Upload folder using huggingface_hub

base/13_128_e3_3e-5/README.md

@@ -0,0 +1,202 @@
+---
+base_model: ibm-granite/granite-3.3-8b-base
+library_name: peft
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.15.2

base/13_128_e3_3e-5/adapter_config.json

@@ -0,0 +1,39 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "ibm-granite/granite-3.3-8b-base",
+  "bias": "none",
+  "corda_config": null,
+  "eva_config": null,
+  "exclude_modules": null,
+  "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_bias": false,
+  "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": [
+    "gate_proj",
+    "o_proj",
+    "down_proj",
+    "k_proj",
+    "v_proj",
+    "up_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

base/13_128_e3_3e-5/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a4a63257f9f22ac6f2a7a7368337860519a1fac4ee247bffd2974a3e91730fa0
+size 791751704

base/13_128_e3_3e-5/latest

@@ -0,0 +1 @@
+global_step1005

base/13_128_e3_3e-5/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8130cdf803111a9a8afcb74dc9f2b0f5e2c9dfae0b3bee2185fa3a8de0e19451
+size 16389

base/13_128_e3_3e-5/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ee0651fade3c9f0a3c59f626faf7b53a4a91f50a5230f28447b014a94b5c7ed9
+size 16389

base/13_128_e3_3e-5/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3b94bd89c963dd4367058fbdebb7ec2af8ad85b41fde89a4a5c685bd68f56e0d
+size 16389

base/13_128_e3_3e-5/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b2ddd34484fddd0b0d7c1d86e90c60a4aeee306105dab18683c69dac1d443c5
+size 16389

base/13_128_e3_3e-5/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:23ad20c5e109644e2b58d7a4919c407787e144cf2f3ceba9050be046dc9b4208
+size 16389

base/13_128_e3_3e-5/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a3755270c6ced7849bf0b6bf729708160baff4b6e22215aacea497646a269fde
+size 16389

base/13_128_e3_3e-5/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61593e44d921783be8aa8fefe09271157a2be48d8e18f9998422b97785301dc7
+size 16389

base/13_128_e3_3e-5/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eaa5ce94d3a1d59051ee1482239c8a7c0c888852461d20aa3608433ca4b28619
+size 16389

base/13_128_e3_3e-5/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73422430e8f6350819498aed01f5cbf056ec1013b62ba19e9a568cadb14758a1
+size 1401

base/13_128_e3_3e-5/special_tokens_map.json

@@ -0,0 +1,45 @@
+{
+  "additional_special_tokens": [
+    "<|endoftext|>",
+    "<fim_prefix>",
+    "<fim_middle>",
+    "<fim_suffix>",
+    "<fim_pad>",
+    "<filename>",
+    "<gh_stars>",
+    "<issue_start>",
+    "<issue_comment>",
+    "<issue_closed>",
+    "<jupyter_start>",
+    "<jupyter_text>",
+    "<jupyter_code>",
+    "<jupyter_output>",
+    "<empty_output>",
+    "<commit_before>",
+    "<commit_msg>",
+    "<commit_after>",
+    "<reponame>"
+  ],
+  "bos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "<reponame>",
+  "unk_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

base/13_128_e3_3e-5/tokenizer_config.json

@@ -0,0 +1,188 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<fim_prefix>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<fim_middle>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<fim_suffix>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<fim_pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "5": {
+      "content": "<filename>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "6": {
+      "content": "<gh_stars>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "7": {
+      "content": "<issue_start>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "8": {
+      "content": "<issue_comment>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "9": {
+      "content": "<issue_closed>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "10": {
+      "content": "<jupyter_start>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "11": {
+      "content": "<jupyter_text>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "12": {
+      "content": "<jupyter_code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "13": {
+      "content": "<jupyter_output>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "14": {
+      "content": "<empty_output>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "15": {
+      "content": "<commit_before>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "16": {
+      "content": "<commit_msg>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "17": {
+      "content": "<commit_after>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "18": {
+      "content": "<reponame>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<|endoftext|>",
+    "<fim_prefix>",
+    "<fim_middle>",
+    "<fim_suffix>",
+    "<fim_pad>",
+    "<filename>",
+    "<gh_stars>",
+    "<issue_start>",
+    "<issue_comment>",
+    "<issue_closed>",
+    "<jupyter_start>",
+    "<jupyter_text>",
+    "<jupyter_code>",
+    "<jupyter_output>",
+    "<empty_output>",
+    "<commit_before>",
+    "<commit_msg>",
+    "<commit_after>",
+    "<reponame>"
+  ],
+  "bos_token": "<|endoftext|>",
+  "clean_up_tokenization_spaces": true,
+  "eos_token": "<|endoftext|>",
+  "extra_special_tokens": {},
+  "model_max_length": 8192,
+  "pad_token": "<reponame>",
+  "padding_side": "left",
+  "tokenizer_class": "GPT2Tokenizer",
+  "unk_token": "<|endoftext|>",
+  "vocab_size": 49152
+}

