0.00019649877196921896,
+      "loss": 0.9986,
+      "step": 59
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7907558585543373,
+      "learning_rate": 0.00019633504414377388,
+      "loss": 0.9201,
+      "step": 60
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7216280408288712,
+      "learning_rate": 0.00019616764647855926,
+      "loss": 0.9976,
+      "step": 61
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6946470891456141,
+      "learning_rate": 0.00019599658535051314,
+      "loss": 0.9008,
+      "step": 62
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6470248283451219,
+      "learning_rate": 0.00019582186727613152,
+      "loss": 0.8226,
+      "step": 63
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8297915622585336,
+      "learning_rate": 0.00019564349891122018,
+      "loss": 0.8825,
+      "step": 64
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7018515834126928,
+      "learning_rate": 0.00019546148705064097,
+      "loss": 0.8521,
+      "step": 65
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6119835758734723,
+      "learning_rate": 0.00019527583862805303,
+      "loss": 0.7872,
+      "step": 66
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6396036538427098,
+      "learning_rate": 0.00019508656071564882,
+      "loss": 0.7887,
+      "step": 67
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.6712059239435435,
+      "learning_rate": 0.00019489366052388441,
+      "loss": 0.8406,
+      "step": 68
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.6498227189328728,
+      "learning_rate": 0.00019469714540120507,
+      "loss": 0.7109,
+      "step": 69
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6950957852561941,
+      "learning_rate": 0.00019449702283376517,
+      "loss": 0.7008,
+      "step": 70
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6415745385783075,
+      "learning_rate": 0.00019429330044514305,
+      "loss": 0.6808,
+      "step": 71
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6774461765802887,
+      "learning_rate": 0.0001940859859960506,
+      "loss": 0.7122,
+      "step": 72
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.6335543398879422,
+      "learning_rate": 0.00019387508738403768,
+      "loss": 0.6826,
+      "step": 73
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6455659601218003,
+      "learning_rate": 0.0001936606126431911,
+      "loss": 0.7342,
+      "step": 74
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.6804108080708727,
+      "learning_rate": 0.00019344256994382878,
+      "loss": 0.6983,
+      "step": 75
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6233570198373359,
+      "learning_rate": 0.00019322096759218836,
+      "loss": 0.6426,
+      "step": 76
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6354196060962453,
+      "learning_rate": 0.00019299581403011082,
+      "loss": 0.6978,
+      "step": 77
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.6723728632702363,
+      "learning_rate": 0.0001927671178347189,
+      "loss": 0.6449,
+      "step": 78
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.6055794839258588,
+      "learning_rate": 0.00019253488771809024,
+      "loss": 0.6608,
+      "step": 79
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6032563228830964,
+      "learning_rate": 0.0001922991325269258,
+      "loss": 0.6691,
+      "step": 80
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5917538532836075,
+      "learning_rate": 0.00019205986124221251,
+      "loss": 0.6418,
+      "step": 81
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6558132078005496,
+      "learning_rate": 0.00019181708297888133,
+      "loss": 0.6562,
+      "step": 82
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.6110330049943966,
+      "learning_rate": 0.00019157080698546,
+      "loss": 0.5855,
+      "step": 83
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.6481622083495842,
+      "learning_rate": 0.00019132104264372063,
+      "loss": 0.628,
+      "step": 84
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5730813607452849,
+      "learning_rate": 0.0001910677994683225,
+      "loss": 0.5476,
+      "step": 85
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.6938507563801335,
+      "learning_rate": 0.00019081108710644932,
+      "loss": 0.6018,
+      "step": 86
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.625439427503205,
+      "learning_rate": 0.00019055091533744202,
+      "loss": 0.5735,
+      "step": 87
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.6628596764324554,
+      "learning_rate": 0.00019028729407242597,
+      "loss": 0.5389,
+      "step": 88
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6112099968245533,
+      "learning_rate": 0.00019002023335393364,
+      "loss": 0.5235,
+      "step": 89
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.6098216223216336,
+      "learning_rate": 0.0001897497433555218,
+      "loss": 0.6058,
+      "step": 90
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.6469247467013166,
+      "learning_rate": 0.0001894758343813842,
+      "loss": 0.5524,
+      "step": 91
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.6344920759870597,
+      "learning_rate": 0.00018919851686595874,
+      "loss": 0.5605,
+      "step": 92
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.6756355159547938,
+      "learning_rate": 0.00018891780137353034,
+      "loss": 0.5096,
+      "step": 93
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.6439314455537293,
+      "learning_rate": 0.00018863369859782825,
+      "loss": 0.5516,
+      "step": 94
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5567728554741562,
+      "learning_rate": 0.0001883462193616187,
+      "loss": 0.4576,
+      "step": 95
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.553595533418767,
+      "learning_rate": 0.00018805537461629265,
+      "loss": 0.4947,
+      "step": 96
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.6200223910647112,
+      "learning_rate": 0.00018776117544144863,
+      "loss": 0.5073,
+      "step": 97
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.6294322114297511,
+      "learning_rate": 0.00018746363304447073,
+      "loss": 0.4938,
+      "step": 98
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6000145257745209,
+      "learning_rate": 0.00018716275876010135,
+      "loss": 0.473,
+      "step": 99
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5927861897994469,
+      "learning_rate": 0.00018685856405000983,
+      "loss": 0.4724,
+      "step": 100
+    },
+    {
+      "epoch": 0.95,
+      "eval_blimp_filtered_avg": 0.7155223880597015,
+      "eval_blimp_filtered_std": 0.005000433138834185,
+      "step": 100
+    },
+    {
+      "epoch": 0.95,
+      "eval_blimp_supplement_avg": 0.8405172413793104,
+      "eval_blimp_supplement_std": 0.016486001732879434,
+      "step": 100
+    },
+    {
+      "epoch": 0.95,
+      "eval_vqa_filtered_avg": 0.52,
+      "eval_vqa_filtered_std": 0.05021167315686779,
+      "step": 100
+    },
+    {
+      "epoch": 0.95,
+      "eval_winoground_filtered_avg": 0.64,
+      "eval_winoground_filtered_std": 0.04824181513244218,
+      "step": 100
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5504516732077648,
+      "learning_rate": 0.00018655106050235548,
+      "loss": 0.4393,
+      "step": 101
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5801589113252366,
+      "learning_rate": 0.00018624025983134644,
+      "loss": 0.468,
+      "step": 102
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5273944337529535,
+      "learning_rate": 0.00018592617387679306,
+      "loss": 0.439,
+      "step": 103
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.508609381383424,
+      "learning_rate": 0.00018560881460365724,
+      "loss": 0.4272,
+      "step": 104
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5396859577867195,
+      "learning_rate": 0.0001852881941015964,
+      "loss": 0.4362,
+      "step": 105
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5122858999271028,
+      "learning_rate": 0.00018496432458450294,
+      "loss": 0.3893,
+      "step": 106
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.49626561438760436,
+      "learning_rate": 0.00018463721839003915,
+      "loss": 0.3498,
+      "step": 107
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.48748413013987063,
+      "learning_rate": 0.000184306887979167,
+      "loss": 0.3256,
+      "step": 108
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5310280563857716,
+      "learning_rate": 0.00018397334593567348,
+      "loss": 0.3225,
+      "step": 109
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.6232514021230662,
+      "learning_rate": 0.00018363660496569127,
+      "loss": 0.3489,
+      "step": 110
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5274577320762,
+      "learning_rate": 0.00018329667789721485,
+      "loss": 0.3123,
+      "step": 111
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5096311315676365,
+      "learning_rate": 0.00018295357767961144,
+      "loss": 0.3325,
+      "step": 112
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.4613577097438129,
+      "learning_rate": 0.00018260731738312818,
+      "loss": 0.2936,
+      "step": 113
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.4997938044342101,
+      "learning_rate": 0.00018225791019839375,
+      "loss": 0.3351,
+      "step": 114
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.538085494988463,
+      "learning_rate": 0.00018190536943591624,
+      "loss": 0.329,
+      "step": 115
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5567068979809859,
+      "learning_rate": 0.00018154970852557603,
+      "loss": 0.318,
+      "step": 116
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5548141608588357,
+      "learning_rate": 0.0001811909410161139,
+      "loss": 0.3289,
+      "step": 117
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.47326466614968965,
+      "learning_rate": 0.0001808290805746153,
+      "loss": 0.3076,
+      "step": 118
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.47629585466918467,
+      "learning_rate": 0.00018046414098598948,
+      "loss": 0.3016,
+      "step": 119
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.44135735344426463,
+      "learning_rate": 0.00018009613615244436,
+      "loss": 0.2704,
+      "step": 120
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5127645747027901,
+      "learning_rate": 0.000179725080092957,
+      "loss": 0.2887,
+      "step": 121
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5209981172771183,
+      "learning_rate": 0.0001793509869427395,
+      "loss": 0.2938,
+      "step": 122
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5481082193558409,
+      "learning_rate": 0.00017897387095270058,
+      "loss": 0.3191,
+      "step": 123
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.4770065158307258,
+      "learning_rate": 0.0001785937464889027,
+      "loss": 0.2795,
+      "step": 124
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.44845204938493194,
+      "learning_rate": 0.0001782106280320147,
+      "loss": 0.2667,
+      "step": 125
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.47824147005907164,
+      "learning_rate": 0.00017782453017676025,
+      "loss": 0.267,
+      "step": 126
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.501015317452837,
+      "learning_rate": 0.00017743546763136187,
+      "loss": 0.2831,
+      "step": 127
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5232536606095718,
+      "learning_rate": 0.00017704345521698058,
+      "loss": 0.2769,
+      "step": 128
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5495388553709665,
+      "learning_rate": 0.00017664850786715136,
+      "loss": 0.3031,
+      "step": 129
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5371555106361774,
+      "learning_rate": 0.00017625064062721415,
+      "loss": 0.2955,
+      "step": 130
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.4716773551397148,
+      "learning_rate": 0.00017584986865374082,
+      "loss": 0.2666,
+      "step": 131
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5089124561646106,
+      "learning_rate": 0.00017544620721395777,
+      "loss": 0.3379,
+      "step": 132
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.4715340007422714,
+      "learning_rate": 0.00017503967168516426,
+      "loss": 0.2771,
+      "step": 133
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.43502563576445413,
+      "learning_rate": 0.0001746302775541467,
+      "loss": 0.2423,
+      "step": 134
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.4967705692007805,
+      "learning_rate": 0.00017421804041658863,
+      "loss": 0.2498,
+      "step": 135
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.49127370733051945,
+      "learning_rate": 0.00017380297597647667,
+      "loss": 0.2616,
+      "step": 136
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.47835649282708265,
+      "learning_rate": 0.00017338510004550223,
+      "loss": 0.241,
+      "step": 137
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.4843464174553606,
+      "learning_rate": 0.00017296442854245915,
+      "loss": 0.2458,
+      "step": 138
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5209405133977896,
+      "learning_rate": 0.00017254097749263734,
+      "loss": 0.2452,
+      "step": 139
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.4709574288825739,
+      "learning_rate": 0.0001721147630272123,
+      "loss": 0.2627,
+      "step": 140
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.4752105435022234,
+      "learning_rate": 0.00017168580138263062,
+      "loss": 0.2527,
+      "step": 141
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.48781843284289905,
+      "learning_rate": 0.00017125410889999134,
+      "loss": 0.2356,
+      "step": 142
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5731736183258567,
+      "learning_rate": 0.00017081970202442362,
+      "loss": 0.2668,
+      "step": 143
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.48105126464697834,
+      "learning_rate": 0.0001703825973044602,
+      "loss": 0.2454,
+      "step": 144
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5280645599674879,
+      "learning_rate": 0.00016994281139140688,
+      "loss": 0.2454,
+      "step": 145
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.47876489284248624,
+      "learning_rate": 0.0001695003610387084,
+      "loss": 0.2463,
+      "step": 146
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.48826354198860017,
+      "learning_rate": 0.00016905526310130999,
+      "loss": 0.2295,
+      "step": 147
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.47715494831436517,
+      "learning_rate": 0.0001686075345350156,
+      "loss": 0.252,
+      "step": 148
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.5152105233009641,
+      "learning_rate": 0.0001681571923958416,
+      "loss": 0.2771,
+      "step": 149
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.4990883717055415,
+      "learning_rate": 0.00016770425383936735,
+      "loss": 0.2497,
+      "step": 150
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.4674093996422124,
+      "learning_rate": 0.00016724873612008155,
+      "loss": 0.2441,
+      "step": 151
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.4432102664091143,
+      "learning_rate": 0.00016679065659072487,
+      "loss": 0.2418,
+      "step": 152
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.4677926556162063,
+      "learning_rate": 0.00016633003270162902,
+      "loss": 0.2483,
+      "step": 153
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.5050389021999718,
+      "learning_rate": 0.00016586688200005193,
+      "loss": 0.225,
+      "step": 154
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.538150442089787,
+      "learning_rate": 0.00016540122212950934,
+      "loss": 0.2629,
+      "step": 155
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.4831894197759429,
+      "learning_rate": 0.00016493307082910249,
+      "loss": 0.2539,
+      "step": 156
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.4864294249801108,
+      "learning_rate": 0.00016446244593284277,
+      "loss": 0.2638,
+      "step": 157
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.46236092553249764,
+      "learning_rate": 0.00016398936536897183,
+      "loss": 0.2255,
+      "step": 158
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.4963120760517666,
+      "learning_rate": 0.00016351384715927898,
+      "loss": 0.2524,
+      "step": 159
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.5210286477375989,
+      "learning_rate": 0.00016303590941841458,
+      "loss": 0.225,
+      "step": 160
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.5288475623534257,
+      "learning_rate": 0.0001625555703531998,
+      "loss": 0.2428,
+      "step": 161
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.4973215047467683,
+      "learning_rate": 0.00016207284826193335,
+      "loss": 0.2522,
+      "step": 162
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.44826317640998203,
+      "learning_rate": 0.00016158776153369402,
+      "loss": 0.2019,
+      "step": 163
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.45392654459830534,
+      "learning_rate": 0.0001611003286476406,
+      "loss": 0.2338,
+      "step": 164
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.4430521150056381,
+      "learning_rate": 0.00016061056817230754,
+      "loss": 0.2273,
+      "step": 165
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.44345119147374473,
+      "learning_rate": 0.00016011849876489776,
+      "loss": 0.211,
+      "step": 166
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4808061249544928,
+      "learning_rate": 0.000159624139170572,
+      "loss": 0.2104,
+      "step": 167
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.5573402749682285,
+      "learning_rate": 0.00015912750822173445,
+      "loss": 0.2492,
+      "step": 168
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.5334950652460796,
+      "learning_rate": 0.00015862862483731574,
+      "loss": 0.2187,
+      "step": 169
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.49497739813798797,
+      "learning_rate": 0.00015812750802205187,
+      "loss": 0.2097,
+      "step": 170
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.44446540691990566,
+      "learning_rate": 0.00015762417686576038,
+      "loss": 0.204,
+      "step": 171
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.42142200135464725,
+      "learning_rate": 0.0001571186505426132,
+      "loss": 0.1989,
+      "step": 172
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.4328533901196503,
+      "learning_rate": 0.00015661094831040598,
+      "loss": 0.2173,
+      "step": 173
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.43093996542664664,
+      "learning_rate": 0.00015610108950982494,
+      "loss": 0.1865,
+      "step": 174
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.4850613308932528,
+      "learning_rate": 0.00015558909356370944,
+      "loss": 0.2181,
+      "step": 175
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.47485870685329246,
+      "learning_rate": 0.00015507497997631266,
+      "loss": 0.2223,
+      "step": 176
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.42085147271583295,
+      "learning_rate": 0.0001545587683325583,
+      "loss": 0.1845,
+      "step": 177
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.4479801309419239,
+      "learning_rate": 0.00015404047829729457,
+      "loss": 0.1987,
+      "step": 178
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.4624584058381783,
+      "learning_rate": 0.00015352012961454507,
+      "loss": 0.217,
+      "step": 179
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.44005765649196454,
+      "learning_rate": 0.00015299774210675657,
+      "loss": 0.1837,
+      "step": 180
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.4508346255489124,
+      "learning_rate": 0.00015247333567404406,
+      "loss": 0.2007,
+      "step": 181
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.40396006791211914,
+      "learning_rate": 0.00015194693029343248,
+      "loss": 0.1866,
+      "step": 182
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.44558839018398966,
+      "learning_rate": 0.00015141854601809581,
+      "loss": 0.1967,
+      "step": 183
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.4337334328022437,
+      "learning_rate": 0.00015088820297659314,
+      "loss": 0.1891,
+      "step": 184
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.4636781912221849,
+      "learning_rate": 0.00015035592137210187,
+      "loss": 0.193,
+      "step": 185
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.47955885394967973,
+      "learning_rate": 0.00014982172148164804,
+      "loss": 0.1793,
+      "step": 186
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.4721310395975314,
+      "learning_rate": 0.00014928562365533392,
+      "loss": 0.186,
+      "step": 187
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.4737141537120664,
+      "learning_rate": 0.00014874764831556285,
+      "loss": 0.2058,
+      "step": 188
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.40830849621087567,
+      "learning_rate": 0.00014820781595626116,
+      "loss": 0.1822,
+      "step": 189
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.4272142710058541,
+      "learning_rate": 0.0001476661471420975,
+      "loss": 0.2057,
+      "step": 190
