Upload folder using huggingface_hub

base/22_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/22_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": [
+    "v_proj",
+    "k_proj",
+    "gate_proj",
+    "o_proj",
+    "up_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

base/22_128_e3_3e-5/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:491094a2a293125b8e6ca4eb43ebc14702636c816338266a3747ee35aa02dcf2
+size 791751704

base/22_128_e3_3e-5/latest

@@ -0,0 +1 @@
+global_step1116

base/22_128_e3_3e-5/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:04fc6ea3f9861b95adf712c2a8b56298cf21919230c9851c73eefe2d7d0ca534
+size 16389

base/22_128_e3_3e-5/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fe7ea071d1f9c70dcb020cacbb126020047a18008aaf0c1895aac0f3cecbcc60
+size 16389

base/22_128_e3_3e-5/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5e49d31365e821a228a37cc67d80df9479feb05beeba506cea11db2576a12c9c
+size 16389

base/22_128_e3_3e-5/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2bebfeac0d29aeb100302f17a2e80bc3faa0c5f101fcd0513f16903db302b08f
+size 16389

base/22_128_e3_3e-5/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2b641e44bf1d40c97ed598b0018ab4cb6247b95544590c146b0bec43a9eac39a
+size 16389

base/22_128_e3_3e-5/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3f10347fcc399d9e33971365662e828b6e7329252a554e3d0aaba96d6b76638b
+size 16389

base/22_128_e3_3e-5/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5ea450af986263f9d2889ff7c8237e0ddf0a86e6b2c80175ca2830ac64cd4cfb
+size 16389

base/22_128_e3_3e-5/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ab0208f64d58e49e492e5036fc9c310258a596c0b7ced626079a5ec13e0ce70a
+size 16389

base/22_128_e3_3e-5/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3ca07ec63e57b77e792aaad097780d19ad143630ed28bf555d041baaa0081ba4
+size 1401

base/22_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/22_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/22_128_e3_3e-5/trainer_state.json