base/13_128_e3_3e-5/trainer_state.json

@@ -0,0 +1,1441 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 1005,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.014925373134328358,
+      "grad_norm": 0.8188077807426453,
+      "learning_rate": 2.3529411764705885e-06,
+      "loss": 1.3284,
+      "step": 5
+    },
+    {
+      "epoch": 0.029850746268656716,
+      "grad_norm": 0.5685378909111023,
+      "learning_rate": 5.294117647058824e-06,
+      "loss": 1.3113,
+      "step": 10
+    },
+    {
+      "epoch": 0.04477611940298507,
+      "grad_norm": 0.5268212556838989,
+      "learning_rate": 8.23529411764706e-06,
+      "loss": 1.2694,
+      "step": 15
+    },
+    {
+      "epoch": 0.05970149253731343,
+      "grad_norm": 0.5032298564910889,
+      "learning_rate": 1.1176470588235295e-05,
+      "loss": 1.2829,
+      "step": 20
+    },
+    {
+      "epoch": 0.07462686567164178,
+      "grad_norm": 0.4195000231266022,
+      "learning_rate": 1.411764705882353e-05,
+      "loss": 1.1982,
+      "step": 25
+    },
+    {
+      "epoch": 0.08955223880597014,
+      "grad_norm": 0.41988900303840637,
+      "learning_rate": 1.7058823529411763e-05,
+      "loss": 1.2091,
+      "step": 30
+    },
+    {
+      "epoch": 0.1044776119402985,
+      "grad_norm": 0.40643084049224854,
+      "learning_rate": 1.9999999999999998e-05,
+      "loss": 1.2288,
+      "step": 35
+    },
+    {
+      "epoch": 0.11940298507462686,
+      "grad_norm": 0.39718306064605713,
+      "learning_rate": 2.2941176470588233e-05,
+      "loss": 1.2179,
+      "step": 40
+    },
+    {
+      "epoch": 0.13432835820895522,
+      "grad_norm": 0.42657822370529175,
+      "learning_rate": 2.5882352941176472e-05,
+      "loss": 1.1305,
+      "step": 45
+    },
+    {
+      "epoch": 0.14925373134328357,
+      "grad_norm": 0.43022823333740234,
+      "learning_rate": 2.8823529411764707e-05,
+      "loss": 1.117,
+      "step": 50
+    },
+    {
+      "epoch": 0.16417910447761194,
+      "grad_norm": 0.4196614921092987,
+      "learning_rate": 2.9999268013221688e-05,
+      "loss": 1.0708,
+      "step": 55
+    },
+    {
+      "epoch": 0.1791044776119403,
+      "grad_norm": 0.4527842402458191,
+      "learning_rate": 2.9994795019399927e-05,
+      "loss": 1.0767,
+      "step": 60
+    },
+    {
+      "epoch": 0.19402985074626866,
+      "grad_norm": 0.4908345639705658,
+      "learning_rate": 2.9986256902246587e-05,
+      "loss": 1.0589,
+      "step": 65
+    },
+    {
+      "epoch": 0.208955223880597,
+      "grad_norm": 0.5150397419929504,
+      "learning_rate": 2.9973655976464824e-05,
+      "loss": 1.0376,
+      "step": 70
+    },
+    {
+      "epoch": 0.22388059701492538,
+      "grad_norm": 0.48812800645828247,
+      "learning_rate": 2.995699565819452e-05,
+      "loss": 0.9836,
+      "step": 75
+    },
+    {
+      "epoch": 0.23880597014925373,
+      "grad_norm": 0.5567641258239746,
+      "learning_rate": 2.993628046408618e-05,
+      "loss": 0.9568,
+      "step": 80
+    },
+    {
+      "epoch": 0.2537313432835821,
+      "grad_norm": 0.5827003121376038,
+      "learning_rate": 2.991151601007646e-05,
+      "loss": 0.9636,
+      "step": 85
+    },
+    {
+      "epoch": 0.26865671641791045,
+      "grad_norm": 0.7133756279945374,
+      "learning_rate": 2.9882709009865653e-05,
+      "loss": 0.9128,
+      "step": 90
+    },
+    {
+      "epoch": 0.2835820895522388,
+      "grad_norm": 0.6302868127822876,
+      "learning_rate": 2.98498672730976e-05,
+      "loss": 0.902,
+      "step": 95
+    },
+    {
+      "epoch": 0.29850746268656714,
+      "grad_norm": 0.7415488958358765,
+      "learning_rate": 2.9812999703242502e-05,
+      "loss": 0.9178,
+      "step": 100
+    },
+    {
+      "epoch": 0.31343283582089554,
+      "grad_norm": 0.6979687809944153,
+      "learning_rate": 2.977211629518312e-05,
+      "loss": 0.8532,
+      "step": 105
+    },
+    {
+      "epoch": 0.3283582089552239,
+      "grad_norm": 0.8314119577407837,
+      "learning_rate": 2.9727228132505178e-05,
+      "loss": 0.903,
+      "step": 110
+    },
+    {
+      "epoch": 0.34328358208955223,
+      "grad_norm": 0.8514019846916199,
+      "learning_rate": 2.967834738449256e-05,
+      "loss": 0.8708,
+      "step": 115
+    },
+    {
+      "epoch": 0.3582089552238806,
+      "grad_norm": 0.865260660648346,
+      "learning_rate": 2.9625487302828198e-05,
+      "loss": 0.8277,
+      "step": 120
+    },
+    {
+      "epoch": 0.373134328358209,
+      "grad_norm": 0.7271809577941895,
+      "learning_rate": 2.956866221800151e-05,
+      "loss": 0.8007,
+      "step": 125
+    },
+    {
+      "epoch": 0.3880597014925373,
+      "grad_norm": 0.8328828811645508,
+      "learning_rate": 2.9507887535423357e-05,
+      "loss": 0.7724,
+      "step": 130
+    },
+    {
+      "epoch": 0.40298507462686567,
+      "grad_norm": 0.7392182350158691,
+      "learning_rate": 2.944317973124962e-05,
+      "loss": 0.7352,
+      "step": 135
+    },
+    {
+      "epoch": 0.417910447761194,
+      "grad_norm": 0.8831366896629333,
+      "learning_rate": 2.937455634791447e-05,
+      "loss": 0.7307,
+      "step": 140
+    },
+    {
+      "epoch": 0.43283582089552236,
+      "grad_norm": 0.891521692276001,
+      "learning_rate": 2.9302035989374565e-05,
+      "loss": 0.7173,
+      "step": 145
+    },
+    {
+      "epoch": 0.44776119402985076,
+      "grad_norm": 0.9438328742980957,