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.4212227727031309,
+      "learning_rate": 0.0001471226625076993,
+      "loss": 0.1845,
+      "step": 191
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.39660108389275345,
+      "learning_rate": 0.0001465773827568671,
+      "loss": 0.1769,
+      "step": 192
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.38828383424285384,
+      "learning_rate": 0.00014603032866178538,
+      "loss": 0.1699,
+      "step": 193
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.3681031142044674,
+      "learning_rate": 0.00014548152106223157,
+      "loss": 0.1456,
+      "step": 194
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.46248659870169556,
+      "learning_rate": 0.00014493098086478196,
+      "loss": 0.1846,
+      "step": 195
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.4437664820090981,
+      "learning_rate": 0.00014437872904201542,
+      "loss": 0.1706,
+      "step": 196
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.4410375026146085,
+      "learning_rate": 0.0001438247866317145,
+      "loss": 0.1757,
+      "step": 197
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.4290870801703047,
+      "learning_rate": 0.00014326917473606366,
+      "loss": 0.1777,
+      "step": 198
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.4812130220306999,
+      "learning_rate": 0.00014271191452084597,
+      "loss": 0.2013,
+      "step": 199
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.4314920290891278,
+      "learning_rate": 0.00014215302721463623,
+      "loss": 0.1857,
+      "step": 200
+    },
+    {
+      "epoch": 1.9,
+      "eval_blimp_filtered_avg": 0.7161194029850746,
+      "eval_blimp_filtered_std": 0.005001692965803923,
+      "step": 200
+    },
+    {
+      "epoch": 1.9,
+      "eval_blimp_supplement_avg": 0.8211206896551724,
+      "eval_blimp_supplement_std": 0.016785621805327337,
+      "step": 200
+    },
+    {
+      "epoch": 1.9,
+      "eval_vqa_filtered_avg": 0.51,
+      "eval_vqa_filtered_std": 0.05024183937956912,
+      "step": 200
+    },
+    {
+      "epoch": 1.9,
+      "eval_winoground_filtered_avg": 0.62,
+      "eval_winoground_filtered_std": 0.04878317312145633,
+      "step": 200
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.41562514975066434,
+      "learning_rate": 0.0001415925341079927,
+      "loss": 0.21,
+      "step": 201
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.37833993286875955,
+      "learning_rate": 0.00014103045655264576,
+      "loss": 0.1659,
+      "step": 202
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.3880529818353851,
+      "learning_rate": 0.00014046681596068466,
+      "loss": 0.1638,
+      "step": 203
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.40159118156434603,
+      "learning_rate": 0.00013990163380374194,
+      "loss": 0.1768,
+      "step": 204
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.4086449128732129,
+      "learning_rate": 0.00013933493161217523,
+      "loss": 0.1544,
+      "step": 205
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3808287729283849,
+      "learning_rate": 0.0001387667309742472,
+      "loss": 0.1366,
+      "step": 206
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.39609061286446773,
+      "learning_rate": 0.0001381970535353032,
+      "loss": 0.1494,
+      "step": 207
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.40847272653729905,
+      "learning_rate": 0.00013762592099694665,
+      "loss": 0.1615,
+      "step": 208
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.4334994696681873,
+      "learning_rate": 0.00013705335511621228,
+      "loss": 0.1542,
+      "step": 209
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.4546384761691546,
+      "learning_rate": 0.00013647937770473737,
+      "loss": 0.1834,
+      "step": 210
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.36130610610645814,
+      "learning_rate": 0.00013590401062793083,
+      "loss": 0.123,
+      "step": 211
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.29975302946848653,
+      "learning_rate": 0.0001353272758041402,
+      "loss": 0.0824,
+      "step": 212
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.29392603086414587,
+      "learning_rate": 0.00013474919520381671,
+      "loss": 0.0836,
+      "step": 213
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.33169221984700814,
+      "learning_rate": 0.00013416979084867852,
+      "loss": 0.0683,
+      "step": 214
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.39192700338704206,
+      "learning_rate": 0.00013358908481087134,
+      "loss": 0.0804,
+      "step": 215
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.42443737109460977,
+      "learning_rate": 0.0001330070992121281,
+      "loss": 0.0797,
+      "step": 216
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.42848813761714244,
+      "learning_rate": 0.00013242385622292592,
+      "loss": 0.0776,
+      "step": 217
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.37448633759803696,
+      "learning_rate": 0.00013183937806164172,
+      "loss": 0.0739,
+      "step": 218
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.3437440816482259,
+      "learning_rate": 0.00013125368699370567,
+      "loss": 0.0652,
+      "step": 219
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.356415907025676,
+      "learning_rate": 0.0001306668053307531,
+      "loss": 0.0778,
+      "step": 220
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.30675625825005026,
+      "learning_rate": 0.00013007875542977448,
+      "loss": 0.0665,
+      "step": 221
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.29794655672460485,
+      "learning_rate": 0.00012948955969226383,
+      "loss": 0.0696,
+      "step": 222
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.30163505061461343,
+      "learning_rate": 0.00012889924056336532,
+      "loss": 0.0705,
+      "step": 223
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.32541739323213426,
+      "learning_rate": 0.00012830782053101805,
+      "loss": 0.0733,
+      "step": 224
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.31121536090331003,
+      "learning_rate": 0.00012771532212509974,
+      "loss": 0.0711,
+      "step": 225
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.34593292210442944,
+      "learning_rate": 0.00012712176791656807,
+      "loss": 0.0788,
+      "step": 226
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.33946278651997686,
+      "learning_rate": 0.0001265271805166012,
+      "loss": 0.0677,
+      "step": 227
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.3400898219352628,
+      "learning_rate": 0.0001259315825757362,
+      "loss": 0.0643,
+      "step": 228
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.3813085350755264,
+      "learning_rate": 0.00012533499678300618,
+      "loss": 0.0761,
+      "step": 229
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.3523012248149677,
+      "learning_rate": 0.00012473744586507604,
+      "loss": 0.0648,
+      "step": 230
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.37842862853695125,
+      "learning_rate": 0.00012413895258537675,
+      "loss": 0.0812,
+      "step": 231
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.39475455813661525,
+      "learning_rate": 0.00012353953974323807,
+      "loss": 0.0801,
+      "step": 232
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.3205081471986943,
+      "learning_rate": 0.00012293923017302002,
+      "loss": 0.0677,
+      "step": 233
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.31006899448135294,
+      "learning_rate": 0.0001223380467432432,
+      "loss": 0.07,
+      "step": 234
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.3048520942780853,
+      "learning_rate": 0.00012173601235571742,
+      "loss": 0.0615,
+      "step": 235
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.3425413653893973,
+      "learning_rate": 0.0001211331499446693,
+      "loss": 0.0658,
+      "step": 236
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.31929344956491607,
+      "learning_rate": 0.00012052948247586873,
+      "loss": 0.0653,
+      "step": 237
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.3414359773691709,
+      "learning_rate": 0.00011992503294575383,
+      "loss": 0.0723,
+      "step": 238
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.32978160245312554,
+      "learning_rate": 0.00011931982438055505,
+      "loss": 0.07,
+      "step": 239
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.33271868205929617,
+      "learning_rate": 0.00011871387983541789,
+      "loss": 0.0672,
+      "step": 240
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.29862145989444433,
+      "learning_rate": 0.00011810722239352467,
+      "loss": 0.0603,
+      "step": 241
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.34485364985513034,
+      "learning_rate": 0.00011749987516521523,
+      "loss": 0.0632,
+      "step": 242
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.3299899118013224,
+      "learning_rate": 0.00011689186128710654,
+      "loss": 0.0601,
+      "step": 243
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.29635972892096896,
+      "learning_rate": 0.00011628320392121117,
+      "loss": 0.0558,
+      "step": 244
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.3414458592363874,
+      "learning_rate": 0.0001156739262540552,
+      "loss": 0.0703,
+      "step": 245
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.3280087622706941,
+      "learning_rate": 0.00011506405149579468,
+      "loss": 0.0657,
+      "step": 246
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.373086375777386,
+      "learning_rate": 0.00011445360287933165,
+      "loss": 0.0668,
+      "step": 247
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.2937645914714354,
+      "learning_rate": 0.00011384260365942904,
+      "loss": 0.0612,
+      "step": 248
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.39022311054047737,
+      "learning_rate": 0.00011323107711182473,
+      "loss": 0.0762,
+      "step": 249
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.3345521008714258,
+      "learning_rate": 0.00011261904653234485,
+      "loss": 0.0711,
+      "step": 250
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.30608871062806836,
+      "learning_rate": 0.00011200653523601652,
+      "loss": 0.0617,
+      "step": 251
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.30714147902477945,
+      "learning_rate": 0.00011139356655617945,
+      "loss": 0.063,
+      "step": 252
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.31051190204375445,
+      "learning_rate": 0.00011078016384359724,
+      "loss": 0.0659,
+      "step": 253
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.3071085278813772,
+      "learning_rate": 0.00011016635046556772,
+      "loss": 0.061,
+      "step": 254
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.3045837343462885,
+      "learning_rate": 0.00010955214980503284,
+      "loss": 0.0597,
+      "step": 255
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.3049959198680976,
+      "learning_rate": 0.00010893758525968789,
+      "loss": 0.0587,
+      "step": 256
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.3168437149994661,
+      "learning_rate": 0.00010832268024109025,
+      "loss": 0.0559,
+      "step": 257
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.3024342626013227,
+      "learning_rate": 0.00010770745817376742,
+      "loss": 0.0583,
+      "step": 258
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3188509232471995,
+      "learning_rate": 0.0001070919424943247,
+      "loss": 0.061,
+      "step": 259
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.3381945814712772,
+      "learning_rate": 0.0001064761566505525,
+      "loss": 0.0648,
+      "step": 260
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.3131931451431926,
+      "learning_rate": 0.00010586012410053292,
+      "loss": 0.0624,
+      "step": 261
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.32809637984753304,
+      "learning_rate": 0.00010524386831174628,
+      "loss": 0.0627,
+      "step": 262
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.2832796499168925,
+      "learning_rate": 0.00010462741276017711,
+      "loss": 0.0535,
+      "step": 263
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.3334141162384235,
+      "learning_rate": 0.00010401078092941971,
+      "loss": 0.061,
+      "step": 264
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.27653747850590626,
+      "learning_rate": 0.00010339399630978373,
+      "loss": 0.0497,
+      "step": 265
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.32205480409336124,
+      "learning_rate": 0.00010277708239739924,
+      "loss": 0.0658,
+      "step": 266
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.310079147965717,
+      "learning_rate": 0.0001021600626933217,
+      "loss": 0.0525,
+      "step": 267
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.31094425691461797,
+      "learning_rate": 0.00010154296070263649,
+      "loss": 0.0619,
+      "step": 268
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.33419799536496597,
+      "learning_rate": 0.00010092579993356386,
+      "loss": 0.0615,
+      "step": 269
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.3343121767672678,
+      "learning_rate": 0.00010030860389656305,
+      "loss": 0.0663,
+      "step": 270
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.3516117623617434,
+      "learning_rate": 9.969139610343696e-05,
+      "loss": 0.0662,
+      "step": 271
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.31796912631433194,
+      "learning_rate": 9.907420006643619e-05,
+      "loss": 0.0624,
+      "step": 272
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.29219460425245597,
+      "learning_rate": 9.845703929736351e-05,
+      "loss": 0.0596,
+      "step": 273
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.316635170830544,
+      "learning_rate": 9.783993730667831e-05,
+      "loss": 0.0659,
+      "step": 274
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.33766616368603597,
+      "learning_rate": 9.722291760260077e-05,
+      "loss": 0.0646,
+      "step": 275
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.31287192455811574,
+      "learning_rate": 9.66060036902163e-05,
+      "loss": 0.0585,
+      "step": 276
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.28964582015181484,
+      "learning_rate": 9.598921907058033e-05,
+      "loss": 0.0543,
+      "step": 277
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.3037919396698326,
+      "learning_rate": 9.53725872398229e-05,
+      "loss": 0.0512,
+      "step": 278
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.3229974938313004,
+      "learning_rate": 9.475613168825374e-05,
+      "loss": 0.0531,
+      "step": 279
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.29881091304580676,
+      "learning_rate": 9.413987589946711e-05,
+      "loss": 0.0569,
+      "step": 280
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.29692909307641674,
+      "learning_rate": 9.352384334944753e-05,
+      "loss": 0.0547,
+      "step": 281
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.33439942628885455,
+      "learning_rate": 9.290805750567532e-05,
+      "loss": 0.0622,
+      "step": 282
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.2991141437988068,
+      "learning_rate": 9.22925418262326e-05,
+      "loss": 0.0464,
+      "step": 283
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.3171911760038229,
+      "learning_rate": 9.167731975890976e-05,
+      "loss": 0.059,
+      "step": 284
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.30072460150102115,
+      "learning_rate": 9.106241474031212e-05,
+      "loss": 0.0559,
+      "step": 285
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.3301896190647226,
+      "learning_rate": 9.04478501949672e-05,
+      "loss": 0.0514,
+      "step": 286
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.3298071637508188,
+      "learning_rate": 8.983364953443227e-05,
+      "loss": 0.0618,
+      "step": 287
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.3497185839244567,
+      "learning_rate": 8.921983615640277e-05,
+      "loss": 0.065,
+      "step": 288
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.33084725547728233,
+      "learning_rate": 8.860643344382056e-05,
+      "loss": 0.0527,
+      "step": 289
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.33012822636415956,
+      "learning_rate": 8.79934647639835e-05,
+      "loss": 0.0666,
+      "step": 290
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.3151687548518561,
+      "learning_rate": 8.738095346765518e-05,
+      "loss": 0.0573,
+      "step": 291
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.30346203875619676,
+      "learning_rate": 8.676892288817531e-05,
+      "loss": 0.0491,
+      "step": 292
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.3133369298353677,
+      "learning_rate": 8.615739634057098e-05,
+      "loss": 0.0595,
+      "step": 293
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.28715782085999497,
+      "learning_rate": 8.554639712066836e-05,
+      "loss": 0.0542,
+      "step": 294
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.2815995010771035,
+      "learning_rate": 8.493594850420537e-05,
+      "loss": 0.0551,
+      "step": 295
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.280576878443274,
+      "learning_rate": 8.432607374594484e-05,
+      "loss": 0.0488,
+      "step": 296
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.298809991890747,
+      "learning_rate": 8.371679607878884e-05,
+      "loss": 0.0544,
+      "step": 297
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.30088222272143067,
+      "learning_rate": 8.310813871289348e-05,
+      "loss": 0.0591,
+      "step": 298
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.3237358977236424,
+      "learning_rate": 8.250012483478478e-05,
+      "loss": 0.0547,
+      "step": 299
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.34075237005827885,
+      "learning_rate": 8.189277760647537e-05,
+      "loss": 0.0566,
+      "step": 300
+    },
+    {
+      "epoch": 2.85,
+      "eval_blimp_filtered_avg": 0.7037313432835821,
+      "eval_blimp_filtered_std": 0.005058972315437875,
+      "step": 300
+    },
+    {
+      "epoch": 2.85,
+      "eval_blimp_supplement_avg": 0.8103448275862069,
+      "eval_blimp_supplement_std": 0.017321145118445798,
+      "step": 300
+    },
+    {
+      "epoch": 2.85,
+      "eval_vqa_filtered_avg": 0.53,
+      "eval_vqa_filtered_std": 0.0501613558046592,
+      "step": 300
+    },
+    {
+      "epoch": 2.85,
+      "eval_winoground_filtered_avg": 0.68,
+      "eval_winoground_filtered_std": 0.046882617226215034,
+      "step": 300
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.3237263865460515,
+      "learning_rate": 8.128612016458215e-05,
+      "loss": 0.059,
+      "step": 301
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.2977357286247905,
+      "learning_rate": 8.068017561944499e-05,
+      "loss": 0.0492,
+      "step": 302
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.29591506818063545,
+      "learning_rate": 8.00749670542462e-05,
+      "loss": 0.052,
+      "step": 303
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.2789469075911483,
+      "learning_rate": 7.94705175241313e-05,
+      "loss": 0.0455,
+      "step": 304
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.2997082343784124,
+      "learning_rate": 7.886685005533072e-05,
+      "loss": 0.0498,
+      "step": 305
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.30157528073661777,
+      "learning_rate": 7.82639876442826e-05,
+      "loss": 0.0567,
+      "step": 306
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.32803298910194756,
+      "learning_rate": 7.76619532567568e-05,
+      "loss": 0.0622,
+      "step": 307
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.28556449374878695,
+      "learning_rate": 7.706076982697999e-05,
+      "loss": 0.0489,
+      "step": 308
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.32287162854623286,
+      "learning_rate": 7.646046025676198e-05,
+      "loss": 0.066,
+      "step": 309
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.3384064716667544,
+      "learning_rate": 7.586104741462325e-05,
+      "loss": 0.0629,
+      "step": 310
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.3005901634146794,
+      "learning_rate": 7.526255413492395e-05,
+      "loss": 0.051,
+      "step": 311
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.2907146546357962,
+      "learning_rate": 7.466500321699383e-05,
+      "loss": 0.0546,
+      "step": 312
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.30779520364750435,
+      "learning_rate": 7.40684174242638e-05,
+      "loss": 0.058,
+      "step": 313
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.29074373091101263,
+      "learning_rate": 7.347281948339879e-05,
+      "loss": 0.0463,
+      "step": 314
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 0.32970798475445445,
+      "learning_rate": 7.287823208343192e-05,
+      "loss": 0.0589,
+      "step": 315
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.2798345327195924,
+      "learning_rate": 7.228467787490028e-05,
+      "loss": 0.0438,
+      "step": 316
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.18326848967204043,
+      "learning_rate": 7.169217946898197e-05,
+      "loss": 0.0225,