@@ -0,0 +1,1595 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 1116,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.013440860215053764,
+      "grad_norm": 1.00059175491333,
+      "learning_rate": 2.1428571428571427e-06,
+      "loss": 1.2699,
+      "step": 5
+    },
+    {
+      "epoch": 0.026881720430107527,
+      "grad_norm": 0.6339553594589233,
+      "learning_rate": 4.821428571428572e-06,
+      "loss": 1.297,
+      "step": 10
+    },
+    {
+      "epoch": 0.04032258064516129,
+      "grad_norm": 0.46756768226623535,
+      "learning_rate": 7.5e-06,
+      "loss": 1.2648,
+      "step": 15
+    },
+    {
+      "epoch": 0.053763440860215055,
+      "grad_norm": 0.5269528031349182,
+      "learning_rate": 1.0178571428571429e-05,
+      "loss": 1.1965,
+      "step": 20
+    },
+    {
+      "epoch": 0.06720430107526881,
+      "grad_norm": 0.46185433864593506,
+      "learning_rate": 1.2857142857142857e-05,
+      "loss": 1.2958,
+      "step": 25
+    },
+    {
+      "epoch": 0.08064516129032258,
+      "grad_norm": 0.43897947669029236,
+      "learning_rate": 1.553571428571429e-05,
+      "loss": 1.1974,
+      "step": 30
+    },
+    {
+      "epoch": 0.09408602150537634,
+      "grad_norm": 0.4118952751159668,
+      "learning_rate": 1.8214285714285712e-05,
+      "loss": 1.1729,
+      "step": 35
+    },
+    {
+      "epoch": 0.10752688172043011,
+      "grad_norm": 0.41677960753440857,
+      "learning_rate": 2.089285714285714e-05,
+      "loss": 1.1983,
+      "step": 40
+    },
+    {
+      "epoch": 0.12096774193548387,
+      "grad_norm": 0.4294178783893585,
+      "learning_rate": 2.357142857142857e-05,
+      "loss": 1.1234,
+      "step": 45
+    },
+    {
+      "epoch": 0.13440860215053763,
+      "grad_norm": 0.4342615604400635,
+      "learning_rate": 2.625e-05,
+      "loss": 1.156,
+      "step": 50
+    },
+    {
+      "epoch": 0.1478494623655914,
+      "grad_norm": 0.454303503036499,
+      "learning_rate": 2.892857142857143e-05,
+      "loss": 1.1284,
+      "step": 55
+    },
+    {
+      "epoch": 0.16129032258064516,
+      "grad_norm": 0.4393405616283417,
+      "learning_rate": 2.9999407089793328e-05,
+      "loss": 1.0955,
+      "step": 60
+    },
+    {
+      "epoch": 0.17473118279569894,
+      "grad_norm": 0.5047406554222107,
+      "learning_rate": 2.99957839193835e-05,
+      "loss": 1.1035,
+      "step": 65
+    },
+    {
+      "epoch": 0.1881720430107527,
+      "grad_norm": 0.5270004868507385,
+      "learning_rate": 2.9988867767784412e-05,
+      "loss": 1.0324,
+      "step": 70
+    },
+    {
+      "epoch": 0.20161290322580644,
+      "grad_norm": 0.5382557511329651,
+      "learning_rate": 2.997866015374007e-05,
+      "loss": 1.066,
+      "step": 75
+    },
+    {
+      "epoch": 0.21505376344086022,
+      "grad_norm": 0.536156952381134,
+      "learning_rate": 2.996516331877925e-05,
+      "loss": 0.9768,
+      "step": 80
+    },
+    {
+      "epoch": 0.22849462365591397,
+      "grad_norm": 0.5498602986335754,
+      "learning_rate": 2.994838022672324e-05,
+      "loss": 0.9506,
+      "step": 85
+    },
+    {
+      "epoch": 0.24193548387096775,
+      "grad_norm": 0.5507628321647644,
+      "learning_rate": 2.9928314563035015e-05,
+      "loss": 0.968,
+      "step": 90
+    },
+    {
+      "epoch": 0.2553763440860215,
+      "grad_norm": 0.5989782810211182,
+      "learning_rate": 2.9904970734009933e-05,
+      "loss": 0.9596,
+      "step": 95
+    },
+    {
+      "epoch": 0.26881720430107525,
+      "grad_norm": 0.6400024890899658,
+      "learning_rate": 2.9878353865808144e-05,
+      "loss": 0.9271,
+      "step": 100
+    },
+    {
+      "epoch": 0.28225806451612906,
+      "grad_norm": 0.6517824530601501,
+      "learning_rate": 2.9848469803328896e-05,
+      "loss": 0.8872,
+      "step": 105
+    },
+    {
+      "epoch": 0.2956989247311828,
+      "grad_norm": 0.6882349848747253,
+      "learning_rate": 2.981532510892707e-05,
+      "loss": 0.9132,
+      "step": 110
+    },
+    {
+      "epoch": 0.30913978494623656,
+      "grad_norm": 0.6823952794075012,
+      "learning_rate": 2.9778927060972075e-05,
+      "loss": 0.8172,
+      "step": 115
+    },
+    {
+      "epoch": 0.3225806451612903,
+      "grad_norm": 0.7629667520523071,
+      "learning_rate": 2.9739283652249625e-05,
+      "loss": 0.8335,
+      "step": 120
+    },
+    {