+      "learning_rate": 2.9225638316065483e-05,
+      "loss": 0.6808,
+      "step": 150
+    },
+    {
+      "epoch": 0.4626865671641791,
+      "grad_norm": 0.8953092694282532,
+      "learning_rate": 2.9145384039571743e-05,
+      "loss": 0.7321,
+      "step": 155
+    },
+    {
+      "epoch": 0.47761194029850745,
+      "grad_norm": 1.0160408020019531,
+      "learning_rate": 2.9061294917011817e-05,
+      "loss": 0.6822,
+      "step": 160
+    },
+    {
+      "epoch": 0.4925373134328358,
+      "grad_norm": 1.0142035484313965,
+      "learning_rate": 2.897339374513975e-05,
+      "loss": 0.6701,
+      "step": 165
+    },
+    {
+      "epoch": 0.5074626865671642,
+      "grad_norm": 0.9683791995048523,
+      "learning_rate": 2.888170435416491e-05,
+      "loss": 0.6709,
+      "step": 170
+    },
+    {
+      "epoch": 0.5223880597014925,
+      "grad_norm": 1.2864644527435303,
+      "learning_rate": 2.878625160129155e-05,
+      "loss": 0.6434,
+      "step": 175
+    },
+    {
+      "epoch": 0.5373134328358209,
+      "grad_norm": 1.0858073234558105,
+      "learning_rate": 2.8687061363979963e-05,
+      "loss": 0.6748,
+      "step": 180
+    },
+    {
+      "epoch": 0.5522388059701493,
+      "grad_norm": 1.2624996900558472,
+      "learning_rate": 2.858416053293105e-05,
+      "loss": 0.6556,
+      "step": 185
+    },
+    {
+      "epoch": 0.5671641791044776,
+      "grad_norm": 1.008945107460022,
+      "learning_rate": 2.84775770047962e-05,
+      "loss": 0.5958,
+      "step": 190
+    },
+    {
+      "epoch": 0.582089552238806,
+      "grad_norm": 1.0350261926651,
+      "learning_rate": 2.8367339674614402e-05,
+      "loss": 0.5855,
+      "step": 195
+    },
+    {
+      "epoch": 0.5970149253731343,
+      "grad_norm": 0.9665529727935791,
+      "learning_rate": 2.825347842797879e-05,
+      "loss": 0.5693,
+      "step": 200
+    },
+    {
+      "epoch": 0.6119402985074627,
+      "grad_norm": 1.122005820274353,
+      "learning_rate": 2.8136024132934552e-05,
+      "loss": 0.5749,
+      "step": 205
+    },
+    {
+      "epoch": 0.6268656716417911,
+      "grad_norm": 0.9779945611953735,
+      "learning_rate": 2.8015008631610545e-05,
+      "loss": 0.5651,
+      "step": 210
+    },
+    {
+      "epoch": 0.6417910447761194,
+      "grad_norm": 1.2046878337860107,
+      "learning_rate": 2.789046473158682e-05,
+      "loss": 0.5597,
+      "step": 215
+    },
+    {
+      "epoch": 0.6567164179104478,
+      "grad_norm": 0.9880019426345825,
+      "learning_rate": 2.7762426197000404e-05,
+      "loss": 0.5514,
+      "step": 220
+    },
+    {
+      "epoch": 0.6716417910447762,
+      "grad_norm": 1.0366430282592773,
+      "learning_rate": 2.763092773939177e-05,
+      "loss": 0.5302,
+      "step": 225
+    },
+    {
+      "epoch": 0.6865671641791045,
+      "grad_norm": 1.0500271320343018,
+      "learning_rate": 2.749600500829448e-05,
+      "loss": 0.5703,
+      "step": 230
+    },
+    {
+      "epoch": 0.7014925373134329,
+      "grad_norm": 1.02364182472229,
+      "learning_rate": 2.7357694581570475e-05,
+      "loss": 0.5346,
+      "step": 235
+    },
+    {
+      "epoch": 0.7164179104477612,
+      "grad_norm": 1.1580158472061157,
+      "learning_rate": 2.7216033955493756e-05,
+      "loss": 0.4864,
+      "step": 240
+    },
+    {
+      "epoch": 0.7313432835820896,
+      "grad_norm": 1.29770028591156,
+      "learning_rate": 2.7071061534585064e-05,
+      "loss": 0.4915,
+      "step": 245
+    },
+    {
+      "epoch": 0.746268656716418,
+      "grad_norm": 1.3903497457504272,
+      "learning_rate": 2.6922816621200302e-05,
+      "loss": 0.4417,
+      "step": 250
+    },
+    {
+      "epoch": 0.7611940298507462,
+      "grad_norm": 1.1154823303222656,
+      "learning_rate": 2.6771339404875602e-05,
+      "loss": 0.489,
+      "step": 255
+    },
+    {
+      "epoch": 0.7761194029850746,
+      "grad_norm": 1.19193434715271,
+      "learning_rate": 2.6616670951431842e-05,
+      "loss": 0.4621,
+      "step": 260
+    },
+    {
+      "epoch": 0.7910447761194029,
+      "grad_norm": 1.3155850172042847,
+      "learning_rate": 2.645885319184159e-05,
+      "loss": 0.4669,
+      "step": 265
+    },
+    {
+      "epoch": 0.8059701492537313,
+      "grad_norm": 1.1470247507095337,
+      "learning_rate": 2.6297928910861546e-05,
+      "loss": 0.4577,
+      "step": 270
+    },
+    {
+      "epoch": 0.8208955223880597,
+      "grad_norm": 1.2841483354568481,
+      "learning_rate": 2.6133941735433496e-05,
+      "loss": 0.4384,
+      "step": 275
+    },
+    {
+      "epoch": 0.835820895522388,
+      "grad_norm": 1.3723411560058594,
+      "learning_rate": 2.596693612285691e-05,
+      "loss": 0.4036,
+      "step": 280
+    },
+    {
+      "epoch": 0.8507462686567164,
+      "grad_norm": 1.1929142475128174,
+      "learning_rate": 2.5796957348736522e-05,
+      "loss": 0.4302,
+      "step": 285
+    },
+    {
+      "epoch": 0.8656716417910447,
+      "grad_norm": 1.1312122344970703,
+      "learning_rate": 2.5624051494707967e-05,
+      "loss": 0.4153,
+      "step": 290
+    },
+    {
+      "epoch": 0.8805970149253731,
+      "grad_norm": 1.616319179534912,
+      "learning_rate": 2.5448265435944957e-05,
+      "loss": 0.3502,
+      "step": 295
+    },
+    {
+      "epoch": 0.8955223880597015,
+      "grad_norm": 1.2079335451126099,
+      "learning_rate": 2.5269646828451323e-05,
+      "loss": 0.4232,
+      "step": 300
+    },