+      "step": 317
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 0.18022372679373735,
+      "learning_rate": 7.110075943663472e-05,
+      "loss": 0.0161,
+      "step": 318
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.1633153575928502,
+      "learning_rate": 7.051044030773618e-05,
+      "loss": 0.0153,
+      "step": 319
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 0.17802284328446474,
+      "learning_rate": 6.992124457022553e-05,
+      "loss": 0.0176,
+      "step": 320
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.17359891604740127,
+      "learning_rate": 6.933319466924693e-05,
+      "loss": 0.0162,
+      "step": 321
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 0.2202987501804585,
+      "learning_rate": 6.874631300629435e-05,
+      "loss": 0.0162,
+      "step": 322
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 0.22277821921264357,
+      "learning_rate": 6.81606219383583e-05,
+      "loss": 0.0187,
+      "step": 323
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.18724963681022663,
+      "learning_rate": 6.757614377707409e-05,
+      "loss": 0.0153,
+      "step": 324
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 0.21995220887794256,
+      "learning_rate": 6.699290078787193e-05,
+      "loss": 0.0188,
+      "step": 325
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.1967935793635855,
+      "learning_rate": 6.641091518912867e-05,
+      "loss": 0.0156,
+      "step": 326
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 0.20661934683104752,
+      "learning_rate": 6.583020915132152e-05,
+      "loss": 0.0158,
+      "step": 327
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.2422474266231083,
+      "learning_rate": 6.525080479618331e-05,
+      "loss": 0.0177,
+      "step": 328
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 0.18354685059507367,
+      "learning_rate": 6.467272419585984e-05,
+      "loss": 0.013,
+      "step": 329
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.22423754187379397,
+      "learning_rate": 6.40959893720692e-05,
+      "loss": 0.0188,
+      "step": 330
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.18994008796265852,
+      "learning_rate": 6.352062229526266e-05,
+      "loss": 0.0132,
+      "step": 331
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 0.24715301748493912,
+      "learning_rate": 6.294664488378776e-05,
+      "loss": 0.015,
+      "step": 332
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 0.17280498203848704,
+      "learning_rate": 6.237407900305335e-05,
+      "loss": 0.0138,
+      "step": 333
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.21773200395950232,
+      "learning_rate": 6.180294646469679e-05,
+      "loss": 0.0155,
+      "step": 334
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 0.2144971485793242,
+      "learning_rate": 6.123326902575282e-05,
+      "loss": 0.0158,
+      "step": 335
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.18331926033535073,
+      "learning_rate": 6.06650683878248e-05,
+      "loss": 0.013,
+      "step": 336
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.1788180130126268,
+      "learning_rate": 6.009836619625809e-05,
+      "loss": 0.0133,
+      "step": 337
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.20337677688861636,
+      "learning_rate": 5.953318403931532e-05,
+      "loss": 0.0129,
+      "step": 338
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 0.20853998405220736,
+      "learning_rate": 5.896954344735426e-05,
+      "loss": 0.0176,
+      "step": 339
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 0.1919639102705018,
+      "learning_rate": 5.840746589200732e-05,
+      "loss": 0.0144,
+      "step": 340
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.2134469059873606,
+      "learning_rate": 5.784697278536379e-05,
+      "loss": 0.0138,
+      "step": 341
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 0.18435084201272836,
+      "learning_rate": 5.728808547915405e-05,
+      "loss": 0.0135,
+      "step": 342
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.19554570393158438,
+      "learning_rate": 5.673082526393634e-05,
+      "loss": 0.015,
+      "step": 343
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 0.18522448379098544,
+      "learning_rate": 5.617521336828556e-05,
+      "loss": 0.0129,
+      "step": 344
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 0.190207008998555,
+      "learning_rate": 5.5621270957984573e-05,
+      "loss": 0.0161,
+      "step": 345
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 0.19594053008897275,
+      "learning_rate": 5.506901913521808e-05,
+      "loss": 0.0162,
+      "step": 346
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.20111569255746164,
+      "learning_rate": 5.451847893776845e-05,
+      "loss": 0.0147,
+      "step": 347
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 0.20867562278084897,
+      "learning_rate": 5.396967133821461e-05,
+      "loss": 0.0154,
+      "step": 348
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 0.16028325232055693,
+      "learning_rate": 5.342261724313292e-05,
+      "loss": 0.0117,
+      "step": 349
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.14992620939570764,
+      "learning_rate": 5.28773374923007e-05,
+      "loss": 0.0106,
+      "step": 350
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.20669460754401175,
+      "learning_rate": 5.2333852857902575e-05,
+      "loss": 0.0161,
+      "step": 351
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 0.21934716169620833,
+      "learning_rate": 5.1792184043738855e-05,
+      "loss": 0.0128,
+      "step": 352
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.18204794157825063,
+      "learning_rate": 5.1252351684437136e-05,
+      "loss": 0.0129,
+      "step": 353
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 0.21363608639584963,
+      "learning_rate": 5.071437634466609e-05,
+      "loss": 0.0105,
+      "step": 354
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 0.15881770971724649,
+      "learning_rate": 5.0178278518351983e-05,
+      "loss": 0.0096,
+      "step": 355
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 0.1980006966366768,
+      "learning_rate": 4.964407862789817e-05,
+      "loss": 0.0119,
+      "step": 356
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 0.21004802159627842,
+      "learning_rate": 4.911179702340688e-05,
+      "loss": 0.0119,
+      "step": 357
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 0.20419756258161648,
+      "learning_rate": 4.85814539819042e-05,
+      "loss": 0.0145,
+      "step": 358
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 0.1565818058300373,
+      "learning_rate": 4.8053069706567554e-05,
+      "loss": 0.0105,
+      "step": 359
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 0.19501698471957343,
+      "learning_rate": 4.752666432595596e-05,
+      "loss": 0.0126,
+      "step": 360
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 0.20941486180216556,
+      "learning_rate": 4.700225789324343e-05,
+      "loss": 0.0105,
+      "step": 361
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.18304197382791004,
+      "learning_rate": 4.647987038545496e-05,
+      "loss": 0.011,
+      "step": 362
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 0.16720171411001336,
+      "learning_rate": 4.595952170270542e-05,
+      "loss": 0.0112,
+      "step": 363
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.22478251297433013,
+      "learning_rate": 4.544123166744172e-05,
+      "loss": 0.0118,
+      "step": 364
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 0.1598572948562243,
+      "learning_rate": 4.492502002368738e-05,
+      "loss": 0.0107,
+      "step": 365
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 0.22373563049772874,
+      "learning_rate": 4.4410906436290566e-05,
+      "loss": 0.0104,
+      "step": 366
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 0.16802667132434534,
+      "learning_rate": 4.38989104901751e-05,
+      "loss": 0.0114,
+      "step": 367
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 0.24550738449688075,
+      "learning_rate": 4.3389051689594e-05,
+      "loss": 0.0121,
+      "step": 368
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.1660066244443363,
+      "learning_rate": 4.288134945738684e-05,
+      "loss": 0.0099,
+      "step": 369
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.1783889244909253,
+      "learning_rate": 4.237582313423962e-05,
+      "loss": 0.0094,
+      "step": 370
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.17141038466777303,
+      "learning_rate": 4.187249197794813e-05,
+      "loss": 0.0095,
+      "step": 371
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 0.1893721805088239,
+      "learning_rate": 4.137137516268426e-05,
+      "loss": 0.013,
+      "step": 372
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 0.16935951673752134,
+      "learning_rate": 4.0872491778265535e-05,
+      "loss": 0.0091,
+      "step": 373
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.13309068523326859,
+      "learning_rate": 4.037586082942805e-05,
+      "loss": 0.0091,
+      "step": 374
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 0.18791651271841342,
+      "learning_rate": 3.988150123510224e-05,
+      "loss": 0.0121,
+      "step": 375
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 0.1559825545952661,
+      "learning_rate": 3.938943182769246e-05,
+      "loss": 0.0102,
+      "step": 376
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 0.2261919531211638,
+      "learning_rate": 3.88996713523594e-05,
+      "loss": 0.0127,
+      "step": 377
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 0.20792420146527377,
+      "learning_rate": 3.841223846630599e-05,
+      "loss": 0.013,
+      "step": 378
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.16486082885129608,
+      "learning_rate": 3.792715173806669e-05,
+      "loss": 0.0105,
+      "step": 379
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 0.1549020176177142,
+      "learning_rate": 3.74444296468002e-05,
+      "loss": 0.0098,
+      "step": 380
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.17250200199106172,
+      "learning_rate": 3.696409058158544e-05,
+      "loss": 0.0109,
+      "step": 381
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 0.1415293330470341,
+      "learning_rate": 3.6486152840721046e-05,
+      "loss": 0.0084,
+      "step": 382
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 0.14461810975420877,
+      "learning_rate": 3.6010634631028226e-05,
+      "loss": 0.0084,
+      "step": 383
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.1557012557289619,
+      "learning_rate": 3.553755406715724e-05,
+      "loss": 0.0089,
+      "step": 384
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 0.15752891661687976,
+      "learning_rate": 3.506692917089751e-05,
+      "loss": 0.0109,
+      "step": 385
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 0.1694876915505117,
+      "learning_rate": 3.459877787049072e-05,
+      "loss": 0.009,
+      "step": 386
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 0.1582663784415179,
+      "learning_rate": 3.413311799994808e-05,
+      "loss": 0.0095,
+      "step": 387
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 0.13693031068741818,
+      "learning_rate": 3.366996729837102e-05,
+      "loss": 0.0092,
+      "step": 388
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 0.14543112940410688,
+      "learning_rate": 3.320934340927513e-05,
+      "loss": 0.0108,
+      "step": 389
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.19389482832864774,
+      "learning_rate": 3.275126387991847e-05,
+      "loss": 0.0098,
+      "step": 390
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.15797165592004603,
+      "learning_rate": 3.229574616063268e-05,
+      "loss": 0.0076,
+      "step": 391
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 0.21281942854700847,
+      "learning_rate": 3.184280760415843e-05,
+      "loss": 0.0142,
+      "step": 392
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 0.12498130411986656,
+      "learning_rate": 3.1392465464984455e-05,
+      "loss": 0.0081,
+      "step": 393
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 0.1152125429659436,
+      "learning_rate": 3.094473689869002e-05,
+      "loss": 0.0058,
+      "step": 394
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.1567733530080216,
+      "learning_rate": 3.0499638961291623e-05,
+      "loss": 0.011,
+      "step": 395
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 0.14500898906990572,
+      "learning_rate": 3.0057188608593147e-05,
+      "loss": 0.0085,
+      "step": 396
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 0.16163974543952728,
+      "learning_rate": 2.9617402695539808e-05,
+      "loss": 0.013,
+      "step": 397
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 0.13868168811451842,
+      "learning_rate": 2.9180297975576364e-05,
+      "loss": 0.0084,
+      "step": 398
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 0.17847032901949134,
+      "learning_rate": 2.8745891100008683e-05,
+      "loss": 0.0121,
+      "step": 399
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.17527442252411723,
+      "learning_rate": 2.83141986173694e-05,
+      "loss": 0.0084,
+      "step": 400
+    },
+    {
+      "epoch": 3.8,
+      "eval_blimp_filtered_avg": 0.7053731343283582,
+      "eval_blimp_filtered_std": 0.005043001462199571,
+      "step": 400
+    },
+    {
+      "epoch": 3.8,
+      "eval_blimp_supplement_avg": 0.8125,
+      "eval_blimp_supplement_std": 0.01736311122127593,
+      "step": 400
+    },
+    {
+      "epoch": 3.8,
+      "eval_vqa_filtered_avg": 0.52,
+      "eval_vqa_filtered_std": 0.05021167315686779,
+      "step": 400
+    },
+    {
+      "epoch": 3.8,
+      "eval_winoground_filtered_avg": 0.64,
+      "eval_winoground_filtered_std": 0.048241815132442176,
+      "step": 400
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 0.14598157841040266,
+      "learning_rate": 2.788523697278773e-05,
+      "loss": 0.0093,
+      "step": 401
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 0.20150542514971506,
+      "learning_rate": 2.7459022507362686e-05,
+      "loss": 0.0122,
+      "step": 402
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 0.18255123614923588,
+      "learning_rate": 2.7035571457540865e-05,
+      "loss": 0.0103,
+      "step": 403
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.16704045474943452,
+      "learning_rate": 2.6614899954497795e-05,
+      "loss": 0.0114,
+      "step": 404
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 0.14683721625679494,
+      "learning_rate": 2.619702402352332e-05,
+      "loss": 0.01,
+      "step": 405
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 0.18144743721435366,
+      "learning_rate": 2.5781959583411374e-05,
+      "loss": 0.0129,
+      "step": 406
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 0.19646570441433073,
+      "learning_rate": 2.5369722445853304e-05,
+      "loss": 0.0143,
+      "step": 407
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 0.1668088181727681,
+      "learning_rate": 2.4960328314835745e-05,
+      "loss": 0.0089,
+      "step": 408
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 0.16111476451284476,
+      "learning_rate": 2.4553792786042262e-05,
+      "loss": 0.0091,
+      "step": 409
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.17729690845562673,
+      "learning_rate": 2.4150131346259197e-05,
+      "loss": 0.0103,
+      "step": 410
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.15155895346947004,
+      "learning_rate": 2.3749359372785883e-05,
+      "loss": 0.0096,
+      "step": 411
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 0.15041370885333255,
+      "learning_rate": 2.3351492132848664e-05,
+      "loss": 0.0085,
+      "step": 412
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 0.12197907148956355,
+      "learning_rate": 2.2956544783019418e-05,
+      "loss": 0.0067,
+      "step": 413
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 0.1788434056496877,
+      "learning_rate": 2.2564532368638146e-05,
+      "loss": 0.01,
+      "step": 414
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 0.19269466130772045,
+      "learning_rate": 2.2175469823239768e-05,
+      "loss": 0.0117,
+      "step": 415
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 0.15780826445252463,
+      "learning_rate": 2.1789371967985338e-05,
+      "loss": 0.0101,
+      "step": 416
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 0.19229144408434373,
+      "learning_rate": 2.140625351109733e-05,
+      "loss": 0.0084,
+      "step": 417
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 0.15474486143047034,
+      "learning_rate": 2.1026129047299436e-05,
+      "loss": 0.0067,
+      "step": 418
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 0.15864166155594778,
+      "learning_rate": 2.0649013057260546e-05,
+      "loss": 0.0098,
+      "step": 419
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 0.22515244613844015,
+      "learning_rate": 2.0274919907043033e-05,
+      "loss": 0.0094,
+      "step": 420
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.18684872878382638,
+      "learning_rate": 1.9903863847555648e-05,
+      "loss": 0.0127,
+      "step": 421
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 0.06270483785922072,
+      "learning_rate": 1.9535859014010526e-05,
+      "loss": 0.0028,
+      "step": 422
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 0.09948637260912774,
+      "learning_rate": 1.917091942538469e-05,
+      "loss": 0.0037,
+      "step": 423
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 0.07530065845248647,
+      "learning_rate": 1.880905898388612e-05,
+      "loss": 0.0039,
+      "step": 424
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 0.054461890750773165,
+      "learning_rate": 1.8450291474423998e-05,
+      "loss": 0.0025,
+      "step": 425
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 0.08002877578075594,
+      "learning_rate": 1.8094630564083736e-05,
+      "loss": 0.0035,
+      "step": 426
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 0.05746226463965698,
+      "learning_rate": 1.7742089801606276e-05,
+      "loss": 0.0025,
+      "step": 427
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 0.0633358139605444,
+      "learning_rate": 1.7392682616871837e-05,
+      "loss": 0.0027,
+      "step": 428
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 0.06509683268742919,
+      "learning_rate": 1.7046422320388556e-05,
+      "loss": 0.0027,
+      "step": 429
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 0.054571154616853274,
+      "learning_rate": 1.6703322102785168e-05,
+      "loss": 0.0026,
+      "step": 430
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 0.06888564779650448,
+      "learning_rate": 1.6363395034308703e-05,
+      "loss": 0.0027,
+      "step": 431
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 0.05307117129834359,
+      "learning_rate": 1.6026654064326553e-05,
+      "loss": 0.0025,
+      "step": 432
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 0.06598879328529111,
+      "learning_rate": 1.5693112020833013e-05,
+      "loss": 0.003,
+      "step": 433
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 0.054752236275106794,
+      "learning_rate": 1.5362781609960852e-05,
+      "loss": 0.0025,
+      "step": 434
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 0.07106963888787232,
+      "learning_rate": 1.5035675415497063e-05,
+      "loss": 0.0031,
+      "step": 435
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 0.052548572683446884,
+      "learning_rate": 1.471180589840363e-05,
+      "loss": 0.0025,
+      "step": 436
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 0.08828036910254508,
+      "learning_rate": 1.4391185396342789e-05,
+      "loss": 0.0038,
+      "step": 437
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 0.09463459893212552,
+      "learning_rate": 1.4073826123206946e-05,
+      "loss": 0.0038,
+      "step": 438
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 0.08002928457971342,
+      "learning_rate": 1.375974016865359e-05,
+      "loss": 0.0031,
+      "step": 439
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 0.07631532690730236,
+      "learning_rate": 1.3448939497644509e-05,
+      "loss": 0.0031,
+      "step": 440
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 0.04831761603516682,
+      "learning_rate": 1.3141435949990188e-05,
+      "loss": 0.0027,
+      "step": 441
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 0.07344003153336562,
+      "learning_rate": 1.2837241239898667e-05,
+      "loss": 0.0032,
+      "step": 442
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 0.08305075630986966,
+      "learning_rate": 1.253636695552931e-05,
+      "loss": 0.003,
+      "step": 443
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 0.1034575433958594,
+      "learning_rate": 1.2238824558551365e-05,
+      "loss": 0.0039,
+      "step": 444
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 0.06655324788558148,
+      "learning_rate": 1.1944625383707374e-05,
+      "loss": 0.003,
+      "step": 445
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 0.0790599253839735,
+      "learning_rate": 1.1653780638381328e-05,
+      "loss": 0.0029,
+      "step": 446
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 0.04198685628145689,