+      "epoch": 0.33602150537634407,
+      "grad_norm": 0.8502324819564819,
+      "learning_rate": 2.9696403588206517e-05,
+      "loss": 0.7916,
+      "step": 125
+    },
+    {
+      "epoch": 0.34946236559139787,
+      "grad_norm": 0.8907939791679382,
+      "learning_rate": 2.9650296285038996e-05,
+      "loss": 0.8016,
+      "step": 130
+    },
+    {
+      "epoch": 0.3629032258064516,
+      "grad_norm": 0.781342625617981,
+      "learning_rate": 2.9600971867625027e-05,
+      "loss": 0.7944,
+      "step": 135
+    },
+    {
+      "epoch": 0.3763440860215054,
+      "grad_norm": 0.9921382665634155,
+      "learning_rate": 2.95484411673009e-05,
+      "loss": 0.7786,
+      "step": 140
+    },
+    {
+      "epoch": 0.3897849462365591,
+      "grad_norm": 0.8458676934242249,
+      "learning_rate": 2.9492715719482776e-05,
+      "loss": 0.7528,
+      "step": 145
+    },
+    {
+      "epoch": 0.4032258064516129,
+      "grad_norm": 0.9752245545387268,
+      "learning_rate": 2.9433807761133542e-05,
+      "loss": 0.7606,
+      "step": 150
+    },
+    {
+      "epoch": 0.4166666666666667,
+      "grad_norm": 0.9003056287765503,
+      "learning_rate": 2.9371730228075684e-05,
+      "loss": 0.7206,
+      "step": 155
+    },
+    {
+      "epoch": 0.43010752688172044,
+      "grad_norm": 1.159876823425293,
+      "learning_rate": 2.9306496752150638e-05,
+      "loss": 0.7178,
+      "step": 160
+    },
+    {
+      "epoch": 0.4435483870967742,
+      "grad_norm": 0.7836427092552185,
+      "learning_rate": 2.9238121658225337e-05,
+      "loss": 0.712,
+      "step": 165
+    },
+    {
+      "epoch": 0.45698924731182794,
+      "grad_norm": 0.8637785315513611,
+      "learning_rate": 2.9166619961046545e-05,
+      "loss": 0.7399,
+      "step": 170
+    },
+    {
+      "epoch": 0.47043010752688175,
+      "grad_norm": 0.9151424765586853,
+      "learning_rate": 2.909200736194372e-05,
+      "loss": 0.6517,
+      "step": 175
+    },
+    {
+      "epoch": 0.4838709677419355,
+      "grad_norm": 0.9069246649742126,
+      "learning_rate": 2.9014300245381095e-05,
+      "loss": 0.6405,
+      "step": 180
+    },
+    {
+      "epoch": 0.49731182795698925,
+      "grad_norm": 1.1238454580307007,
+      "learning_rate": 2.8933515675359766e-05,
+      "loss": 0.6341,
+      "step": 185
+    },
+    {
+      "epoch": 0.510752688172043,
+      "grad_norm": 0.9921523928642273,
+      "learning_rate": 2.8849671391670518e-05,
+      "loss": 0.6437,
+      "step": 190
+    },
+    {
+      "epoch": 0.5241935483870968,
+      "grad_norm": 1.0404704809188843,
+      "learning_rate": 2.8762785805998295e-05,
+      "loss": 0.576,
+      "step": 195
+    },
+    {
+      "epoch": 0.5376344086021505,
+      "grad_norm": 1.1066476106643677,
+      "learning_rate": 2.86728779978791e-05,
+      "loss": 0.6435,
+      "step": 200
+    },
+    {
+      "epoch": 0.5510752688172043,
+      "grad_norm": 0.9322255253791809,
+      "learning_rate": 2.8579967710510264e-05,
+      "loss": 0.6314,
+      "step": 205
+    },
+    {
+      "epoch": 0.5645161290322581,
+      "grad_norm": 1.0714216232299805,
+      "learning_rate": 2.8484075346414936e-05,
+      "loss": 0.5775,
+      "step": 210
+    },
+    {
+      "epoch": 0.5779569892473119,
+      "grad_norm": 1.1178566217422485,
+      "learning_rate": 2.838522196296182e-05,
+      "loss": 0.5588,
+      "step": 215
+    },
+    {
+      "epoch": 0.5913978494623656,
+      "grad_norm": 1.3558969497680664,
+      "learning_rate": 2.8283429267741138e-05,
+      "loss": 0.5886,
+      "step": 220
+    },
+    {
+      "epoch": 0.6048387096774194,
+      "grad_norm": 1.2517259120941162,
+      "learning_rate": 2.8178719613797747e-05,
+      "loss": 0.5579,
+      "step": 225
+    },
+    {
+      "epoch": 0.6182795698924731,
+      "grad_norm": 0.9852866530418396,
+      "learning_rate": 2.807111599472254e-05,
+      "loss": 0.5611,
+      "step": 230
+    },
+    {
+      "epoch": 0.6317204301075269,
+      "grad_norm": 1.151684045791626,
+      "learning_rate": 2.7960642039603235e-05,
+      "loss": 0.5035,
+      "step": 235
+    },
+    {
+      "epoch": 0.6451612903225806,
+      "grad_norm": 1.1494412422180176,
+      "learning_rate": 2.7847322007835546e-05,
+      "loss": 0.5562,
+      "step": 240
+    },
+    {
+      "epoch": 0.6586021505376344,
+      "grad_norm": 1.0049399137496948,