+    {
+      "epoch": 0.9104477611940298,
+      "grad_norm": 1.0986040830612183,
+      "learning_rate": 2.5088244096141355e-05,
+      "loss": 0.3958,
+      "step": 305
+    },
+    {
+      "epoch": 0.9253731343283582,
+      "grad_norm": 1.223262071609497,
+      "learning_rate": 2.490410641771196e-05,
+      "loss": 0.3896,
+      "step": 310
+    },
+    {
+      "epoch": 0.9402985074626866,
+      "grad_norm": 1.1202898025512695,
+      "learning_rate": 2.4717283713310224e-05,
+      "loss": 0.3723,
+      "step": 315
+    },
+    {
+      "epoch": 0.9552238805970149,
+      "grad_norm": 1.308760404586792,
+      "learning_rate": 2.4527826630999922e-05,
+      "loss": 0.3912,
+      "step": 320
+    },
+    {
+      "epoch": 0.9701492537313433,
+      "grad_norm": 1.2027877569198608,
+      "learning_rate": 2.4335786533030736e-05,
+      "loss": 0.3645,
+      "step": 325
+    },
+    {
+      "epoch": 0.9850746268656716,
+      "grad_norm": 1.2411190271377563,
+      "learning_rate": 2.414121548191381e-05,
+      "loss": 0.3909,
+      "step": 330
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.1736737489700317,
+      "learning_rate": 2.39441662263075e-05,
+      "loss": 0.3601,
+      "step": 335
+    },
+    {
+      "epoch": 1.0149253731343284,
+      "grad_norm": 1.2270152568817139,
+      "learning_rate": 2.374469218671708e-05,
+      "loss": 0.3083,
+      "step": 340
+    },
+    {
+      "epoch": 1.0298507462686568,
+      "grad_norm": 1.158692717552185,
+      "learning_rate": 2.3542847441012325e-05,
+      "loss": 0.2991,
+      "step": 345
+    },
+    {
+      "epoch": 1.044776119402985,
+      "grad_norm": 1.4135785102844238,
+      "learning_rate": 2.333868670976688e-05,
+      "loss": 0.3071,
+      "step": 350
+    },
+    {
+      "epoch": 1.0597014925373134,
+      "grad_norm": 1.3567718267440796,
+      "learning_rate": 2.3132265341423382e-05,
+      "loss": 0.2703,
+      "step": 355
+    },
+    {
+      "epoch": 1.0746268656716418,
+      "grad_norm": 1.2302590608596802,
+      "learning_rate": 2.292363929728836e-05,
+      "loss": 0.3014,
+      "step": 360
+    },
+    {
+      "epoch": 1.0895522388059702,
+      "grad_norm": 1.150296688079834,
+      "learning_rate": 2.2712865136361037e-05,
+      "loss": 0.3241,
+      "step": 365
+    },
+    {
+      "epoch": 1.1044776119402986,
+      "grad_norm": 1.1916894912719727,
+      "learning_rate": 2.25e-05,
+      "loss": 0.2576,
+      "step": 370
+    },
+    {
+      "epoch": 1.1194029850746268,
+      "grad_norm": 1.1171743869781494,
+      "learning_rate": 2.2285101596432084e-05,
+      "loss": 0.2947,
+      "step": 375
+    },
+    {
+      "epoch": 1.1343283582089552,
+      "grad_norm": 1.3033281564712524,
+      "learning_rate": 2.2068228185107524e-05,
+      "loss": 0.2711,
+      "step": 380
+    },
+    {
+      "epoch": 1.1492537313432836,
+      "grad_norm": 1.2771257162094116,
+      "learning_rate": 2.1849438560905697e-05,
+      "loss": 0.2699,
+      "step": 385
+    },
+    {
+      "epoch": 1.164179104477612,
+      "grad_norm": 1.5561299324035645,
+      "learning_rate": 2.162879203819568e-05,
+      "loss": 0.2648,
+      "step": 390
+    },
+    {
+      "epoch": 1.1791044776119404,
+      "grad_norm": 1.2045551538467407,
+      "learning_rate": 2.1406348434755994e-05,
+      "loss": 0.2774,
+      "step": 395
+    },
+    {
+      "epoch": 1.1940298507462686,
+      "grad_norm": 1.4329160451889038,
+      "learning_rate": 2.118216805555788e-05,
+      "loss": 0.2682,
+      "step": 400
+    },
+    {
+      "epoch": 1.208955223880597,
+      "grad_norm": 1.13307785987854,
+      "learning_rate": 2.0956311676416464e-05,
+      "loss": 0.2836,
+      "step": 405
+    },
+    {
+      "epoch": 1.2238805970149254,
+      "grad_norm": 1.2032175064086914,
+      "learning_rate": 2.0728840527514294e-05,
+      "loss": 0.2344,
+      "step": 410
+    },
+    {
+      "epoch": 1.2388059701492538,
+      "grad_norm": 1.0836251974105835,
+      "learning_rate": 2.0499816276801724e-05,
+      "loss": 0.2483,
+      "step": 415
+    },
+    {
+      "epoch": 1.2537313432835822,
+      "grad_norm": 1.1372569799423218,
+      "learning_rate": 2.0269301013278555e-05,
+      "loss": 0.2417,
+      "step": 420
+    },
+    {
+      "epoch": 1.2686567164179103,
+      "grad_norm": 1.2705116271972656,
+      "learning_rate": 2.0037357230161587e-05,
+      "loss": 0.2388,
+      "step": 425
+    },
+    {
+      "epoch": 1.2835820895522387,
+      "grad_norm": 1.2042105197906494,
+      "learning_rate": 1.9804047807942564e-05,
+      "loss": 0.2464,
+      "step": 430
+    },
+    {
+      "epoch": 1.2985074626865671,
+      "grad_norm": 1.094473958015442,
+      "learning_rate": 1.956943599734112e-05,
+      "loss": 0.2159,
+      "step": 435
+    },
+    {
+      "epoch": 1.3134328358208955,
+      "grad_norm": 1.3166625499725342,
+      "learning_rate": 1.9333585402157365e-05,
+      "loss": 0.2572,
+      "step": 440
+    },
+    {
+      "epoch": 1.328358208955224,
+      "grad_norm": 1.3254424333572388,
+      "learning_rate": 1.9096559962028746e-05,
+      "loss": 0.2036,
+      "step": 445
+    },
+    {
+      "epoch": 1.3432835820895521,
+      "grad_norm": 1.0283764600753784,
+      "learning_rate": 1.88584239350959e-05,
+      "loss": 0.2087,
+      "step": 450
+    },
+    {
+      "epoch": 1.3582089552238805,
+      "grad_norm": 1.400997281074524,