+      "learning_rate": 1.1366301402171775e-05,
+      "loss": 0.0017,
+      "step": 447
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 0.06439353264983554,
+      "learning_rate": 1.1082198626469686e-05,
+      "loss": 0.0024,
+      "step": 448
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 0.07762450043477247,
+      "learning_rate": 1.0801483134041268e-05,
+      "loss": 0.0027,
+      "step": 449
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 0.07856883953783565,
+      "learning_rate": 1.0524165618615845e-05,
+      "loss": 0.0033,
+      "step": 450
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 0.07929308057852809,
+      "learning_rate": 1.0250256644478195e-05,
+      "loss": 0.003,
+      "step": 451
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 0.0587512154822952,
+      "learning_rate": 9.979766646066368e-06,
+      "loss": 0.0027,
+      "step": 452
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 0.06109551507247056,
+      "learning_rate": 9.71270592757404e-06,
+      "loss": 0.0032,
+      "step": 453
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 0.05909029031199419,
+      "learning_rate": 9.449084662557982e-06,
+      "loss": 0.0026,
+      "step": 454
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 0.0814055458144323,
+      "learning_rate": 9.188912893550695e-06,
+      "loss": 0.0026,
+      "step": 455
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 0.07735385332942207,
+      "learning_rate": 8.932200531677537e-06,
+      "loss": 0.0028,
+      "step": 456
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 0.08519595591969155,
+      "learning_rate": 8.678957356279371e-06,
+      "loss": 0.0024,
+      "step": 457
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 0.055031384326470804,
+      "learning_rate": 8.429193014540015e-06,
+      "loss": 0.0026,
+      "step": 458
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 0.05387324401647046,
+      "learning_rate": 8.182917021118663e-06,
+      "loss": 0.0026,
+      "step": 459
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 0.07168879976269556,
+      "learning_rate": 7.940138757787507e-06,
+      "loss": 0.0032,
+      "step": 460
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 0.07661756681904786,
+      "learning_rate": 7.700867473074224e-06,
+      "loss": 0.0035,
+      "step": 461
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 0.09486930411075328,
+      "learning_rate": 7.46511228190977e-06,
+      "loss": 0.0049,
+      "step": 462
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 0.0679530025111762,
+      "learning_rate": 7.232882165281141e-06,
+      "loss": 0.0026,
+      "step": 463
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 0.06514922044267304,
+      "learning_rate": 7.004185969889187e-06,
+      "loss": 0.0027,
+      "step": 464
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 0.06706026131022384,
+      "learning_rate": 6.7790324078116364e-06,
+      "loss": 0.0027,
+      "step": 465
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 0.07709046890424658,
+      "learning_rate": 6.557430056171221e-06,
+      "loss": 0.0033,
+      "step": 466
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 0.051443041020356704,
+      "learning_rate": 6.339387356808912e-06,
+      "loss": 0.0026,
+      "step": 467
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 0.060318722923432995,
+      "learning_rate": 6.124912615962341e-06,
+      "loss": 0.0028,
+      "step": 468
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 0.062212012735137795,
+      "learning_rate": 5.9140140039494084e-06,
+      "loss": 0.0025,
+      "step": 469
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 0.06556299474776538,
+      "learning_rate": 5.706699554856964e-06,
+      "loss": 0.0023,
+      "step": 470
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 0.08649267044276539,
+      "learning_rate": 5.502977166234857e-06,
+      "loss": 0.0035,
+      "step": 471
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 0.08526822145924882,
+      "learning_rate": 5.302854598794937e-06,
+      "loss": 0.003,
+      "step": 472
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 0.04133711118453636,
+      "learning_rate": 5.106339476115596e-06,
+      "loss": 0.0019,
+      "step": 473
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 0.05708577094578342,
+      "learning_rate": 4.913439284351207e-06,
+      "loss": 0.0026,
+      "step": 474
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 0.07367912633186298,
+      "learning_rate": 4.724161371946978e-06,
+      "loss": 0.0029,
+      "step": 475
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 0.08135320771271103,
+      "learning_rate": 4.538512949359075e-06,
+      "loss": 0.0027,
+      "step": 476
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 0.0849858165893086,
+      "learning_rate": 4.356501088779841e-06,
+      "loss": 0.0027,
+      "step": 477
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 0.05260609110954984,
+      "learning_rate": 4.178132723868477e-06,
+      "loss": 0.0019,
+      "step": 478
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 0.0795477617292828,
+      "learning_rate": 4.003414649486892e-06,
+      "loss": 0.0032,
+      "step": 479
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 0.08161922179718771,
+      "learning_rate": 3.832353521440768e-06,
+      "loss": 0.0026,
+      "step": 480
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 0.06830643544893618,
+      "learning_rate": 3.6649558562261375e-06,
+      "loss": 0.0032,
+      "step": 481
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 0.08641205617098656,
+      "learning_rate": 3.501228030781034e-06,
+      "loss": 0.0028,
+      "step": 482
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 0.04921706287498077,
+      "learning_rate": 3.341176282242653e-06,
+      "loss": 0.0021,
+      "step": 483
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 0.05901589705081983,
+      "learning_rate": 3.184806707709698e-06,
+      "loss": 0.0027,
+      "step": 484
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 0.08562934355546689,
+      "learning_rate": 3.0321252640100885e-06,
+      "loss": 0.0035,
+      "step": 485
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 0.056139936545776606,
+      "learning_rate": 2.88313776747412e-06,
+      "loss": 0.0027,
+      "step": 486
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 0.06574452787357139,
+      "learning_rate": 2.7378498937128404e-06,
+      "loss": 0.0031,
+      "step": 487
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 0.06295208396607756,
+      "learning_rate": 2.5962671774018234e-06,
+      "loss": 0.0029,
+      "step": 488
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 0.06348707610420529,
+      "learning_rate": 2.458395012070369e-06,
+      "loss": 0.0027,
+      "step": 489
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 0.06438459591992919,
+      "learning_rate": 2.3242386498960266e-06,
+      "loss": 0.003,
+      "step": 490
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 0.0936033257355208,
+      "learning_rate": 2.1938032015044964e-06,
+      "loss": 0.0053,
+      "step": 491
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 0.0712704009642112,
+      "learning_rate": 2.067093635774975e-06,
+      "loss": 0.0033,
+      "step": 492
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 0.05278839840964536,
+      "learning_rate": 1.9441147796508407e-06,
+      "loss": 0.0025,
+      "step": 493
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 0.05158800004403027,
+      "learning_rate": 1.8248713179557786e-06,
+      "loss": 0.002,
+      "step": 494
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 0.06302315225352234,
+      "learning_rate": 1.7093677932153218e-06,
+      "loss": 0.002,
+      "step": 495
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 0.09014451602286425,
+      "learning_rate": 1.5976086054838025e-06,
+      "loss": 0.0031,
+      "step": 496
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 0.08249201483869177,
+      "learning_rate": 1.4895980121767627e-06,
+      "loss": 0.0029,
+      "step": 497
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 0.07887788932672342,
+      "learning_rate": 1.3853401279086854e-06,
+      "loss": 0.0028,
+      "step": 498
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 0.09271365227044996,
+      "learning_rate": 1.2848389243363512e-06,
+      "loss": 0.0026,
+      "step": 499
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 0.05191622392926365,
+      "learning_rate": 1.1880982300074838e-06,
+      "loss": 0.0027,
+      "step": 500
+    },
+    {
+      "epoch": 4.75,
+      "eval_blimp_filtered_avg": 0.7105970149253731,
+      "eval_blimp_filtered_std": 0.005015059082306442,
+      "step": 500
+    },
+    {
+      "epoch": 4.75,
+      "eval_blimp_supplement_avg": 0.8146551724137931,
+      "eval_blimp_supplement_std": 0.01739418193453382,
+      "step": 500
+    },
+    {
+      "epoch": 4.75,
+      "eval_vqa_filtered_avg": 0.52,
+      "eval_vqa_filtered_std": 0.05021167315686779,
+      "step": 500
+    },
+    {
+      "epoch": 4.75,
+      "eval_winoground_filtered_avg": 0.64,
+      "eval_winoground_filtered_std": 0.048241815132442176,
+      "step": 500
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 525,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "total_flos": 394333829201920.0,
+  "train_batch_size": 40,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-500/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:08b5f65153ee777fbd3d7179f8557cd50d2f8195b4001cff264a353dcb84007a
+size 7032

checkpoint-500/zero_to_fp32.py

@@ -0,0 +1,587 @@
+#!/usr/bin/env python
+
+# Copyright (c) Microsoft Corporation.
+# SPDX-License-Identifier: Apache-2.0
+
+# DeepSpeed Team
+
+# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
+# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
+# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
+# application.
+#
+# example: python zero_to_fp32.py . pytorch_model.bin
+
+import argparse
+import torch
+import glob
+import math
+import os
+import re
+from collections import OrderedDict
+from dataclasses import dataclass
+
+# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
+# DeepSpeed data structures it has to be available in the current python environment.
+from deepspeed.utils import logger
+from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
+                                            FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
+                                            FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
+
+
+@dataclass
+class zero_model_state:
+    buffers: dict()
+    param_shapes: dict()
+    shared_params: list
+    ds_version: int
+    frozen_param_shapes: dict()
+    frozen_param_fragments: dict()
+
+
+debug = 0
+
+# load to cpu
+device = torch.device('cpu')
+
+
+def atoi(text):
+    return int(text) if text.isdigit() else text
+
+
+def natural_keys(text):
+    '''
+    alist.sort(key=natural_keys) sorts in human order
+    http://nedbatchelder.com/blog/200712/human_sorting.html
+    (See Toothy's implementation in the comments)
+    '''
+    return [atoi(c) for c in re.split(r'(\d+)', text)]
+
+
+def get_model_state_file(checkpoint_dir, zero_stage):
+    if not os.path.isdir(checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
+
+    # there should be only one file
+    if zero_stage <= 2:
+        file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
+    elif zero_stage == 3:
+        file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
+
+    if not os.path.exists(file):
+        raise FileNotFoundError(f"can't find model states file at '{file}'")
+
+    return file
+
+
+def get_checkpoint_files(checkpoint_dir, glob_pattern):
+    # XXX: need to test that this simple glob rule works for multi-node setup too
+    ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
+
+    if len(ckpt_files) == 0:
+        raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
+
+    return ckpt_files
+
+
+def get_optim_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
+
+
+def get_model_state_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
+
+
+def parse_model_states(files):
+    zero_model_states = []
+    for file in files:
+        state_dict = torch.load(file, map_location=device)
+
+        if BUFFER_NAMES not in state_dict:
+            raise ValueError(f"{file} is not a model state checkpoint")
+        buffer_names = state_dict[BUFFER_NAMES]
+        if debug:
+            print("Found buffers:", buffer_names)
+
+        # recover just the buffers while restoring them to fp32 if they were saved in fp16
+        buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
+        param_shapes = state_dict[PARAM_SHAPES]
+
+        # collect parameters that are included in param_shapes
+        param_names = []
+        for s in param_shapes:
+            for name in s.keys():
+                param_names.append(name)
+
+        # update with frozen parameters
+        frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
+        if frozen_param_shapes is not None:
+            if debug:
+                print(f"Found frozen_param_shapes: {frozen_param_shapes}")
+            param_names += list(frozen_param_shapes.keys())
+
+        # handle shared params
+        shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
+
+        ds_version = state_dict.get(DS_VERSION, None)
+
+        frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
+
+        z_model_state = zero_model_state(buffers=buffers,
+                                         param_shapes=param_shapes,
+                                         shared_params=shared_params,
+                                         ds_version=ds_version,
+                                         frozen_param_shapes=frozen_param_shapes,
+                                         frozen_param_fragments=frozen_param_fragments)
+        zero_model_states.append(z_model_state)
+
+    return zero_model_states
+
+
+def parse_optim_states(files, ds_checkpoint_dir):
+
+    total_files = len(files)
+    state_dicts = []
+    for f in files:
+        state_dict = torch.load(f, map_location=device)
+        # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
+        # and also handle the case where it was already removed by another helper script
+        state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
+        state_dicts.append(state_dict)
+
+    if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
+        raise ValueError(f"{files[0]} is not a zero checkpoint")
+    zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
+    world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
+
+    # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
+    # parameters can be different from data parallelism for non-expert parameters. So we can just
+    # use the max of the partition_count to get the dp world_size.
+
+    if type(world_size) is list:
+        world_size = max(world_size)
+
+    if world_size != total_files:
+        raise ValueError(
+            f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
+            "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
+        )
+
+    # the groups are named differently in each stage
+    if zero_stage <= 2:
+        fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
+    elif zero_stage == 3:
+        fp32_groups_key = FP32_FLAT_GROUPS
+    else:
+        raise ValueError(f"unknown zero stage {zero_stage}")
+
+    if zero_stage <= 2:
+        fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
+    elif zero_stage == 3:
+        # if there is more than one param group, there will be multiple flattened tensors - one
+        # flattened tensor per group - for simplicity merge them into a single tensor
+        #
+        # XXX: could make the script more memory efficient for when there are multiple groups - it
+        # will require matching the sub-lists of param_shapes for each param group flattened tensor
+
+        fp32_flat_groups = [
+            torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
+        ]
+
+    return zero_stage, world_size, fp32_flat_groups
+
+
+def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir):
+    """
+    Returns fp32 state_dict reconstructed from ds checkpoint
+
+    Args:
+        - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
+
+    """
+    print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
+
+    optim_files = get_optim_files(ds_checkpoint_dir)
+    zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
+    print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
+
+    model_files = get_model_state_files(ds_checkpoint_dir)
+
+    zero_model_states = parse_model_states(model_files)
+    print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
+
+    if zero_stage <= 2:
+        return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states)
+    elif zero_stage == 3:
+        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states)
+
+
+def _zero2_merge_frozen_params(state_dict, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+    frozen_param_fragments = zero_model_states[0].frozen_param_fragments
+
+    if debug:
+        num_elem = sum(s.numel() for s in frozen_param_shapes.values())
+        print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        state_dict[name] = frozen_param_fragments[name]
+
+        if debug:
+            print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+
+    # Reconstruction protocol:
+    #
+    # XXX: document this
+
+    if debug:
+        for i in range(world_size):
+            for j in range(len(fp32_flat_groups[0])):
+                print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
+
+    # XXX: memory usage doubles here (zero2)
+    num_param_groups = len(fp32_flat_groups[0])
+    merged_single_partition_of_fp32_groups = []
+    for i in range(num_param_groups):
+        merged_partitions = [sd[i] for sd in fp32_flat_groups]
+        full_single_fp32_vector = torch.cat(merged_partitions, 0)
+        merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
+    avail_numel = sum(
+        [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
+
+    if debug:
+        wanted_params = sum([len(shapes) for shapes in param_shapes])
+        wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
+        # not asserting if there is a mismatch due to possible padding
+        print(f"Have {avail_numel} numels to process.")
+        print(f"Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    total_numel = 0
+    total_params = 0
+    for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
+        offset = 0
+        avail_numel = full_single_fp32_vector.numel()
+        for name, shape in shapes.items():
+
+            unpartitioned_numel = shape.numel()
+            total_numel += unpartitioned_numel
+            total_params += 1
+
+            if debug:
+                print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+            state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
+            offset += unpartitioned_numel
+
+        # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
+        # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
+        # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
+        # live optimizer object, so we are checking that the numbers are within the right range
+        align_to = 2 * world_size
+
+        def zero2_align(x):
+            return align_to * math.ceil(x / align_to)
+
+        if debug:
+            print(f"original offset={offset}, avail_numel={avail_numel}")
+
+        offset = zero2_align(offset)
+        avail_numel = zero2_align(avail_numel)
+
+        if debug:
+            print(f"aligned  offset={offset}, avail_numel={avail_numel}")
+
+        # Sanity check
+        if offset != avail_numel:
+            raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    _zero2_merge_frozen_params(state_dict, zero_model_states)
+
+    _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def zero3_partitioned_param_info(unpartitioned_numel, world_size):
+    remainder = unpartitioned_numel % world_size
+    padding_numel = (world_size - remainder) if remainder else 0
+    partitioned_numel = math.ceil(unpartitioned_numel / world_size)
+    return partitioned_numel, padding_numel
+
+
+def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    if debug:
+        for i in range(world_size):
+            num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
+            print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in zero_model_states[0].frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
+        state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+    avail_numel = fp32_flat_groups[0].numel() * world_size
+    # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
+    # param, re-consolidating each param, while dealing with padding if any
+
+    # merge list of dicts, preserving order
+    param_shapes = {k: v for d in param_shapes for k, v in d.items()}
+
+    if debug:
+        for i in range(world_size):
+            print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
+
+        wanted_params = len(param_shapes)
+        wanted_numel = sum(shape.numel() for shape in param_shapes.values())
+        # not asserting if there is a mismatch due to possible padding
+        avail_numel = fp32_flat_groups[0].numel() * world_size
+        print(f"Trainable params: Have {avail_numel} numels to process.")