+      "learning_rate": 1.861924188058205e-05,
+      "loss": 0.2561,
+      "step": 455
+    },
+    {
+      "epoch": 1.373134328358209,
+      "grad_norm": 1.210942268371582,
+      "learning_rate": 1.8379078641290923e-05,
+      "loss": 0.2139,
+      "step": 460
+    },
+    {
+      "epoch": 1.3880597014925373,
+      "grad_norm": 1.2460658550262451,
+      "learning_rate": 1.8137999326027696e-05,
+      "loss": 0.2392,
+      "step": 465
+    },
+    {
+      "epoch": 1.4029850746268657,
+      "grad_norm": 1.171144962310791,
+      "learning_rate": 1.7896069291947827e-05,
+      "loss": 0.2056,
+      "step": 470
+    },
+    {
+      "epoch": 1.417910447761194,
+      "grad_norm": 1.1853833198547363,
+      "learning_rate": 1.7653354126838593e-05,
+      "loss": 0.1908,
+      "step": 475
+    },
+    {
+      "epoch": 1.4328358208955223,
+      "grad_norm": 1.2758232355117798,
+      "learning_rate": 1.7409919631338124e-05,
+      "loss": 0.2527,
+      "step": 480
+    },
+    {
+      "epoch": 1.4477611940298507,
+      "grad_norm": 1.1914968490600586,
+      "learning_rate": 1.7165831801096635e-05,
+      "loss": 0.1981,
+      "step": 485
+    },
+    {
+      "epoch": 1.462686567164179,
+      "grad_norm": 1.2703306674957275,
+      "learning_rate": 1.6921156808884904e-05,
+      "loss": 0.2373,
+      "step": 490
+    },
+    {
+      "epoch": 1.4776119402985075,
+      "grad_norm": 1.1430697441101074,
+      "learning_rate": 1.6675960986654675e-05,
+      "loss": 0.2205,
+      "step": 495
+    },
+    {
+      "epoch": 1.4925373134328357,
+      "grad_norm": 1.155730962753296,
+      "learning_rate": 1.6430310807555884e-05,
+      "loss": 0.1748,
+      "step": 500
+    },
+    {
+      "epoch": 1.5074626865671643,
+      "grad_norm": 1.3574665784835815,
+      "learning_rate": 1.618427286791568e-05,
+      "loss": 0.1823,
+      "step": 505
+    },
+    {
+      "epoch": 1.5223880597014925,
+      "grad_norm": 1.1159979104995728,
+      "learning_rate": 1.593791386918396e-05,
+      "loss": 0.1822,
+      "step": 510
+    },
+    {
+      "epoch": 1.537313432835821,
+      "grad_norm": 1.2768901586532593,
+      "learning_rate": 1.5691300599850495e-05,
+      "loss": 0.1716,
+      "step": 515
+    },
+    {
+      "epoch": 1.5522388059701493,
+      "grad_norm": 1.2257966995239258,
+      "learning_rate": 1.5444499917338398e-05,
+      "loss": 0.1992,
+      "step": 520
+    },
+    {
+      "epoch": 1.5671641791044775,
+      "grad_norm": 1.0807833671569824,
+      "learning_rate": 1.5197578729878915e-05,
+      "loss": 0.1803,
+      "step": 525
+    },
+    {
+      "epoch": 1.582089552238806,
+      "grad_norm": 1.1698600053787231,
+      "learning_rate": 1.4950603978372467e-05,
+      "loss": 0.1868,
+      "step": 530
+    },
+    {
+      "epoch": 1.5970149253731343,
+      "grad_norm": 1.0760620832443237,
+      "learning_rate": 1.4703642618240806e-05,
+      "loss": 0.175,
+      "step": 535
+    },
+    {
+      "epoch": 1.6119402985074627,
+      "grad_norm": 1.0615389347076416,
+      "learning_rate": 1.4456761601275254e-05,
+      "loss": 0.1842,
+      "step": 540
+    },
+    {
+      "epoch": 1.626865671641791,
+      "grad_norm": 1.0602505207061768,
+      "learning_rate": 1.4210027857485932e-05,
+      "loss": 0.1482,
+      "step": 545
+    },
+    {
+      "epoch": 1.6417910447761193,
+      "grad_norm": 1.105310082435608,
+      "learning_rate": 1.3963508276956832e-05,
+      "loss": 0.1869,
+      "step": 550
+    },
+    {
+      "epoch": 1.6567164179104479,
+      "grad_norm": 1.1317061185836792,
+      "learning_rate": 1.371726969171182e-05,
+      "loss": 0.1519,
+      "step": 555
+    },
+    {
+      "epoch": 1.671641791044776,
+      "grad_norm": 1.1249550580978394,
+      "learning_rate": 1.34713788575963e-05,
+      "loss": 0.1511,
+      "step": 560
+    },
+    {
+      "epoch": 1.6865671641791045,
+      "grad_norm": 1.2642542123794556,
+      "learning_rate": 1.3225902436179515e-05,
+      "loss": 0.1724,
+      "step": 565
+    },
+    {
+      "epoch": 1.7014925373134329,
+      "grad_norm": 1.315899133682251,
+      "learning_rate": 1.2980906976682508e-05,
+      "loss": 0.158,
+      "step": 570
+    },
+    {
+      "epoch": 1.716417910447761,
+      "grad_norm": 1.4055341482162476,
+      "learning_rate": 1.2736458897936432e-05,
+      "loss": 0.1522,
+      "step": 575
+    },
+    {
+      "epoch": 1.7313432835820897,
+      "grad_norm": 1.2732903957366943,
+      "learning_rate": 1.2492624470376253e-05,
+      "loss": 0.1546,
+      "step": 580
+    },
+    {
+      "epoch": 1.7462686567164178,
+      "grad_norm": 1.145630121231079,
+      "learning_rate": 1.22494697980747e-05,
+      "loss": 0.1701,
+      "step": 585
+    },
+    {
+      "epoch": 1.7611940298507462,
+      "grad_norm": 1.1732102632522583,
+      "learning_rate": 1.20070608008213e-05,
+      "loss": 0.1646,
+      "step": 590
+    },
+    {
+      "epoch": 1.7761194029850746,
+      "grad_norm": 1.1512342691421509,
+      "learning_rate": 1.1765463196251349e-05,
+      "loss": 0.164,
+      "step": 595
+    },
+    {
+      "epoch": 1.7910447761194028,
+      "grad_norm": 1.1634297370910645,
+      "learning_rate": 1.1524742482029728e-05,
+      "loss": 0.1338,
+      "step": 600
+    },
+    {
+      "epoch": 1.8059701492537314,
+      "grad_norm": 1.004382610321045,
+      "learning_rate": 1.1284963918094346e-05,
+      "loss": 0.1523,
+      "step": 605
+    },