+        print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    offset = 0
+    total_numel = 0
+    total_params = 0
+    for name, shape in param_shapes.items():
+
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+        total_params += 1
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+        # XXX: memory usage doubles here
+        state_dict[name] = torch.cat(
+            tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
+            0).narrow(0, 0, unpartitioned_numel).view(shape)
+        offset += partitioned_numel
+
+    offset *= world_size
+
+    # Sanity check
+    if offset != avail_numel:
+        raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
+
+    _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
+    ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
+    via a model hub.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
+
+    Returns:
+        - pytorch ``state_dict``
+
+    Note: this approach may not work if your application doesn't have sufficient free CPU memory and
+    you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
+    the checkpoint.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        # do the training and checkpoint saving
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
+        model = model.cpu() # move to cpu
+        model.load_state_dict(state_dict)
+        # submit to model hub or save the model to share with others
+
+    In this example the ``model`` will no longer be usable in the deepspeed context of the same
+    application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
+
+    """
+    if tag is None:
+        latest_path = os.path.join(checkpoint_dir, 'latest')
+        if os.path.isfile(latest_path):
+            with open(latest_path, 'r') as fd:
+                tag = fd.read().strip()
+        else:
+            raise ValueError(f"Unable to find 'latest' file at {latest_path}")
+
+    ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
+
+    if not os.path.isdir(ds_checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
+
+    return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir)
+
+
+def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
+    loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+    """
+
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+    print(f"Saving fp32 state dict to {output_file}")
+    torch.save(state_dict, output_file)
+
+
+def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
+    """
+    1. Put the provided model to cpu
+    2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
+    3. Load it into the provided model
+
+    Args:
+        - ``model``: the model object to update
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+
+    Returns:
+        - ``model`: modified model
+
+    Make sure you have plenty of CPU memory available before you call this function. If you don't
+    have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
+    conveniently placed for you in the checkpoint folder.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
+        model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
+        # submit to model hub or save the model to share with others
+
+    Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
+    of the same application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    """
+    logger.info(f"Extracting fp32 weights")
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+
+    logger.info(f"Overwriting model with fp32 weights")
+    model = model.cpu()
+    model.load_state_dict(state_dict, strict=False)
+
+    return model
+
+
+if __name__ == "__main__":
+
+    parser = argparse.ArgumentParser()
+    parser.add_argument("checkpoint_dir",
+                        type=str,
+                        help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
+    parser.add_argument(
+        "output_file",
+        type=str,
+        help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
+    parser.add_argument("-t",
+                        "--tag",
+                        type=str,
+                        default=None,
+                        help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
+    parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
+    args = parser.parse_args()
+
+    debug = args.debug
+
+    convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, args.output_file, tag=args.tag)

config.json

@@ -0,0 +1,49 @@
+{
+  "_name_or_path": "liuhaotian/llava-v1.5-7b",
+  "architectures": [
+    "LlavaLlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "freeze_mm_mlp_adapter": false,
+  "freeze_mm_vision_resampler": false,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "image_aspect_ratio": "pad",
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "max_length": 4096,
+  "max_position_embeddings": 4096,
+  "mm_hidden_size": 1024,
+  "mm_patch_merge_type": "flat",
+  "mm_projector_lr": 2e-05,
+  "mm_projector_type": "mlp2x_gelu",
+  "mm_resampler_type": null,
+  "mm_use_im_patch_token": false,
+  "mm_use_im_start_end": false,
+  "mm_vision_select_feature": "patch",
+  "mm_vision_select_layer": -2,
+  "mm_vision_tower": "openai/clip-vit-large-patch14-336",
+  "model_type": "llava_llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 32,
+  "pad_token_id": 0,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": null,
+  "rope_theta": 10000.0,
+  "tie_word_embeddings": false,
+  "tokenizer_model_max_length": 2048,
+  "tokenizer_padding_side": "right",
+  "torch_dtype": "float16",
+  "transformers_version": "4.39.3",
+  "tune_mm_mlp_adapter": false,
+  "tune_mm_vision_resampler": false,
+  "unfreeze_mm_vision_tower": false,
+  "use_cache": true,
+  "use_mm_proj": true,
+  "vocab_size": 32000
+}

non_lora_trainables.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4256d30527b68662b35b87c465a72acf392ee1cb4c248edee2238d360ce1f04e
+size 41961648

special_tokens_map.json

@@ -0,0 +1,24 @@
+{
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "<unk>",
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
+size 499723

tokenizer_config.json

@@ -0,0 +1,43 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "add_prefix_space": true,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "</s>",
+  "legacy": false,
+  "model_max_length": 2048,
+  "pad_token": "<unk>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}

trainer_state.json

@@ -0,0 +1,3825 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.98812351543943,
+  "eval_steps": 100,
+  "global_step": 525,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.008147055910771,
+      "learning_rate": 1.25e-05,
+      "loss": 4.2415,
+      "step": 1
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.04569203441769,
+      "learning_rate": 2.5e-05,
+      "loss": 4.3121,
+      "step": 2
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.865746651377984,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 4.3208,
+      "step": 3
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.6407193073379105,
+      "learning_rate": 5e-05,
+      "loss": 3.8848,
+      "step": 4
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.451159328560232,
+      "learning_rate": 6.25e-05,
+      "loss": 3.4391,
+      "step": 5
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.8259504797317525,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 3.0656,
+      "step": 6
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1881779175566867,
+      "learning_rate": 8.75e-05,
+      "loss": 2.8135,
+      "step": 7
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.614839668966139,
+      "learning_rate": 0.0001,
+      "loss": 2.7319,
+      "step": 8
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.5198673994210212,
+      "learning_rate": 0.00011250000000000001,
+      "loss": 2.6903,
+      "step": 9
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0044025931610727,
+      "learning_rate": 0.000125,
+      "loss": 2.584,
+      "step": 10
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.1531821793787296,
+      "learning_rate": 0.0001375,
+      "loss": 2.586,
+      "step": 11
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6210600474209341,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 2.5298,
+      "step": 12
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5025244204180619,
+      "learning_rate": 0.00016250000000000002,
+      "loss": 2.4665,
+      "step": 13
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5058788641352842,
+      "learning_rate": 0.000175,
+      "loss": 2.4194,
+      "step": 14
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.44571801666869537,
+      "learning_rate": 0.0001875,
+      "loss": 2.3531,
+      "step": 15
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.44028009268534757,
+      "learning_rate": 0.0002,
+      "loss": 2.2749,
+      "step": 16
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.42473118020142525,
+      "learning_rate": 0.00019999809527270051,
+      "loss": 2.2587,
+      "step": 17
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.465029302165452,
+      "learning_rate": 0.0001999923811633618,
+      "loss": 2.2196,
+      "step": 18
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.49040381415815754,
+      "learning_rate": 0.00019998285788966027,
+      "loss": 2.2061,
+      "step": 19
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4160855034634493,
+      "learning_rate": 0.00019996952581438068,
+      "loss": 2.1173,
+      "step": 20
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.45625369964232165,
+      "learning_rate": 0.00019995238544540241,
+      "loss": 2.1267,
+      "step": 21
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.42551849567803673,
+      "learning_rate": 0.00019993143743568,
+      "loss": 2.0976,
+      "step": 22
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5100052595965069,
+      "learning_rate": 0.0001999066825832184,
+      "loss": 2.0428,
+      "step": 23
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4717525078599394,
+      "learning_rate": 0.00019987812183104247,
+      "loss": 2.0068,
+      "step": 24
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.5596905853419681,
+      "learning_rate": 0.0001998457562671611,
+      "loss": 2.0303,
+      "step": 25
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4931645550169434,
+      "learning_rate": 0.00019980958712452577,
+      "loss": 1.9722,
+      "step": 26
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4433810930704678,
+      "learning_rate": 0.0001997696157809835,
+      "loss": 1.957,
+      "step": 27
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5522396650266582,
+      "learning_rate": 0.0001997258437592245,
+      "loss": 1.915,
+      "step": 28
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.49861222066728145,
+      "learning_rate": 0.00019967827272672408,
+      "loss": 1.8303,
+      "step": 29
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6169911964169147,
+      "learning_rate": 0.00019962690449567912,
+      "loss": 1.8454,
+      "step": 30
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5639780725078123,
+      "learning_rate": 0.000199571741022939,
+      "loss": 1.8068,
+      "step": 31
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6302805853808786,
+      "learning_rate": 0.0001995127844099313,
+      "loss": 1.7166,
+      "step": 32
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6494693483139545,
+      "learning_rate": 0.00019945003690258125,
+      "loss": 1.6433,
+      "step": 33
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.7598443409498918,
+      "learning_rate": 0.00019938350089122682,
+      "loss": 1.7081,
+      "step": 34
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6512764391881087,
+      "learning_rate": 0.00019931317891052708,
+      "loss": 1.6436,
+      "step": 35
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6953537359048508,
+      "learning_rate": 0.00019923907363936593,
+      "loss": 1.5862,
+      "step": 36
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6011387829084072,
+      "learning_rate": 0.00019916118790075008,
+      "loss": 1.5432,
+      "step": 37
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.659130437748028,
+      "learning_rate": 0.00019907952466170138,
+      "loss": 1.5132,
+      "step": 38
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7211467253555573,
+      "learning_rate": 0.00019899408703314385,
+      "loss": 1.506,
+      "step": 39
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7006890038987398,
+      "learning_rate": 0.0001989048782697851,
+      "loss": 1.4498,
+      "step": 40
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.64642158324997,
+      "learning_rate": 0.00019881190176999255,
+      "loss": 1.4478,
+      "step": 41
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6608085069521318,
+      "learning_rate": 0.00019871516107566366,
+      "loss": 1.3542,
+      "step": 42
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7707478188072372,
+      "learning_rate": 0.0001986146598720913,
+      "loss": 1.3309,
+      "step": 43
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8119298049916807,
+      "learning_rate": 0.00019851040198782326,
+      "loss": 1.345,
+      "step": 44
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7712308653234212,
+      "learning_rate": 0.0001984023913945162,
+      "loss": 1.3076,
+      "step": 45
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.682341709525683,
+      "learning_rate": 0.0001982906322067847,
+      "loss": 1.2565,
+      "step": 46
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7071991083514119,
+      "learning_rate": 0.00019817512868204425,
+      "loss": 1.1796,
+      "step": 47
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.745222014713615,
+      "learning_rate": 0.00019805588522034916,
+      "loss": 1.1649,
+      "step": 48
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7158459299510994,
+      "learning_rate": 0.00019793290636422505,
+      "loss": 1.2109,
+      "step": 49
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7335821144549012,
+      "learning_rate": 0.00019780619679849552,
+      "loss": 1.1475,
+      "step": 50
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7804306024320766,
+      "learning_rate": 0.000197675761350104,
+      "loss": 1.1068,
+      "step": 51
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8274924156959725,
+      "learning_rate": 0.00019754160498792965,
+      "loss": 1.1839,
+      "step": 52
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.8840482383868431,
+      "learning_rate": 0.0001974037328225982,
+      "loss": 1.0928,
+      "step": 53
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7224652999279871,
+      "learning_rate": 0.00019726215010628718,
+      "loss": 1.0299,
+      "step": 54
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7109288879933862,
+      "learning_rate": 0.0001971168622325259,
+      "loss": 1.0436,
+      "step": 55
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7650325966583326,
+      "learning_rate": 0.00019696787473598993,
+      "loss": 1.041,
+      "step": 56
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7307809391946058,
+      "learning_rate": 0.00019681519329229033,
+      "loss": 1.0195,
+      "step": 57
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6873943623441443,
+      "learning_rate": 0.00019665882371775733,
+      "loss": 0.972,
+      "step": 58
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8185924734616268,
+      "learning_rate": 0.00019649877196921896,
+      "loss": 0.9986,
+      "step": 59
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7907558585543373,
+      "learning_rate": 0.00019633504414377388,
+      "loss": 0.9201,
+      "step": 60
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7216280408288712,
+      "learning_rate": 0.00019616764647855926,
+      "loss": 0.9976,
+      "step": 61
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6946470891456141,
+      "learning_rate": 0.00019599658535051314,
+      "loss": 0.9008,
+      "step": 62
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6470248283451219,
+      "learning_rate": 0.00019582186727613152,
+      "loss": 0.8226,
+      "step": 63
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8297915622585336,
+      "learning_rate": 0.00019564349891122018,
+      "loss": 0.8825,
+      "step": 64
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7018515834126928,
+      "learning_rate": 0.00019546148705064097,
+      "loss": 0.8521,
+      "step": 65
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6119835758734723,
+      "learning_rate": 0.00019527583862805303,
+      "loss": 0.7872,
+      "step": 66
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6396036538427098,
+      "learning_rate": 0.00019508656071564882,
+      "loss": 0.7887,
+      "step": 67
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.6712059239435435,
+      "learning_rate": 0.00019489366052388441,
+      "loss": 0.8406,
+      "step": 68
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.6498227189328728,
+      "learning_rate": 0.00019469714540120507,
+      "loss": 0.7109,
+      "step": 69
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6950957852561941,
+      "learning_rate": 0.00019449702283376517,
+      "loss": 0.7008,
+      "step": 70
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6415745385783075,
+      "learning_rate": 0.00019429330044514305,
+      "loss": 0.6808,
+      "step": 71
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6774461765802887,
+      "learning_rate": 0.0001940859859960506,
+      "loss": 0.7122,
+      "step": 72
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.6335543398879422,
+      "learning_rate": 0.00019387508738403768,
+      "loss": 0.6826,
+      "step": 73
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6455659601218003,
+      "learning_rate": 0.0001936606126431911,
+      "loss": 0.7342,
+      "step": 74
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.6804108080708727,
+      "learning_rate": 0.00019344256994382878,
+      "loss": 0.6983,
+      "step": 75
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6233570198373359,
+      "learning_rate": 0.00019322096759218836,
+      "loss": 0.6426,
+      "step": 76
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6354196060962453,
+      "learning_rate": 0.00019299581403011082,
+      "loss": 0.6978,
+      "step": 77
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.6723728632702363,
+      "learning_rate": 0.0001927671178347189,
+      "loss": 0.6449,
+      "step": 78
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.6055794839258588,
+      "learning_rate": 0.00019253488771809024,
+      "loss": 0.6608,
+      "step": 79
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6032563228830964,
+      "learning_rate": 0.0001922991325269258,
+      "loss": 0.6691,
+      "step": 80
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5917538532836075,
+      "learning_rate": 0.00019205986124221251,
+      "loss": 0.6418,
+      "step": 81
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6558132078005496,
+      "learning_rate": 0.00019181708297888133,
+      "loss": 0.6562,
+      "step": 82
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.6110330049943966,
+      "learning_rate": 0.00019157080698546,
+      "loss": 0.5855,
+      "step": 83
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.6481622083495842,
+      "learning_rate": 0.00019132104264372063,
+      "loss": 0.628,
+      "step": 84
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5730813607452849,
+      "learning_rate": 0.0001910677994683225,
+      "loss": 0.5476,
+      "step": 85
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.6938507563801335,
+      "learning_rate": 0.00019081108710644932,
+      "loss": 0.6018,
+      "step": 86
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.625439427503205,
+      "learning_rate": 0.00019055091533744202,
+      "loss": 0.5735,
+      "step": 87
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.6628596764324554,
+      "learning_rate": 0.00019028729407242597,
+      "loss": 0.5389,
+      "step": 88
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6112099968245533,
+      "learning_rate": 0.00019002023335393364,
+      "loss": 0.5235,
+      "step": 89
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.6098216223216336,
+      "learning_rate": 0.0001897497433555218,
+      "loss": 0.6058,
+      "step": 90
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.6469247467013166,
+      "learning_rate": 0.0001894758343813842,
+      "loss": 0.5524,
+      "step": 91
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.6344920759870597,
+      "learning_rate": 0.00018919851686595874,
+      "loss": 0.5605,
+      "step": 92
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.6756355159547938,
+      "learning_rate": 0.00018891780137353034,
+      "loss": 0.5096,
+      "step": 93
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.6439314455537293,
+      "learning_rate": 0.00018863369859782825,
+      "loss": 0.5516,
+      "step": 94
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5567728554741562,
+      "learning_rate": 0.0001883462193616187,
+      "loss": 0.4576,
+      "step": 95
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.553595533418767,
+      "learning_rate": 0.00018805537461629265,
+      "loss": 0.4947,
+      "step": 96
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.6200223910647112,
+      "learning_rate": 0.00018776117544144863,
+      "loss": 0.5073,
+      "step": 97
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.6294322114297511,
+      "learning_rate": 0.00018746363304447073,
+      "loss": 0.4938,
+      "step": 98
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6000145257745209,
+      "learning_rate": 0.00018716275876010135,
+      "loss": 0.473,
+      "step": 99
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5927861897994469,
+      "learning_rate": 0.00018685856405000983,
+      "loss": 0.4724,
+      "step": 100
+    },
+    {
+      "epoch": 0.95,
+      "eval_blimp_filtered_avg": 0.7155223880597015,
+      "eval_blimp_filtered_std": 0.005000433138834185,
+      "step": 100
+    },
+    {
+      "epoch": 0.95,
+      "eval_blimp_supplement_avg": 0.8405172413793104,
+      "eval_blimp_supplement_std": 0.016486001732879434,
+      "step": 100
+    },
+    {
+      "epoch": 0.95,
+      "eval_vqa_filtered_avg": 0.52,
+      "eval_vqa_filtered_std": 0.05021167315686779,
+      "step": 100
+    },
+    {
+      "epoch": 0.95,
+      "eval_winoground_filtered_avg": 0.64,
+      "eval_winoground_filtered_std": 0.04824181513244218,
+      "step": 100
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5504516732077648,
+      "learning_rate": 0.00018655106050235548,
+      "loss": 0.4393,
+      "step": 101
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5801589113252366,
+      "learning_rate": 0.00018624025983134644,
+      "loss": 0.468,
+      "step": 102
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5273944337529535,
+      "learning_rate": 0.00018592617387679306,
+      "loss": 0.439,
+      "step": 103
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.508609381383424,
+      "learning_rate": 0.00018560881460365724,
+      "loss": 0.4272,
+      "step": 104
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5396859577867195,
+      "learning_rate": 0.0001852881941015964,
+      "loss": 0.4362,
+      "step": 105
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5122858999271028,
+      "learning_rate": 0.00018496432458450294,
+      "loss": 0.3893,
+      "step": 106
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.49626561438760436,
+      "learning_rate": 0.00018463721839003915,
+      "loss": 0.3498,
+      "step": 107
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.48748413013987063,
+      "learning_rate": 0.000184306887979167,
+      "loss": 0.3256,
+      "step": 108
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5310280563857716,
+      "learning_rate": 0.00018397334593567348,
+      "loss": 0.3225,
+      "step": 109
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.6232514021230662,
+      "learning_rate": 0.00018363660496569127,
+      "loss": 0.3489,
+      "step": 110
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5274577320762,
+      "learning_rate": 0.00018329667789721485,
+      "loss": 0.3123,
+      "step": 111
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5096311315676365,
+      "learning_rate": 0.00018295357767961144,
+      "loss": 0.3325,
+      "step": 112
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.4613577097438129,
+      "learning_rate": 0.00018260731738312818,
+      "loss": 0.2936,
+      "step": 113
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.4997938044342101,
+      "learning_rate": 0.00018225791019839375,
+      "loss": 0.3351,
+      "step": 114
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.538085494988463,
+      "learning_rate": 0.00018190536943591624,
+      "loss": 0.329,
+      "step": 115
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5567068979809859,
+      "learning_rate": 0.00018154970852557603,
+      "loss": 0.318,
+      "step": 116
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5548141608588357,
+      "learning_rate": 0.0001811909410161139,
+      "loss": 0.3289,
+      "step": 117
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.47326466614968965,
+      "learning_rate": 0.0001808290805746153,
+      "loss": 0.3076,
+      "step": 118
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.47629585466918467,
+      "learning_rate": 0.00018046414098598948,
+      "loss": 0.3016,
+      "step": 119
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.44135735344426463,
+      "learning_rate": 0.00018009613615244436,
+      "loss": 0.2704,
+      "step": 120
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5127645747027901,
+      "learning_rate": 0.000179725080092957,
+      "loss": 0.2887,
+      "step": 121
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5209981172771183,
+      "learning_rate": 0.0001793509869427395,
+      "loss": 0.2938,
+      "step": 122
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5481082193558409,
+      "learning_rate": 0.00017897387095270058,
+      "loss": 0.3191,
+      "step": 123
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.4770065158307258,
+      "learning_rate": 0.0001785937464889027,
+      "loss": 0.2795,
+      "step": 124
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.44845204938493194,
+      "learning_rate": 0.0001782106280320147,
+      "loss": 0.2667,
+      "step": 125
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.47824147005907164,
+      "learning_rate": 0.00017782453017676025,
+      "loss": 0.267,
+      "step": 126
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.501015317452837,
+      "learning_rate": 0.00017743546763136187,
+      "loss": 0.2831,
+      "step": 127
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5232536606095718,
+      "learning_rate": 0.00017704345521698058,
+      "loss": 0.2769,
+      "step": 128
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5495388553709665,
+      "learning_rate": 0.00017664850786715136,
+      "loss": 0.3031,
+      "step": 129
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5371555106361774,
+      "learning_rate": 0.00017625064062721415,
+      "loss": 0.2955,
+      "step": 130
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.4716773551397148,
+      "learning_rate": 0.00017584986865374082,
+      "loss": 0.2666,
+      "step": 131
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5089124561646106,
+      "learning_rate": 0.00017544620721395777,
+      "loss": 0.3379,
+      "step": 132
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.4715340007422714,
+      "learning_rate": 0.00017503967168516426,
+      "loss": 0.2771,
+      "step": 133
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.43502563576445413,
+      "learning_rate": 0.0001746302775541467,
+      "loss": 0.2423,
+      "step": 134
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.4967705692007805,
+      "learning_rate": 0.00017421804041658863,
+      "loss": 0.2498,
+      "step": 135
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.49127370733051945,
+      "learning_rate": 0.00017380297597647667,
+      "loss": 0.2616,
+      "step": 136
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.47835649282708265,
+      "learning_rate": 0.00017338510004550223,
+      "loss": 0.241,
+      "step": 137
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.4843464174553606,
+      "learning_rate": 0.00017296442854245915,
+      "loss": 0.2458,
+      "step": 138
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5209405133977896,
+      "learning_rate": 0.00017254097749263734,
+      "loss": 0.2452,
+      "step": 139
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.4709574288825739,
+      "learning_rate": 0.0001721147630272123,
+      "loss": 0.2627,
+      "step": 140
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.4752105435022234,
+      "learning_rate": 0.00017168580138263062,
+      "loss": 0.2527,
+      "step": 141
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.48781843284289905,
+      "learning_rate": 0.00017125410889999134,
+      "loss": 0.2356,
+      "step": 142
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5731736183258567,
+      "learning_rate": 0.00017081970202442362,
+      "loss": 0.2668,
+      "step": 143
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.48105126464697834,
+      "learning_rate": 0.0001703825973044602,
+      "loss": 0.2454,
+      "step": 144
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5280645599674879,
+      "learning_rate": 0.00016994281139140688,
+      "loss": 0.2454,
+      "step": 145
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.47876489284248624,
+      "learning_rate": 0.0001695003610387084,
+      "loss": 0.2463,
+      "step": 146
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.48826354198860017,
+      "learning_rate": 0.00016905526310130999,
+      "loss": 0.2295,
+      "step": 147
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.47715494831436517,
+      "learning_rate": 0.0001686075345350156,
+      "loss": 0.252,
+      "step": 148
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.5152105233009641,
+      "learning_rate": 0.0001681571923958416,
+      "loss": 0.2771,
+      "step": 149
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.4990883717055415,
+      "learning_rate": 0.00016770425383936735,
+      "loss": 0.2497,
+      "step": 150
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.4674093996422124,
+      "learning_rate": 0.00016724873612008155,
+      "loss": 0.2441,
+      "step": 151
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.4432102664091143,
+      "learning_rate": 0.00016679065659072487,
+      "loss": 0.2418,
+      "step": 152
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.4677926556162063,
+      "learning_rate": 0.00016633003270162902,
+      "loss": 0.2483,
+      "step": 153
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.5050389021999718,
+      "learning_rate": 0.00016586688200005193,
+      "loss": 0.225,
+      "step": 154
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.538150442089787,
+      "learning_rate": 0.00016540122212950934,
+      "loss": 0.2629,
+      "step": 155
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.4831894197759429,
+      "learning_rate": 0.00016493307082910249,
+      "loss": 0.2539,
+      "step": 156
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.4864294249801108,
+      "learning_rate": 0.00016446244593284277,
+      "loss": 0.2638,
+      "step": 157
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.46236092553249764,
+      "learning_rate": 0.00016398936536897183,
+      "loss": 0.2255,
+      "step": 158
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.4963120760517666,
+      "learning_rate": 0.00016351384715927898,
+      "loss": 0.2524,
+      "step": 159
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.5210286477375989,
+      "learning_rate": 0.00016303590941841458,
+      "loss": 0.225,
+      "step": 160
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.5288475623534257,
+      "learning_rate": 0.0001625555703531998,
+      "loss": 0.2428,
+      "step": 161
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.4973215047467683,
+      "learning_rate": 0.00016207284826193335,
+      "loss": 0.2522,
+      "step": 162
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.44826317640998203,
+      "learning_rate": 0.00016158776153369402,
+      "loss": 0.2019,
+      "step": 163
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.45392654459830534,
+      "learning_rate": 0.0001611003286476406,
+      "loss": 0.2338,
+      "step": 164
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.4430521150056381,
+      "learning_rate": 0.00016061056817230754,
+      "loss": 0.2273,
+      "step": 165
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.44345119147374473,
+      "learning_rate": 0.00016011849876489776,
+      "loss": 0.211,
+      "step": 166
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4808061249544928,
+      "learning_rate": 0.000159624139170572,
+      "loss": 0.2104,
+      "step": 167
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.5573402749682285,
+      "learning_rate": 0.00015912750822173445,
+      "loss": 0.2492,
+      "step": 168
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.5334950652460796,
+      "learning_rate": 0.00015862862483731574,
+      "loss": 0.2187,
+      "step": 169
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.49497739813798797,
+      "learning_rate": 0.00015812750802205187,
+      "loss": 0.2097,
+      "step": 170
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.44446540691990566,
+      "learning_rate": 0.00015762417686576038,
+      "loss": 0.204,
+      "step": 171
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.42142200135464725,
+      "learning_rate": 0.0001571186505426132,
+      "loss": 0.1989,
+      "step": 172
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.4328533901196503,
+      "learning_rate": 0.00015661094831040598,
+      "loss": 0.2173,
+      "step": 173
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.43093996542664664,
+      "learning_rate": 0.00015610108950982494,
+      "loss": 0.1865,
+      "step": 174
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.4850613308932528,
+      "learning_rate": 0.00015558909356370944,
+      "loss": 0.2181,
+      "step": 175
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.47485870685329246,
+      "learning_rate": 0.00015507497997631266,
+      "loss": 0.2223,
+      "step": 176
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.42085147271583295,
+      "learning_rate": 0.0001545587683325583,
+      "loss": 0.1845,
+      "step": 177
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.4479801309419239,
+      "learning_rate": 0.00015404047829729457,
+      "loss": 0.1987,
+      "step": 178
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.4624584058381783,
+      "learning_rate": 0.00015352012961454507,
+      "loss": 0.217,
+      "step": 179
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.44005765649196454,
+      "learning_rate": 0.00015299774210675657,
+      "loss": 0.1837,
+      "step": 180
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.4508346255489124,
+      "learning_rate": 0.00015247333567404406,
+      "loss": 0.2007,
+      "step": 181
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.40396006791211914,
+      "learning_rate": 0.00015194693029343248,
+      "loss": 0.1866,
+      "step": 182
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.44558839018398966,
+      "learning_rate": 0.00015141854601809581,
+      "loss": 0.1967,
+      "step": 183
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.4337334328022437,
+      "learning_rate": 0.00015088820297659314,
+      "loss": 0.1891,
+      "step": 184
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.4636781912221849,
+      "learning_rate": 0.00015035592137210187,
+      "loss": 0.193,
+      "step": 185
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.47955885394967973,
+      "learning_rate": 0.00014982172148164804,
+      "loss": 0.1793,
+      "step": 186
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.4721310395975314,
+      "learning_rate": 0.00014928562365533392,
+      "loss": 0.186,
+      "step": 187
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.4737141537120664,
+      "learning_rate": 0.00014874764831556285,
+      "loss": 0.2058,
+      "step": 188
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.40830849621087567,
+      "learning_rate": 0.00014820781595626116,
+      "loss": 0.1822,
+      "step": 189
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.4272142710058541,
+      "learning_rate": 0.0001476661471420975,
+      "loss": 0.2057,
+      "step": 190
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.4212227727031309,
+      "learning_rate": 0.0001471226625076993,
+      "loss": 0.1845,
+      "step": 191
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.39660108389275345,
+      "learning_rate": 0.0001465773827568671,
+      "loss": 0.1769,
+      "step": 192
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.38828383424285384,
+      "learning_rate": 0.00014603032866178538,
+      "loss": 0.1699,
+      "step": 193
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.3681031142044674,
+      "learning_rate": 0.00014548152106223157,
+      "loss": 0.1456,
+      "step": 194
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.46248659870169556,
+      "learning_rate": 0.00014493098086478196,
+      "loss": 0.1846,
+      "step": 195
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.4437664820090981,
+      "learning_rate": 0.00014437872904201542,
+      "loss": 0.1706,
+      "step": 196
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.4410375026146085,
+      "learning_rate": 0.0001438247866317145,
+      "loss": 0.1757,
+      "step": 197
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.4290870801703047,
+      "learning_rate": 0.00014326917473606366,
+      "loss": 0.1777,
+      "step": 198
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.4812130220306999,
+      "learning_rate": 0.00014271191452084597,
+      "loss": 0.2013,
+      "step": 199
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.4314920290891278,
+      "learning_rate": 0.00014215302721463623,
+      "loss": 0.1857,
+      "step": 200
+    },
+    {
+      "epoch": 1.9,
+      "eval_blimp_filtered_avg": 0.7161194029850746,
+      "eval_blimp_filtered_std": 0.005001692965803923,
+      "step": 200
+    },
+    {
+      "epoch": 1.9,
+      "eval_blimp_supplement_avg": 0.8211206896551724,
+      "eval_blimp_supplement_std": 0.016785621805327337,
+      "step": 200
+    },
+    {
+      "epoch": 1.9,
+      "eval_vqa_filtered_avg": 0.51,
+      "eval_vqa_filtered_std": 0.05024183937956912,
+      "step": 200
+    },
+    {
+      "epoch": 1.9,
+      "eval_winoground_filtered_avg": 0.62,
+      "eval_winoground_filtered_std": 0.04878317312145633,
+      "step": 200
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.41562514975066434,
+      "learning_rate": 0.0001415925341079927,
+      "loss": 0.21,
+      "step": 201
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.37833993286875955,
+      "learning_rate": 0.00014103045655264576,
+      "loss": 0.1659,
+      "step": 202
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.3880529818353851,
+      "learning_rate": 0.00014046681596068466,
+      "loss": 0.1638,
+      "step": 203
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.40159118156434603,
+      "learning_rate": 0.00013990163380374194,
+      "loss": 0.1768,
+      "step": 204
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.4086449128732129,
+      "learning_rate": 0.00013933493161217523,
+      "loss": 0.1544,
+      "step": 205
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3808287729283849,
+      "learning_rate": 0.0001387667309742472,
+      "loss": 0.1366,
+      "step": 206
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.39609061286446773,
+      "learning_rate": 0.0001381970535353032,
+      "loss": 0.1494,
+      "step": 207
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.40847272653729905,
+      "learning_rate": 0.00013762592099694665,
+      "loss": 0.1615,
+      "step": 208
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.4334994696681873,
+      "learning_rate": 0.00013705335511621228,
+      "loss": 0.1542,
+      "step": 209
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.4546384761691546,
+      "learning_rate": 0.00013647937770473737,
+      "loss": 0.1834,
+      "step": 210
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.36130610610645814,
+      "learning_rate": 0.00013590401062793083,
+      "loss": 0.123,
+      "step": 211
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.29975302946848653,
+      "learning_rate": 0.0001353272758041402,
+      "loss": 0.0824,
+      "step": 212
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.29392603086414587,
+      "learning_rate": 0.00013474919520381671,
+      "loss": 0.0836,
+      "step": 213
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.33169221984700814,
+      "learning_rate": 0.00013416979084867852,
+      "loss": 0.0683,
+      "step": 214
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.39192700338704206,
+      "learning_rate": 0.00013358908481087134,
+      "loss": 0.0804,
+      "step": 215
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.42443737109460977,
+      "learning_rate": 0.0001330070992121281,
+      "loss": 0.0797,
+      "step": 216
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.42848813761714244,
+      "learning_rate": 0.00013242385622292592,
+      "loss": 0.0776,
+      "step": 217
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.37448633759803696,
+      "learning_rate": 0.00013183937806164172,
+      "loss": 0.0739,
+      "step": 218
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.3437440816482259,
+      "learning_rate": 0.00013125368699370567,
+      "loss": 0.0652,
+      "step": 219
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.356415907025676,
+      "learning_rate": 0.0001306668053307531,
+      "loss": 0.0778,
+      "step": 220
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.30675625825005026,
+      "learning_rate": 0.00013007875542977448,
+      "loss": 0.0665,
+      "step": 221
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.29794655672460485,
+      "learning_rate": 0.00012948955969226383,
+      "loss": 0.0696,
+      "step": 222
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.30163505061461343,
+      "learning_rate": 0.00012889924056336532,
+      "loss": 0.0705,
+      "step": 223
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.32541739323213426,
+      "learning_rate": 0.00012830782053101805,
+      "loss": 0.0733,
+      "step": 224
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.31121536090331003,
+      "learning_rate": 0.00012771532212509974,
+      "loss": 0.0711,
+      "step": 225
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.34593292210442944,
+      "learning_rate": 0.00012712176791656807,
+      "loss": 0.0788,
+      "step": 226
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.33946278651997686,
+      "learning_rate": 0.0001265271805166012,
+      "loss": 0.0677,
+      "step": 227
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.3400898219352628,
+      "learning_rate": 0.0001259315825757362,
+      "loss": 0.0643,
+      "step": 228
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.3813085350755264,
+      "learning_rate": 0.00012533499678300618,
+      "loss": 0.0761,
+      "step": 229
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.3523012248149677,
+      "learning_rate": 0.00012473744586507604,
+      "loss": 0.0648,
+      "step": 230
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.37842862853695125,
+      "learning_rate": 0.00012413895258537675,
+      "loss": 0.0812,
+      "step": 231
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.39475455813661525,
+      "learning_rate": 0.00012353953974323807,
+      "loss": 0.0801,
+      "step": 232
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.3205081471986943,
+      "learning_rate": 0.00012293923017302002,
+      "loss": 0.0677,
+      "step": 233
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.31006899448135294,
+      "learning_rate": 0.0001223380467432432,
+      "loss": 0.07,
+      "step": 234
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.3048520942780853,
+      "learning_rate": 0.00012173601235571742,
+      "loss": 0.0615,
+      "step": 235
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.3425413653893973,
+      "learning_rate": 0.0001211331499446693,
+      "loss": 0.0658,
+      "step": 236
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.31929344956491607,
+      "learning_rate": 0.00012052948247586873,
+      "loss": 0.0653,
+      "step": 237
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.3414359773691709,
+      "learning_rate": 0.00011992503294575383,
+      "loss": 0.0723,
+      "step": 238
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.32978160245312554,
+      "learning_rate": 0.00011931982438055505,
+      "loss": 0.07,
+      "step": 239
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.33271868205929617,
+      "learning_rate": 0.00011871387983541789,
+      "loss": 0.0672,
+      "step": 240
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.29862145989444433,
+      "learning_rate": 0.00011810722239352467,
+      "loss": 0.0603,
+      "step": 241
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.34485364985513034,
+      "learning_rate": 0.00011749987516521523,
+      "loss": 0.0632,
+      "step": 242
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.3299899118013224,
+      "learning_rate": 0.00011689186128710654,
+      "loss": 0.0601,
+      "step": 243
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.29635972892096896,
+      "learning_rate": 0.00011628320392121117,
+      "loss": 0.0558,
+      "step": 244
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.3414458592363874,
+      "learning_rate": 0.0001156739262540552,
+      "loss": 0.0703,
+      "step": 245
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.3280087622706941,
+      "learning_rate": 0.00011506405149579468,
+      "loss": 0.0657,
+      "step": 246
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.373086375777386,
+      "learning_rate": 0.00011445360287933165,
+      "loss": 0.0668,
+      "step": 247
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.2937645914714354,
+      "learning_rate": 0.00011384260365942904,
+      "loss": 0.0612,
+      "step": 248