+    {
+      "epoch": 1.8208955223880596,
+      "grad_norm": 1.097747564315796,
+      "learning_rate": 1.104619250896399e-05,
+      "loss": 0.1506,
+      "step": 610
+    },
+    {
+      "epoch": 1.835820895522388,
+      "grad_norm": 1.0649802684783936,
+      "learning_rate": 1.0808492986115476e-05,
+      "loss": 0.1329,
+      "step": 615
+    },
+    {
+      "epoch": 1.8507462686567164,
+      "grad_norm": 1.1538245677947998,
+      "learning_rate": 1.0571929790434792e-05,
+      "loss": 0.1384,
+      "step": 620
+    },
+    {
+      "epoch": 1.8656716417910446,
+      "grad_norm": 1.0772777795791626,
+      "learning_rate": 1.0336567054747033e-05,
+      "loss": 0.1569,
+      "step": 625
+    },
+    {
+      "epoch": 1.8805970149253732,
+      "grad_norm": 1.218000888824463,
+      "learning_rate": 1.0102468586429808e-05,
+      "loss": 0.1386,
+      "step": 630
+    },
+    {
+      "epoch": 1.8955223880597014,
+      "grad_norm": 0.9585780501365662,
+      "learning_rate": 9.86969785011497e-06,
+      "loss": 0.1365,
+      "step": 635
+    },
+    {
+      "epoch": 1.9104477611940298,
+      "grad_norm": 1.0707758665084839,
+      "learning_rate": 9.638317950483167e-06,
+      "loss": 0.123,
+      "step": 640
+    },
+    {
+      "epoch": 1.9253731343283582,
+      "grad_norm": 1.0073219537734985,
+      "learning_rate": 9.408391615156023e-06,
+      "loss": 0.1154,
+      "step": 645
+    },
+    {
+      "epoch": 1.9402985074626866,
+      "grad_norm": 1.2402931451797485,
+      "learning_rate": 9.179981177690566e-06,
+      "loss": 0.1187,
+      "step": 650
+    },
+    {
+      "epoch": 1.955223880597015,
+      "grad_norm": 1.1681808233261108,
+      "learning_rate": 8.953148560680419e-06,
+      "loss": 0.1368,
+      "step": 655
+    },
+    {
+      "epoch": 1.9701492537313432,
+      "grad_norm": 1.307678461074829,
+      "learning_rate": 8.727955258968462e-06,
+      "loss": 0.1258,
+      "step": 660
+    },
+    {
+      "epoch": 1.9850746268656716,
+      "grad_norm": 1.1485263109207153,
+      "learning_rate": 8.504462322975442e-06,
+      "loss": 0.1211,
+      "step": 665
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.208245038986206,
+      "learning_rate": 8.282730342149059e-06,
+      "loss": 0.1116,
+      "step": 670
+    },
+    {
+      "epoch": 2.014925373134328,
+      "grad_norm": 1.0575000047683716,
+      "learning_rate": 8.062819428538009e-06,
+      "loss": 0.0959,
+      "step": 675
+    },
+    {
+      "epoch": 2.029850746268657,
+      "grad_norm": 0.9653350114822388,
+      "learning_rate": 7.844789200495517e-06,
+      "loss": 0.0867,
+      "step": 680
+    },
+    {
+      "epoch": 2.044776119402985,
+      "grad_norm": 0.9760949015617371,
+      "learning_rate": 7.628698766516625e-06,
+      "loss": 0.0881,
+      "step": 685
+    },
+    {
+      "epoch": 2.0597014925373136,
+      "grad_norm": 0.8332158923149109,
+      "learning_rate": 7.414606709213735e-06,
+      "loss": 0.0866,
+      "step": 690
+    },
+    {
+      "epoch": 2.074626865671642,
+      "grad_norm": 1.0246304273605347,
+      "learning_rate": 7.202571069434772e-06,
+      "loss": 0.0884,
+      "step": 695
+    },
+    {
+      "epoch": 2.08955223880597,
+      "grad_norm": 0.9097149968147278,
+      "learning_rate": 6.992649330528146e-06,
+      "loss": 0.0939,
+      "step": 700
+    },
+    {
+      "epoch": 2.1044776119402986,
+      "grad_norm": 0.9704083800315857,
+      "learning_rate": 6.78489840275887e-06,
+      "loss": 0.0853,
+      "step": 705
+    },
+    {
+      "epoch": 2.1194029850746268,
+      "grad_norm": 1.02879798412323,
+      "learning_rate": 6.579374607880116e-06,
+      "loss": 0.0809,
+      "step": 710
+    },
+    {
+      "epoch": 2.1343283582089554,
+      "grad_norm": 0.945453405380249,
+      "learning_rate": 6.376133663864196e-06,
+      "loss": 0.0779,
+      "step": 715
+    },
+    {
+      "epoch": 2.1492537313432836,
+      "grad_norm": 1.0824517011642456,
+      "learning_rate": 6.175230669797306e-06,
+      "loss": 0.0929,
+      "step": 720
+    },
+    {
+      "epoch": 2.1641791044776117,
+      "grad_norm": 1.3560867309570312,
+      "learning_rate": 5.976720090942066e-06,
+      "loss": 0.0902,
+      "step": 725
+    },
+    {
+      "epoch": 2.1791044776119404,
+      "grad_norm": 0.9563027024269104,
+      "learning_rate": 5.780655743971844e-06,
+      "loss": 0.074,
+      "step": 730
+    },
+    {
+      "epoch": 2.1940298507462686,
+      "grad_norm": 0.9136431217193604,
+      "learning_rate": 5.587090782380912e-06,
+      "loss": 0.0802,
+      "step": 735
+    },
+    {
+      "epoch": 2.208955223880597,
+      "grad_norm": 0.8077270984649658,
+      "learning_rate": 5.3960776820744415e-06,
+      "loss": 0.09,
+      "step": 740
+    },
+    {
+      "epoch": 2.2238805970149254,
+      "grad_norm": 0.9026997089385986,
+      "learning_rate": 5.207668227142178e-06,
+      "loss": 0.0755,
+      "step": 745
+    },
+    {
+      "epoch": 2.2388059701492535,
+      "grad_norm": 0.8953799605369568,
+      "learning_rate": 5.021913495819593e-06,
+      "loss": 0.0824,
+      "step": 750
+    },
+    {
+      "epoch": 2.253731343283582,
+      "grad_norm": 0.7142133116722107,
+      "learning_rate": 4.838863846640524e-06,
+      "loss": 0.0664,
+      "step": 755
+    },
+    {
+      "epoch": 2.2686567164179103,
+      "grad_norm": 0.7506063580513,
+      "learning_rate": 4.6585689047848264e-06,
+      "loss": 0.0815,
+      "step": 760
+    },
+    {
+      "epoch": 2.283582089552239,
+      "grad_norm": 0.9790431261062622,
+      "learning_rate": 4.481077548624871e-06,
+      "loss": 0.0774,
+      "step": 765
+    },
+    {
+      "epoch": 2.298507462686567,
+      "grad_norm": 0.9579931497573853,
+      "learning_rate": 4.306437896474523e-06,
+      "loss": 0.0886,
+      "step": 770
+    },
+    {
+      "epoch": 2.3134328358208958,
+      "grad_norm": 0.7290546894073486,
+      "learning_rate": 4.134697293544158e-06,
+      "loss": 0.0818,
+      "step": 775
+    },
+    {
+      "epoch": 2.328358208955224,
+      "grad_norm": 0.8074793219566345,
+      "learning_rate": 3.965902299105245e-06,
+      "loss": 0.0773,
+      "step": 780
+    },
+    {
+      "epoch": 2.343283582089552,
+      "grad_norm": 0.9090767502784729,
+      "learning_rate": 3.8000986738680245e-06,
+      "loss": 0.079,
+      "step": 785
+    },
+    {
+      "epoch": 2.3582089552238807,
+      "grad_norm": 0.8598201870918274,
+      "learning_rate": 3.637331367575698e-06,
+      "loss": 0.0762,
+      "step": 790
+    },
+    {
+      "epoch": 2.373134328358209,
+      "grad_norm": 0.8872838020324707,
+      "learning_rate": 3.4776445068184365e-06,
+      "loss": 0.0771,
+      "step": 795
+    },
+    {
+      "epoch": 2.388059701492537,
+      "grad_norm": 0.766930341720581,
+      "learning_rate": 3.32108138307054e-06,
+      "loss": 0.0652,
+      "step": 800
+    },
+    {
+      "epoch": 2.4029850746268657,
+      "grad_norm": 0.6352797746658325,
+      "learning_rate": 3.1676844409540607e-06,
+      "loss": 0.0695,
+      "step": 805
+    },
+    {
+      "epoch": 2.417910447761194,
+      "grad_norm": 0.7674069404602051,
+      "learning_rate": 3.017495266731942e-06,
+      "loss": 0.0672,
+      "step": 810
+    },
+    {
+      "epoch": 2.4328358208955225,
+      "grad_norm": 0.9534958004951477,
+      "learning_rate": 2.8705545770338758e-06,
+      "loss": 0.0763,
+      "step": 815
+    },
+    {
+      "epoch": 2.4477611940298507,
+      "grad_norm": 0.7554170489311218,
+      "learning_rate": 2.7269022078179638e-06,
+      "loss": 0.0715,
+      "step": 820
+    },
+    {
+      "epoch": 2.4626865671641793,
+      "grad_norm": 0.7556818723678589,
+      "learning_rate": 2.5865771035710777e-06,
+      "loss": 0.0687,
+      "step": 825
+    },
+    {
+      "epoch": 2.4776119402985075,
+      "grad_norm": 0.8016027212142944,
+      "learning_rate": 2.449617306750913e-06,
+      "loss": 0.0733,
+      "step": 830
+    },
+    {
+      "epoch": 2.4925373134328357,
+      "grad_norm": 0.9080985188484192,
+      "learning_rate": 2.3160599474726073e-06,
+      "loss": 0.0629,
+      "step": 835
+    },
+    {
+      "epoch": 2.5074626865671643,
+      "grad_norm": 0.773644208908081,
+      "learning_rate": 2.1859412334426853e-06,
+      "loss": 0.0601,
+      "step": 840
+    },
+    {
+      "epoch": 2.5223880597014925,
+      "grad_norm": 1.204464077949524,
+      "learning_rate": 2.0592964401430377e-06,
+      "loss": 0.0815,
+      "step": 845
+    },
+    {
+      "epoch": 2.5373134328358207,
+      "grad_norm": 0.8199998736381531,
+      "learning_rate": 1.936159901267682e-06,
+      "loss": 0.063,
+      "step": 850
+    },
+    {
+      "epoch": 2.5522388059701493,
+      "grad_norm": 0.7511867880821228,
+      "learning_rate": 1.8165649994148203e-06,
+      "loss": 0.069,
+      "step": 855
+    },
+    {
+      "epoch": 2.5671641791044775,
+      "grad_norm": 0.8140180706977844,
+      "learning_rate": 1.7005441570367164e-06,
+      "loss": 0.0715,
+      "step": 860
+    },
+    {
+      "epoch": 2.582089552238806,
+      "grad_norm": 0.7719194293022156,
+      "learning_rate": 1.5881288276499211e-06,
+      "loss": 0.0619,
+      "step": 865
+    },
+    {
+      "epoch": 2.5970149253731343,
+      "grad_norm": 0.6067129373550415,
+      "learning_rate": 1.4793494873081504e-06,
+      "loss": 0.0634,
+      "step": 870
+    },
+    {
+      "epoch": 2.611940298507463,
+      "grad_norm": 0.7557328939437866,
+      "learning_rate": 1.374235626340128e-06,
+      "loss": 0.0799,
+      "step": 875
+    },
+    {
+      "epoch": 2.626865671641791,
+      "grad_norm": 0.6791887283325195,
+      "learning_rate": 1.2728157413547232e-06,
+      "loss": 0.0658,
+      "step": 880
+    },
+    {
+      "epoch": 2.6417910447761193,
+      "grad_norm": 0.6470426917076111,
+      "learning_rate": 1.1751173275154403e-06,
+      "loss": 0.0624,
+      "step": 885
+    },
+    {
+      "epoch": 2.656716417910448,
+      "grad_norm": 0.7489967942237854,
+      "learning_rate": 1.0811668710864098e-06,
+      "loss": 0.0591,
+      "step": 890
+    },
+    {
+      "epoch": 2.671641791044776,
+      "grad_norm": 0.7958751916885376,
+      "learning_rate": 9.909898422519198e-07,
+      "loss": 0.0778,
+      "step": 895
+    },
+    {
+      "epoch": 2.6865671641791042,
+      "grad_norm": 0.6287710666656494,
+      "learning_rate": 9.046106882113753e-07,
+      "loss": 0.0525,
+      "step": 900
+    },
+    {
+      "epoch": 2.701492537313433,
+      "grad_norm": 0.8145782351493835,
+      "learning_rate": 8.220528265516125e-07,
+      "loss": 0.0643,
+      "step": 905
+    },
+    {
+      "epoch": 2.716417910447761,
+      "grad_norm": 0.5965122580528259,
+      "learning_rate": 7.433386388983343e-07,
+      "loss": 0.0538,
+      "step": 910
+    },
+    {