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.39022311054047737,
+      "learning_rate": 0.00011323107711182473,
+      "loss": 0.0762,
+      "step": 249
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.3345521008714258,
+      "learning_rate": 0.00011261904653234485,
+      "loss": 0.0711,
+      "step": 250
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.30608871062806836,
+      "learning_rate": 0.00011200653523601652,
+      "loss": 0.0617,
+      "step": 251
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.30714147902477945,
+      "learning_rate": 0.00011139356655617945,
+      "loss": 0.063,
+      "step": 252
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.31051190204375445,
+      "learning_rate": 0.00011078016384359724,
+      "loss": 0.0659,
+      "step": 253
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.3071085278813772,
+      "learning_rate": 0.00011016635046556772,
+      "loss": 0.061,
+      "step": 254
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.3045837343462885,
+      "learning_rate": 0.00010955214980503284,
+      "loss": 0.0597,
+      "step": 255
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.3049959198680976,
+      "learning_rate": 0.00010893758525968789,
+      "loss": 0.0587,
+      "step": 256
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.3168437149994661,
+      "learning_rate": 0.00010832268024109025,
+      "loss": 0.0559,
+      "step": 257
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.3024342626013227,
+      "learning_rate": 0.00010770745817376742,
+      "loss": 0.0583,
+      "step": 258
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3188509232471995,
+      "learning_rate": 0.0001070919424943247,
+      "loss": 0.061,
+      "step": 259
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.3381945814712772,
+      "learning_rate": 0.0001064761566505525,
+      "loss": 0.0648,
+      "step": 260
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.3131931451431926,
+      "learning_rate": 0.00010586012410053292,
+      "loss": 0.0624,
+      "step": 261
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.32809637984753304,
+      "learning_rate": 0.00010524386831174628,
+      "loss": 0.0627,
+      "step": 262
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.2832796499168925,
+      "learning_rate": 0.00010462741276017711,
+      "loss": 0.0535,
+      "step": 263
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.3334141162384235,
+      "learning_rate": 0.00010401078092941971,
+      "loss": 0.061,
+      "step": 264
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.27653747850590626,
+      "learning_rate": 0.00010339399630978373,
+      "loss": 0.0497,
+      "step": 265
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.32205480409336124,
+      "learning_rate": 0.00010277708239739924,
+      "loss": 0.0658,
+      "step": 266
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.310079147965717,
+      "learning_rate": 0.0001021600626933217,
+      "loss": 0.0525,
+      "step": 267
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.31094425691461797,
+      "learning_rate": 0.00010154296070263649,
+      "loss": 0.0619,
+      "step": 268
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.33419799536496597,
+      "learning_rate": 0.00010092579993356386,
+      "loss": 0.0615,
+      "step": 269
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.3343121767672678,
+      "learning_rate": 0.00010030860389656305,
+      "loss": 0.0663,
+      "step": 270
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.3516117623617434,
+      "learning_rate": 9.969139610343696e-05,
+      "loss": 0.0662,
+      "step": 271
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.31796912631433194,
+      "learning_rate": 9.907420006643619e-05,
+      "loss": 0.0624,
+      "step": 272
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.29219460425245597,
+      "learning_rate": 9.845703929736351e-05,
+      "loss": 0.0596,
+      "step": 273
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.316635170830544,
+      "learning_rate": 9.783993730667831e-05,
+      "loss": 0.0659,
+      "step": 274
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.33766616368603597,
+      "learning_rate": 9.722291760260077e-05,
+      "loss": 0.0646,
+      "step": 275
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.31287192455811574,
+      "learning_rate": 9.66060036902163e-05,
+      "loss": 0.0585,
+      "step": 276
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.28964582015181484,
+      "learning_rate": 9.598921907058033e-05,
+      "loss": 0.0543,
+      "step": 277
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.3037919396698326,
+      "learning_rate": 9.53725872398229e-05,
+      "loss": 0.0512,
+      "step": 278
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.3229974938313004,
+      "learning_rate": 9.475613168825374e-05,
+      "loss": 0.0531,
+      "step": 279
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.29881091304580676,
+      "learning_rate": 9.413987589946711e-05,
+      "loss": 0.0569,
+      "step": 280
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.29692909307641674,
+      "learning_rate": 9.352384334944753e-05,
+      "loss": 0.0547,
+      "step": 281
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.33439942628885455,
+      "learning_rate": 9.290805750567532e-05,
+      "loss": 0.0622,
+      "step": 282
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.2991141437988068,
+      "learning_rate": 9.22925418262326e-05,
+      "loss": 0.0464,
+      "step": 283
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.3171911760038229,
+      "learning_rate": 9.167731975890976e-05,
+      "loss": 0.059,
+      "step": 284
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.30072460150102115,
+      "learning_rate": 9.106241474031212e-05,
+      "loss": 0.0559,
+      "step": 285
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.3301896190647226,
+      "learning_rate": 9.04478501949672e-05,
+      "loss": 0.0514,
+      "step": 286
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.3298071637508188,
+      "learning_rate": 8.983364953443227e-05,
+      "loss": 0.0618,
+      "step": 287
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.3497185839244567,
+      "learning_rate": 8.921983615640277e-05,
+      "loss": 0.065,
+      "step": 288
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.33084725547728233,
+      "learning_rate": 8.860643344382056e-05,
+      "loss": 0.0527,
+      "step": 289
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.33012822636415956,
+      "learning_rate": 8.79934647639835e-05,
+      "loss": 0.0666,
+      "step": 290
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.3151687548518561,
+      "learning_rate": 8.738095346765518e-05,
+      "loss": 0.0573,
+      "step": 291
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.30346203875619676,
+      "learning_rate": 8.676892288817531e-05,
+      "loss": 0.0491,
+      "step": 292
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.3133369298353677,
+      "learning_rate": 8.615739634057098e-05,
+      "loss": 0.0595,
+      "step": 293
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.28715782085999497,
+      "learning_rate": 8.554639712066836e-05,
+      "loss": 0.0542,
+      "step": 294
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.2815995010771035,
+      "learning_rate": 8.493594850420537e-05,
+      "loss": 0.0551,
+      "step": 295
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.280576878443274,
+      "learning_rate": 8.432607374594484e-05,
+      "loss": 0.0488,
+      "step": 296
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.298809991890747,
+      "learning_rate": 8.371679607878884e-05,
+      "loss": 0.0544,
+      "step": 297
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.30088222272143067,
+      "learning_rate": 8.310813871289348e-05,
+      "loss": 0.0591,
+      "step": 298
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.3237358977236424,
+      "learning_rate": 8.250012483478478e-05,
+      "loss": 0.0547,
+      "step": 299
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.34075237005827885,
+      "learning_rate": 8.189277760647537e-05,
+      "loss": 0.0566,
+      "step": 300
+    },
+    {
+      "epoch": 2.85,
+      "eval_blimp_filtered_avg": 0.7037313432835821,
+      "eval_blimp_filtered_std": 0.005058972315437875,
+      "step": 300
+    },
+    {
+      "epoch": 2.85,
+      "eval_blimp_supplement_avg": 0.8103448275862069,
+      "eval_blimp_supplement_std": 0.017321145118445798,
+      "step": 300
+    },
+    {
+      "epoch": 2.85,
+      "eval_vqa_filtered_avg": 0.53,
+      "eval_vqa_filtered_std": 0.0501613558046592,
+      "step": 300
+    },
+    {
+      "epoch": 2.85,
+      "eval_winoground_filtered_avg": 0.68,
+      "eval_winoground_filtered_std": 0.046882617226215034,
+      "step": 300
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.3237263865460515,
+      "learning_rate": 8.128612016458215e-05,
+      "loss": 0.059,
+      "step": 301
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.2977357286247905,
+      "learning_rate": 8.068017561944499e-05,
+      "loss": 0.0492,
+      "step": 302
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.29591506818063545,
+      "learning_rate": 8.00749670542462e-05,
+      "loss": 0.052,
+      "step": 303
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.2789469075911483,
+      "learning_rate": 7.94705175241313e-05,
+      "loss": 0.0455,
+      "step": 304
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.2997082343784124,
+      "learning_rate": 7.886685005533072e-05,
+      "loss": 0.0498,
+      "step": 305
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.30157528073661777,
+      "learning_rate": 7.82639876442826e-05,
+      "loss": 0.0567,
+      "step": 306
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.32803298910194756,
+      "learning_rate": 7.76619532567568e-05,
+      "loss": 0.0622,
+      "step": 307
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.28556449374878695,
+      "learning_rate": 7.706076982697999e-05,
+      "loss": 0.0489,
+      "step": 308
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.32287162854623286,
+      "learning_rate": 7.646046025676198e-05,
+      "loss": 0.066,
+      "step": 309
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.3384064716667544,
+      "learning_rate": 7.586104741462325e-05,
+      "loss": 0.0629,
+      "step": 310
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.3005901634146794,
+      "learning_rate": 7.526255413492395e-05,
+      "loss": 0.051,
+      "step": 311
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.2907146546357962,
+      "learning_rate": 7.466500321699383e-05,
+      "loss": 0.0546,
+      "step": 312
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.30779520364750435,
+      "learning_rate": 7.40684174242638e-05,
+      "loss": 0.058,
+      "step": 313
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.29074373091101263,
+      "learning_rate": 7.347281948339879e-05,
+      "loss": 0.0463,
+      "step": 314
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 0.32970798475445445,
+      "learning_rate": 7.287823208343192e-05,
+      "loss": 0.0589,
+      "step": 315
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.2798345327195924,
+      "learning_rate": 7.228467787490028e-05,
+      "loss": 0.0438,
+      "step": 316
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.18326848967204043,
+      "learning_rate": 7.169217946898197e-05,
+      "loss": 0.0225,
+      "step": 317
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 0.18022372679373735,
+      "learning_rate": 7.110075943663472e-05,
+      "loss": 0.0161,
+      "step": 318
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.1633153575928502,
+      "learning_rate": 7.051044030773618e-05,
+      "loss": 0.0153,
+      "step": 319
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 0.17802284328446474,
+      "learning_rate": 6.992124457022553e-05,
+      "loss": 0.0176,
+      "step": 320
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.17359891604740127,
+      "learning_rate": 6.933319466924693e-05,
+      "loss": 0.0162,
+      "step": 321
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 0.2202987501804585,
+      "learning_rate": 6.874631300629435e-05,
+      "loss": 0.0162,
+      "step": 322
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 0.22277821921264357,
+      "learning_rate": 6.81606219383583e-05,
+      "loss": 0.0187,
+      "step": 323
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.18724963681022663,
+      "learning_rate": 6.757614377707409e-05,
+      "loss": 0.0153,
+      "step": 324
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 0.21995220887794256,
+      "learning_rate": 6.699290078787193e-05,
+      "loss": 0.0188,
+      "step": 325
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.1967935793635855,
+      "learning_rate": 6.641091518912867e-05,
+      "loss": 0.0156,
+      "step": 326
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 0.20661934683104752,
+      "learning_rate": 6.583020915132152e-05,
+      "loss": 0.0158,
+      "step": 327
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.2422474266231083,
+      "learning_rate": 6.525080479618331e-05,
+      "loss": 0.0177,
+      "step": 328
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 0.18354685059507367,
+      "learning_rate": 6.467272419585984e-05,
+      "loss": 0.013,
+      "step": 329
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.22423754187379397,
+      "learning_rate": 6.40959893720692e-05,
+      "loss": 0.0188,
+      "step": 330
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.18994008796265852,
+      "learning_rate": 6.352062229526266e-05,
+      "loss": 0.0132,
+      "step": 331
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 0.24715301748493912,
+      "learning_rate": 6.294664488378776e-05,
+      "loss": 0.015,
+      "step": 332
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 0.17280498203848704,
+      "learning_rate": 6.237407900305335e-05,
+      "loss": 0.0138,
+      "step": 333
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.21773200395950232,
+      "learning_rate": 6.180294646469679e-05,
+      "loss": 0.0155,
+      "step": 334
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 0.2144971485793242,
+      "learning_rate": 6.123326902575282e-05,
+      "loss": 0.0158,
+      "step": 335
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.18331926033535073,
+      "learning_rate": 6.06650683878248e-05,
+      "loss": 0.013,
+      "step": 336
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.1788180130126268,
+      "learning_rate": 6.009836619625809e-05,
+      "loss": 0.0133,
+      "step": 337
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.20337677688861636,
+      "learning_rate": 5.953318403931532e-05,
+      "loss": 0.0129,
+      "step": 338
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 0.20853998405220736,
+      "learning_rate": 5.896954344735426e-05,
+      "loss": 0.0176,
+      "step": 339
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 0.1919639102705018,
+      "learning_rate": 5.840746589200732e-05,
+      "loss": 0.0144,
+      "step": 340
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.2134469059873606,
+      "learning_rate": 5.784697278536379e-05,
+      "loss": 0.0138,
+      "step": 341
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 0.18435084201272836,
+      "learning_rate": 5.728808547915405e-05,
+      "loss": 0.0135,
+      "step": 342
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.19554570393158438,
+      "learning_rate": 5.673082526393634e-05,
+      "loss": 0.015,
+      "step": 343
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 0.18522448379098544,
+      "learning_rate": 5.617521336828556e-05,
+      "loss": 0.0129,
+      "step": 344
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 0.190207008998555,
+      "learning_rate": 5.5621270957984573e-05,
+      "loss": 0.0161,
+      "step": 345
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 0.19594053008897275,
+      "learning_rate": 5.506901913521808e-05,
+      "loss": 0.0162,
+      "step": 346
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.20111569255746164,
+      "learning_rate": 5.451847893776845e-05,
+      "loss": 0.0147,
+      "step": 347
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 0.20867562278084897,
+      "learning_rate": 5.396967133821461e-05,
+      "loss": 0.0154,
+      "step": 348
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 0.16028325232055693,
+      "learning_rate": 5.342261724313292e-05,
+      "loss": 0.0117,
+      "step": 349
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.14992620939570764,
+      "learning_rate": 5.28773374923007e-05,
+      "loss": 0.0106,
+      "step": 350
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.20669460754401175,
+      "learning_rate": 5.2333852857902575e-05,
+      "loss": 0.0161,
+      "step": 351
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 0.21934716169620833,
+      "learning_rate": 5.1792184043738855e-05,
+      "loss": 0.0128,
+      "step": 352
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.18204794157825063,
+      "learning_rate": 5.1252351684437136e-05,
+      "loss": 0.0129,
+      "step": 353
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 0.21363608639584963,
+      "learning_rate": 5.071437634466609e-05,
+      "loss": 0.0105,
+      "step": 354
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 0.15881770971724649,
+      "learning_rate": 5.0178278518351983e-05,
+      "loss": 0.0096,
+      "step": 355
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 0.1980006966366768,
+      "learning_rate": 4.964407862789817e-05,
+      "loss": 0.0119,
+      "step": 356
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 0.21004802159627842,
+      "learning_rate": 4.911179702340688e-05,
+      "loss": 0.0119,
+      "step": 357
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 0.20419756258161648,
+      "learning_rate": 4.85814539819042e-05,
+      "loss": 0.0145,
+      "step": 358
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 0.1565818058300373,
+      "learning_rate": 4.8053069706567554e-05,
+      "loss": 0.0105,
+      "step": 359
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 0.19501698471957343,
+      "learning_rate": 4.752666432595596e-05,
+      "loss": 0.0126,
+      "step": 360
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 0.20941486180216556,
+      "learning_rate": 4.700225789324343e-05,
+      "loss": 0.0105,
+      "step": 361
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.18304197382791004,
+      "learning_rate": 4.647987038545496e-05,
+      "loss": 0.011,
+      "step": 362
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 0.16720171411001336,
+      "learning_rate": 4.595952170270542e-05,
+      "loss": 0.0112,
+      "step": 363
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.22478251297433013,
+      "learning_rate": 4.544123166744172e-05,
+      "loss": 0.0118,
+      "step": 364
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 0.1598572948562243,
+      "learning_rate": 4.492502002368738e-05,
+      "loss": 0.0107,
+      "step": 365
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 0.22373563049772874,
+      "learning_rate": 4.4410906436290566e-05,
+      "loss": 0.0104,
+      "step": 366
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 0.16802667132434534,
+      "learning_rate": 4.38989104901751e-05,
+      "loss": 0.0114,
+      "step": 367
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 0.24550738449688075,
+      "learning_rate": 4.3389051689594e-05,
+      "loss": 0.0121,
+      "step": 368
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.1660066244443363,
+      "learning_rate": 4.288134945738684e-05,
+      "loss": 0.0099,
+      "step": 369
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.1783889244909253,
+      "learning_rate": 4.237582313423962e-05,
+      "loss": 0.0094,
+      "step": 370
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.17141038466777303,
+      "learning_rate": 4.187249197794813e-05,
+      "loss": 0.0095,
+      "step": 371
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 0.1893721805088239,
+      "learning_rate": 4.137137516268426e-05,
+      "loss": 0.013,
+      "step": 372
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 0.16935951673752134,
+      "learning_rate": 4.0872491778265535e-05,
+      "loss": 0.0091,
+      "step": 373
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.13309068523326859,
+      "learning_rate": 4.037586082942805e-05,
+      "loss": 0.0091,
+      "step": 374
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 0.18791651271841342,
+      "learning_rate": 3.988150123510224e-05,
+      "loss": 0.0121,
+      "step": 375
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 0.1559825545952661,
+      "learning_rate": 3.938943182769246e-05,
+      "loss": 0.0102,
+      "step": 376
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 0.2261919531211638,
+      "learning_rate": 3.88996713523594e-05,
+      "loss": 0.0127,
+      "step": 377
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 0.20792420146527377,
+      "learning_rate": 3.841223846630599e-05,
+      "loss": 0.013,
+      "step": 378
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.16486082885129608,
+      "learning_rate": 3.792715173806669e-05,
+      "loss": 0.0105,
+      "step": 379
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 0.1549020176177142,
+      "learning_rate": 3.74444296468002e-05,
+      "loss": 0.0098,
+      "step": 380
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.17250200199106172,
+      "learning_rate": 3.696409058158544e-05,
+      "loss": 0.0109,
+      "step": 381
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 0.1415293330470341,
+      "learning_rate": 3.6486152840721046e-05,
+      "loss": 0.0084,
+      "step": 382
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 0.14461810975420877,
+      "learning_rate": 3.6010634631028226e-05,
+      "loss": 0.0084,
+      "step": 383
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.1557012557289619,
+      "learning_rate": 3.553755406715724e-05,
+      "loss": 0.0089,
+      "step": 384
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 0.15752891661687976,
+      "learning_rate": 3.506692917089751e-05,
+      "loss": 0.0109,
+      "step": 385
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 0.1694876915505117,
+      "learning_rate": 3.459877787049072e-05,
+      "loss": 0.009,
+      "step": 386
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 0.1582663784415179,
+      "learning_rate": 3.413311799994808e-05,
+      "loss": 0.0095,
+      "step": 387
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 0.13693031068741818,
+      "learning_rate": 3.366996729837102e-05,
+      "loss": 0.0092,
+      "step": 388
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 0.14543112940410688,
+      "learning_rate": 3.320934340927513e-05,
+      "loss": 0.0108,
+      "step": 389
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.19389482832864774,
+      "learning_rate": 3.275126387991847e-05,
+      "loss": 0.0098,
+      "step": 390
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.15797165592004603,
+      "learning_rate": 3.229574616063268e-05,
+      "loss": 0.0076,
+      "step": 391