+      "epoch": 2.7313432835820897,
+      "grad_norm": 0.5697555541992188,
+      "learning_rate": 6.684894648484069e-07,
+      "loss": 0.0594,
+      "step": 915
+    },
+    {
+      "epoch": 2.746268656716418,
+      "grad_norm": 0.6798402667045593,
+      "learning_rate": 5.975255961846343e-07,
+      "loss": 0.0727,
+      "step": 920
+    },
+    {
+      "epoch": 2.7611940298507465,
+      "grad_norm": 0.8096169233322144,
+      "learning_rate": 5.304662713746205e-07,
+      "loss": 0.0671,
+      "step": 925
+    },
+    {
+      "epoch": 2.7761194029850746,
+      "grad_norm": 0.6525831818580627,
+      "learning_rate": 4.6732967035517326e-07,
+      "loss": 0.067,
+      "step": 930
+    },
+    {
+      "epoch": 2.791044776119403,
+      "grad_norm": 0.6912659406661987,
+      "learning_rate": 4.081329096036829e-07,
+      "loss": 0.0625,
+      "step": 935
+    },
+    {
+      "epoch": 2.8059701492537314,
+      "grad_norm": 0.8461073637008667,
+      "learning_rate": 3.528920374977979e-07,
+      "loss": 0.0656,
+      "step": 940
+    },
+    {
+      "epoch": 2.8208955223880596,
+      "grad_norm": 0.6365325450897217,
+      "learning_rate": 3.0162202996468156e-07,
+      "loss": 0.061,
+      "step": 945
+    },
+    {
+      "epoch": 2.835820895522388,
+      "grad_norm": 0.7500727772712708,
+      "learning_rate": 2.5433678642100664e-07,
+      "loss": 0.0696,
+      "step": 950
+    },
+    {
+      "epoch": 2.8507462686567164,
+      "grad_norm": 0.7384939193725586,
+      "learning_rate": 2.110491260047792e-07,
+      "loss": 0.0686,
+      "step": 955
+    },
+    {
+      "epoch": 2.8656716417910446,
+      "grad_norm": 0.7971112132072449,
+      "learning_rate": 1.7177078410005041e-07,
+      "loss": 0.0685,
+      "step": 960
+    },
+    {
+      "epoch": 2.8805970149253732,
+      "grad_norm": 0.7816891670227051,
+      "learning_rate": 1.3651240915542652e-07,
+      "loss": 0.0733,
+      "step": 965
+    },
+    {
+      "epoch": 2.8955223880597014,
+      "grad_norm": 0.6905750036239624,
+      "learning_rate": 1.0528355979724624e-07,
+      "loss": 0.0658,
+      "step": 970
+    },
+    {
+      "epoch": 2.91044776119403,
+      "grad_norm": 0.7273623943328857,
+      "learning_rate": 7.809270223821552e-08,
+      "loss": 0.0525,
+      "step": 975
+    },
+    {
+      "epoch": 2.925373134328358,
+      "grad_norm": 0.7233467698097229,
+      "learning_rate": 5.4947207982204985e-08,
+      "loss": 0.0618,
+      "step": 980
+    },
+    {
+      "epoch": 2.9402985074626864,
+      "grad_norm": 0.6842172741889954,
+      "learning_rate": 3.585335182580529e-08,
+      "loss": 0.0791,
+      "step": 985
+    },
+    {
+      "epoch": 2.955223880597015,
+      "grad_norm": 0.5739861130714417,
+      "learning_rate": 2.0816310157227846e-08,
+      "loss": 0.0645,
+      "step": 990
+    },
+    {
+      "epoch": 2.970149253731343,
+      "grad_norm": 0.6204884052276611,
+      "learning_rate": 9.840159552969019e-09,
+      "loss": 0.058,
+      "step": 995
+    },
+    {
+      "epoch": 2.9850746268656714,
+      "grad_norm": 0.6620525121688843,
+      "learning_rate": 2.9278756726375257e-09,
+      "loss": 0.0693,
+      "step": 1000
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.7125095725059509,
+      "learning_rate": 8.133245225305785e-11,
+      "loss": 0.0753,
+      "step": 1005
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1005,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 2000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.7241489809940152e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

base/13_128_e3_3e-5/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:33887bbf57c52513335e040b37efe3bba936d6268f2f79f6da442c658a5e6570
+size 8273

base/13_128_e3_3e-5/zero_to_fp32.py

@@ -0,0 +1,604 @@
+#!/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, exclude_frozen_parameters):
+    """
+    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,
+                                                          exclude_frozen_parameters)
+    elif zero_stage == 3:
+        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+
+
+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 _has_callable(obj, fn):
+    attr = getattr(obj, fn, None)
+    return callable(attr)
+
+
+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() if _has_callable(shape, 'numel') else math.prod(shape)
+            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,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _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,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _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, exclude_frozen_parameters=False):
+    """
+    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``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+
+    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, exclude_frozen_parameters)
+
+
+def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False):
+    """
+    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``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+    """
+
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters)
+    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("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
+    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,
+                                               exclude_frozen_parameters=args.exclude_frozen_parameters)