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 0.21281942854700847,
+      "learning_rate": 3.184280760415843e-05,
+      "loss": 0.0142,
+      "step": 392
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 0.12498130411986656,
+      "learning_rate": 3.1392465464984455e-05,
+      "loss": 0.0081,
+      "step": 393
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 0.1152125429659436,
+      "learning_rate": 3.094473689869002e-05,
+      "loss": 0.0058,
+      "step": 394
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.1567733530080216,
+      "learning_rate": 3.0499638961291623e-05,
+      "loss": 0.011,
+      "step": 395
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 0.14500898906990572,
+      "learning_rate": 3.0057188608593147e-05,
+      "loss": 0.0085,
+      "step": 396
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 0.16163974543952728,
+      "learning_rate": 2.9617402695539808e-05,
+      "loss": 0.013,
+      "step": 397
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 0.13868168811451842,
+      "learning_rate": 2.9180297975576364e-05,
+      "loss": 0.0084,
+      "step": 398
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 0.17847032901949134,
+      "learning_rate": 2.8745891100008683e-05,
+      "loss": 0.0121,
+      "step": 399
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.17527442252411723,
+      "learning_rate": 2.83141986173694e-05,
+      "loss": 0.0084,
+      "step": 400
+    },
+    {
+      "epoch": 3.8,
+      "eval_blimp_filtered_avg": 0.7053731343283582,
+      "eval_blimp_filtered_std": 0.005043001462199571,
+      "step": 400
+    },
+    {
+      "epoch": 3.8,
+      "eval_blimp_supplement_avg": 0.8125,
+      "eval_blimp_supplement_std": 0.01736311122127593,
+      "step": 400
+    },
+    {
+      "epoch": 3.8,
+      "eval_vqa_filtered_avg": 0.52,
+      "eval_vqa_filtered_std": 0.05021167315686779,
+      "step": 400
+    },
+    {
+      "epoch": 3.8,
+      "eval_winoground_filtered_avg": 0.64,
+      "eval_winoground_filtered_std": 0.048241815132442176,
+      "step": 400
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 0.14598157841040266,
+      "learning_rate": 2.788523697278773e-05,
+      "loss": 0.0093,
+      "step": 401
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 0.20150542514971506,
+      "learning_rate": 2.7459022507362686e-05,
+      "loss": 0.0122,
+      "step": 402
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 0.18255123614923588,
+      "learning_rate": 2.7035571457540865e-05,
+      "loss": 0.0103,
+      "step": 403
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.16704045474943452,
+      "learning_rate": 2.6614899954497795e-05,
+      "loss": 0.0114,
+      "step": 404
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 0.14683721625679494,
+      "learning_rate": 2.619702402352332e-05,
+      "loss": 0.01,
+      "step": 405
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 0.18144743721435366,
+      "learning_rate": 2.5781959583411374e-05,
+      "loss": 0.0129,
+      "step": 406
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 0.19646570441433073,
+      "learning_rate": 2.5369722445853304e-05,
+      "loss": 0.0143,
+      "step": 407
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 0.1668088181727681,
+      "learning_rate": 2.4960328314835745e-05,
+      "loss": 0.0089,
+      "step": 408
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 0.16111476451284476,
+      "learning_rate": 2.4553792786042262e-05,
+      "loss": 0.0091,
+      "step": 409
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.17729690845562673,
+      "learning_rate": 2.4150131346259197e-05,
+      "loss": 0.0103,
+      "step": 410
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.15155895346947004,
+      "learning_rate": 2.3749359372785883e-05,
+      "loss": 0.0096,
+      "step": 411
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 0.15041370885333255,
+      "learning_rate": 2.3351492132848664e-05,
+      "loss": 0.0085,
+      "step": 412
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 0.12197907148956355,
+      "learning_rate": 2.2956544783019418e-05,
+      "loss": 0.0067,
+      "step": 413
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 0.1788434056496877,
+      "learning_rate": 2.2564532368638146e-05,
+      "loss": 0.01,
+      "step": 414
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 0.19269466130772045,
+      "learning_rate": 2.2175469823239768e-05,
+      "loss": 0.0117,
+      "step": 415
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 0.15780826445252463,
+      "learning_rate": 2.1789371967985338e-05,
+      "loss": 0.0101,
+      "step": 416
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 0.19229144408434373,
+      "learning_rate": 2.140625351109733e-05,
+      "loss": 0.0084,
+      "step": 417
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 0.15474486143047034,
+      "learning_rate": 2.1026129047299436e-05,
+      "loss": 0.0067,
+      "step": 418
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 0.15864166155594778,
+      "learning_rate": 2.0649013057260546e-05,
+      "loss": 0.0098,
+      "step": 419
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 0.22515244613844015,
+      "learning_rate": 2.0274919907043033e-05,
+      "loss": 0.0094,
+      "step": 420
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.18684872878382638,
+      "learning_rate": 1.9903863847555648e-05,
+      "loss": 0.0127,
+      "step": 421
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 0.06270483785922072,
+      "learning_rate": 1.9535859014010526e-05,
+      "loss": 0.0028,
+      "step": 422
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 0.09948637260912774,
+      "learning_rate": 1.917091942538469e-05,
+      "loss": 0.0037,
+      "step": 423
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 0.07530065845248647,
+      "learning_rate": 1.880905898388612e-05,
+      "loss": 0.0039,
+      "step": 424
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 0.054461890750773165,
+      "learning_rate": 1.8450291474423998e-05,
+      "loss": 0.0025,
+      "step": 425
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 0.08002877578075594,
+      "learning_rate": 1.8094630564083736e-05,
+      "loss": 0.0035,
+      "step": 426
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 0.05746226463965698,
+      "learning_rate": 1.7742089801606276e-05,
+      "loss": 0.0025,
+      "step": 427
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 0.0633358139605444,
+      "learning_rate": 1.7392682616871837e-05,
+      "loss": 0.0027,
+      "step": 428
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 0.06509683268742919,
+      "learning_rate": 1.7046422320388556e-05,
+      "loss": 0.0027,
+      "step": 429
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 0.054571154616853274,
+      "learning_rate": 1.6703322102785168e-05,
+      "loss": 0.0026,
+      "step": 430
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 0.06888564779650448,
+      "learning_rate": 1.6363395034308703e-05,
+      "loss": 0.0027,
+      "step": 431
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 0.05307117129834359,
+      "learning_rate": 1.6026654064326553e-05,
+      "loss": 0.0025,
+      "step": 432
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 0.06598879328529111,
+      "learning_rate": 1.5693112020833013e-05,
+      "loss": 0.003,
+      "step": 433
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 0.054752236275106794,
+      "learning_rate": 1.5362781609960852e-05,
+      "loss": 0.0025,
+      "step": 434
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 0.07106963888787232,
+      "learning_rate": 1.5035675415497063e-05,
+      "loss": 0.0031,
+      "step": 435
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 0.052548572683446884,
+      "learning_rate": 1.471180589840363e-05,
+      "loss": 0.0025,
+      "step": 436
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 0.08828036910254508,
+      "learning_rate": 1.4391185396342789e-05,
+      "loss": 0.0038,
+      "step": 437
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 0.09463459893212552,
+      "learning_rate": 1.4073826123206946e-05,
+      "loss": 0.0038,
+      "step": 438
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 0.08002928457971342,
+      "learning_rate": 1.375974016865359e-05,
+      "loss": 0.0031,
+      "step": 439
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 0.07631532690730236,
+      "learning_rate": 1.3448939497644509e-05,
+      "loss": 0.0031,
+      "step": 440
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 0.04831761603516682,
+      "learning_rate": 1.3141435949990188e-05,
+      "loss": 0.0027,
+      "step": 441
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 0.07344003153336562,
+      "learning_rate": 1.2837241239898667e-05,
+      "loss": 0.0032,
+      "step": 442
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 0.08305075630986966,
+      "learning_rate": 1.253636695552931e-05,
+      "loss": 0.003,
+      "step": 443
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 0.1034575433958594,
+      "learning_rate": 1.2238824558551365e-05,
+      "loss": 0.0039,
+      "step": 444
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 0.06655324788558148,
+      "learning_rate": 1.1944625383707374e-05,
+      "loss": 0.003,
+      "step": 445
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 0.0790599253839735,
+      "learning_rate": 1.1653780638381328e-05,
+      "loss": 0.0029,
+      "step": 446
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 0.04198685628145689,
+      "learning_rate": 1.1366301402171775e-05,
+      "loss": 0.0017,
+      "step": 447
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 0.06439353264983554,
+      "learning_rate": 1.1082198626469686e-05,
+      "loss": 0.0024,
+      "step": 448
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 0.07762450043477247,
+      "learning_rate": 1.0801483134041268e-05,
+      "loss": 0.0027,
+      "step": 449
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 0.07856883953783565,
+      "learning_rate": 1.0524165618615845e-05,
+      "loss": 0.0033,
+      "step": 450
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 0.07929308057852809,
+      "learning_rate": 1.0250256644478195e-05,
+      "loss": 0.003,
+      "step": 451
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 0.0587512154822952,
+      "learning_rate": 9.979766646066368e-06,
+      "loss": 0.0027,
+      "step": 452
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 0.06109551507247056,
+      "learning_rate": 9.71270592757404e-06,
+      "loss": 0.0032,
+      "step": 453
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 0.05909029031199419,
+      "learning_rate": 9.449084662557982e-06,
+      "loss": 0.0026,
+      "step": 454
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 0.0814055458144323,
+      "learning_rate": 9.188912893550695e-06,
+      "loss": 0.0026,
+      "step": 455
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 0.07735385332942207,
+      "learning_rate": 8.932200531677537e-06,
+      "loss": 0.0028,
+      "step": 456
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 0.08519595591969155,
+      "learning_rate": 8.678957356279371e-06,
+      "loss": 0.0024,
+      "step": 457
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 0.055031384326470804,
+      "learning_rate": 8.429193014540015e-06,
+      "loss": 0.0026,
+      "step": 458
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 0.05387324401647046,
+      "learning_rate": 8.182917021118663e-06,
+      "loss": 0.0026,
+      "step": 459
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 0.07168879976269556,
+      "learning_rate": 7.940138757787507e-06,
+      "loss": 0.0032,
+      "step": 460
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 0.07661756681904786,
+      "learning_rate": 7.700867473074224e-06,
+      "loss": 0.0035,
+      "step": 461
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 0.09486930411075328,
+      "learning_rate": 7.46511228190977e-06,
+      "loss": 0.0049,
+      "step": 462
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 0.0679530025111762,
+      "learning_rate": 7.232882165281141e-06,
+      "loss": 0.0026,
+      "step": 463
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 0.06514922044267304,
+      "learning_rate": 7.004185969889187e-06,
+      "loss": 0.0027,
+      "step": 464
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 0.06706026131022384,
+      "learning_rate": 6.7790324078116364e-06,
+      "loss": 0.0027,
+      "step": 465
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 0.07709046890424658,
+      "learning_rate": 6.557430056171221e-06,
+      "loss": 0.0033,
+      "step": 466
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 0.051443041020356704,
+      "learning_rate": 6.339387356808912e-06,
+      "loss": 0.0026,
+      "step": 467
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 0.060318722923432995,
+      "learning_rate": 6.124912615962341e-06,
+      "loss": 0.0028,
+      "step": 468
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 0.062212012735137795,
+      "learning_rate": 5.9140140039494084e-06,
+      "loss": 0.0025,
+      "step": 469
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 0.06556299474776538,
+      "learning_rate": 5.706699554856964e-06,
+      "loss": 0.0023,
+      "step": 470
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 0.08649267044276539,
+      "learning_rate": 5.502977166234857e-06,
+      "loss": 0.0035,
+      "step": 471
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 0.08526822145924882,
+      "learning_rate": 5.302854598794937e-06,
+      "loss": 0.003,
+      "step": 472
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 0.04133711118453636,
+      "learning_rate": 5.106339476115596e-06,
+      "loss": 0.0019,
+      "step": 473
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 0.05708577094578342,
+      "learning_rate": 4.913439284351207e-06,
+      "loss": 0.0026,
+      "step": 474
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 0.07367912633186298,
+      "learning_rate": 4.724161371946978e-06,
+      "loss": 0.0029,
+      "step": 475
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 0.08135320771271103,
+      "learning_rate": 4.538512949359075e-06,
+      "loss": 0.0027,
+      "step": 476
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 0.0849858165893086,
+      "learning_rate": 4.356501088779841e-06,
+      "loss": 0.0027,
+      "step": 477
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 0.05260609110954984,
+      "learning_rate": 4.178132723868477e-06,
+      "loss": 0.0019,
+      "step": 478
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 0.0795477617292828,
+      "learning_rate": 4.003414649486892e-06,
+      "loss": 0.0032,
+      "step": 479
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 0.08161922179718771,
+      "learning_rate": 3.832353521440768e-06,
+      "loss": 0.0026,
+      "step": 480
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 0.06830643544893618,
+      "learning_rate": 3.6649558562261375e-06,
+      "loss": 0.0032,
+      "step": 481
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 0.08641205617098656,
+      "learning_rate": 3.501228030781034e-06,
+      "loss": 0.0028,
+      "step": 482
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 0.04921706287498077,
+      "learning_rate": 3.341176282242653e-06,
+      "loss": 0.0021,
+      "step": 483
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 0.05901589705081983,
+      "learning_rate": 3.184806707709698e-06,
+      "loss": 0.0027,
+      "step": 484
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 0.08562934355546689,
+      "learning_rate": 3.0321252640100885e-06,
+      "loss": 0.0035,
+      "step": 485
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 0.056139936545776606,
+      "learning_rate": 2.88313776747412e-06,
+      "loss": 0.0027,
+      "step": 486
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 0.06574452787357139,
+      "learning_rate": 2.7378498937128404e-06,
+      "loss": 0.0031,
+      "step": 487
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 0.06295208396607756,
+      "learning_rate": 2.5962671774018234e-06,
+      "loss": 0.0029,
+      "step": 488
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 0.06348707610420529,
+      "learning_rate": 2.458395012070369e-06,
+      "loss": 0.0027,
+      "step": 489
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 0.06438459591992919,
+      "learning_rate": 2.3242386498960266e-06,
+      "loss": 0.003,
+      "step": 490
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 0.0936033257355208,
+      "learning_rate": 2.1938032015044964e-06,
+      "loss": 0.0053,
+      "step": 491
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 0.0712704009642112,
+      "learning_rate": 2.067093635774975e-06,
+      "loss": 0.0033,
+      "step": 492
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 0.05278839840964536,
+      "learning_rate": 1.9441147796508407e-06,
+      "loss": 0.0025,
+      "step": 493
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 0.05158800004403027,
+      "learning_rate": 1.8248713179557786e-06,
+      "loss": 0.002,
+      "step": 494
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 0.06302315225352234,
+      "learning_rate": 1.7093677932153218e-06,
+      "loss": 0.002,
+      "step": 495
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 0.09014451602286425,
+      "learning_rate": 1.5976086054838025e-06,
+      "loss": 0.0031,
+      "step": 496
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 0.08249201483869177,
+      "learning_rate": 1.4895980121767627e-06,
+      "loss": 0.0029,
+      "step": 497
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 0.07887788932672342,
+      "learning_rate": 1.3853401279086854e-06,
+      "loss": 0.0028,
+      "step": 498
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 0.09271365227044996,
+      "learning_rate": 1.2848389243363512e-06,
+      "loss": 0.0026,
+      "step": 499
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 0.05191622392926365,
+      "learning_rate": 1.1880982300074838e-06,
+      "loss": 0.0027,
+      "step": 500
+    },
+    {
+      "epoch": 4.75,
+      "eval_blimp_filtered_avg": 0.7105970149253731,
+      "eval_blimp_filtered_std": 0.005015059082306442,
+      "step": 500
+    },
+    {
+      "epoch": 4.75,
+      "eval_blimp_supplement_avg": 0.8146551724137931,
+      "eval_blimp_supplement_std": 0.01739418193453382,
+      "step": 500
+    },
+    {
+      "epoch": 4.75,
+      "eval_vqa_filtered_avg": 0.52,
+      "eval_vqa_filtered_std": 0.05021167315686779,
+      "step": 500
+    },
+    {
+      "epoch": 4.75,
+      "eval_winoground_filtered_avg": 0.64,
+      "eval_winoground_filtered_std": 0.048241815132442176,
+      "step": 500
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 0.06004512680198857,
+      "learning_rate": 1.0951217302148986e-06,
+      "loss": 0.0021,
+      "step": 501
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 0.07576379765393293,
+      "learning_rate": 1.0059129668561707e-06,
+      "loss": 0.0027,
+      "step": 502
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 0.0655321501764931,
+      "learning_rate": 9.204753382986097e-07,
+      "loss": 0.0029,
+      "step": 503
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 0.06668565079155468,
+      "learning_rate": 8.388120992499083e-07,
+      "loss": 0.0024,
+      "step": 504
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 0.08295379764022878,
+      "learning_rate": 7.609263606340622e-07,
+      "loss": 0.003,
+      "step": 505
+    },
+    {
+      "epoch": 4.81,
+      "grad_norm": 0.05830372848137469,
+      "learning_rate": 6.868210894729332e-07,
+      "loss": 0.0027,
+      "step": 506
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 0.04555270319966449,
+      "learning_rate": 6.164991087731831e-07,
+      "loss": 0.0021,
+      "step": 507
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 0.057930715171302063,
+      "learning_rate": 5.499630974187375e-07,
+      "loss": 0.0024,
+      "step": 508
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 0.09648171217668358,
+      "learning_rate": 4.872155900687347e-07,
+      "loss": 0.0032,
+      "step": 509
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 0.08324119499167887,
+      "learning_rate": 4.2825897706100235e-07,
+      "loss": 0.0018,
+      "step": 510
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 0.05280884195269513,
+      "learning_rate": 3.7309550432090835e-07,
+      "loss": 0.003,
+      "step": 511
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 0.06550697689715686,
+      "learning_rate": 3.217272732759402e-07,
+      "loss": 0.0029,
+      "step": 512
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 0.07122072726515956,
+      "learning_rate": 2.741562407755138e-07,
+      "loss": 0.0026,
+      "step": 513
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 0.08524060823989948,
+      "learning_rate": 2.3038421901651064e-07,
+      "loss": 0.0032,
+      "step": 514
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 0.06597786647502633,
+      "learning_rate": 1.9041287547424403e-07,
+      "loss": 0.0026,
+      "step": 515
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 0.0640231658570345,
+      "learning_rate": 1.5424373283889904e-07,
+      "loss": 0.0025,
+      "step": 516
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 0.06369562949011548,
+      "learning_rate": 1.2187816895752324e-07,
+      "loss": 0.003,
+      "step": 517
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 0.05040741786604575,
+      "learning_rate": 9.3317416781602e-08,
+      "loss": 0.0021,
+      "step": 518
+    },
+    {
+      "epoch": 4.93,
+      "grad_norm": 0.059709231647531516,
+      "learning_rate": 6.856256432000718e-08,
+      "loss": 0.0024,
+      "step": 519
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 0.07196915675318658,
+      "learning_rate": 4.7614554597608105e-08,
+      "loss": 0.0033,
+      "step": 520
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 0.057403114857655216,
+      "learning_rate": 3.047418561933357e-08,
+      "loss": 0.002,
+      "step": 521
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 0.08027211044033893,
+      "learning_rate": 1.7142110339740668e-08,
+      "loss": 0.003,
+      "step": 522
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 0.09851204603686081,
+      "learning_rate": 7.618836638190186e-09,
+      "loss": 0.0028,
+      "step": 523
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 0.0505999817391235,
+      "learning_rate": 1.904727299473219e-09,
+      "loss": 0.0023,
+      "step": 524
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 0.059018226862226256,
+      "learning_rate": 0.0,
+      "loss": 0.0029,
+      "step": 525
+    },
+    {
+      "epoch": 4.99,
+      "step": 525,
+      "total_flos": 415734656204800.0,
+      "train_loss": 0.33918485829939266,
+      "train_runtime": 37482.4854,
+      "train_samples_per_second": 8.975,
+      "train_steps_per_second": 0.014
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 525,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "total_flos": 415734656204800.0,
+  "train_batch_size": 40,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:08b5f65153ee777fbd3d7179f8557cd50d2f8195b4001cff264a353dcb84007a
+size 7032