+      "learning_rate": 2.773118078379597e-05,
+      "loss": 0.5434,
+      "step": 245
+    },
+    {
+      "epoch": 0.6720430107526881,
+      "grad_norm": 1.088126301765442,
+      "learning_rate": 2.7612243871377342e-05,
+      "loss": 0.5455,
+      "step": 250
+    },
+    {
+      "epoch": 0.6854838709677419,
+      "grad_norm": 1.015211820602417,
+      "learning_rate": 2.749053738838834e-05,
+      "loss": 0.5391,
+      "step": 255
+    },
+    {
+      "epoch": 0.6989247311827957,
+      "grad_norm": 1.0612668991088867,
+      "learning_rate": 2.7366088060818154e-05,
+      "loss": 0.4849,
+      "step": 260
+    },
+    {
+      "epoch": 0.7123655913978495,
+      "grad_norm": 1.104041337966919,
+      "learning_rate": 2.7238923216967666e-05,
+      "loss": 0.5178,
+      "step": 265
+    },
+    {
+      "epoch": 0.7258064516129032,
+      "grad_norm": 1.1188374757766724,
+      "learning_rate": 2.7109070781448283e-05,
+      "loss": 0.4852,
+      "step": 270
+    },
+    {
+      "epoch": 0.739247311827957,
+      "grad_norm": 1.0869883298873901,
+      "learning_rate": 2.6976559269049875e-05,
+      "loss": 0.5125,
+      "step": 275
+    },
+    {
+      "epoch": 0.7526881720430108,
+      "grad_norm": 1.067639946937561,
+      "learning_rate": 2.6841417778479132e-05,
+      "loss": 0.467,
+      "step": 280
+    },
+    {
+      "epoch": 0.7661290322580645,
+      "grad_norm": 1.2798089981079102,
+      "learning_rate": 2.670367598596963e-05,
+      "loss": 0.4695,
+      "step": 285
+    },
+    {
+      "epoch": 0.7795698924731183,
+      "grad_norm": 1.0810679197311401,
+      "learning_rate": 2.6563364138765137e-05,
+      "loss": 0.5091,
+      "step": 290
+    },
+    {
+      "epoch": 0.793010752688172,
+      "grad_norm": 0.9999101758003235,
+      "learning_rate": 2.6420513048477503e-05,
+      "loss": 0.4423,
+      "step": 295
+    },
+    {
+      "epoch": 0.8064516129032258,
+      "grad_norm": 1.343894362449646,
+      "learning_rate": 2.6275154084320622e-05,
+      "loss": 0.4506,
+      "step": 300
+    },
+    {
+      "epoch": 0.8198924731182796,
+      "grad_norm": 1.4252490997314453,
+      "learning_rate": 2.6127319166221923e-05,
+      "loss": 0.4541,
+      "step": 305
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 1.169596552848816,
+      "learning_rate": 2.5977040757812997e-05,
+      "loss": 0.4665,
+      "step": 310
+    },
+    {
+      "epoch": 0.8467741935483871,
+      "grad_norm": 1.2513952255249023,
+      "learning_rate": 2.5824351859300748e-05,
+      "loss": 0.4287,
+      "step": 315
+    },
+    {
+      "epoch": 0.8602150537634409,
+      "grad_norm": 1.36992347240448,
+      "learning_rate": 2.5669286000220757e-05,
+      "loss": 0.4217,
+      "step": 320
+    },
+    {
+      "epoch": 0.8736559139784946,
+      "grad_norm": 1.3834331035614014,
+      "learning_rate": 2.551187723207443e-05,
+      "loss": 0.4051,
+      "step": 325
+    },
+    {
+      "epoch": 0.8870967741935484,
+      "grad_norm": 1.3238064050674438,
+      "learning_rate": 2.5352160120851464e-05,
+      "loss": 0.3747,
+      "step": 330
+    },
+    {
+      "epoch": 0.9005376344086021,
+      "grad_norm": 1.1013132333755493,
+      "learning_rate": 2.519016973943939e-05,
+      "loss": 0.4215,
+      "step": 335
+    },
+    {
+      "epoch": 0.9139784946236559,
+      "grad_norm": 1.2075629234313965,
+      "learning_rate": 2.5025941659921783e-05,
+      "loss": 0.3955,
+      "step": 340
+    },
+    {
+      "epoch": 0.9274193548387096,
+      "grad_norm": 1.1348271369934082,
+      "learning_rate": 2.485951194576685e-05,
+      "loss": 0.4145,
+      "step": 345
+    },
+    {
+      "epoch": 0.9408602150537635,
+      "grad_norm": 1.4727540016174316,
+      "learning_rate": 2.469091714390811e-05,
+      "loss": 0.365,
+      "step": 350
+    },
+    {
+      "epoch": 0.9543010752688172,
+      "grad_norm": 1.3754587173461914,
+      "learning_rate": 2.4520194276718938e-05,
+      "loss": 0.4179,
+      "step": 355
+    },
+    {
+      "epoch": 0.967741935483871,
+      "grad_norm": 1.2584078311920166,
+      "learning_rate": 2.4347380833882634e-05,
+      "loss": 0.3835,
+      "step": 360
+    },
+    {
+      "epoch": 0.9811827956989247,
+      "grad_norm": 0.9971347451210022,
+      "learning_rate": 2.417251476415998e-05,
+      "loss": 0.3789,
+      "step": 365
+    },
+    {
+      "epoch": 0.9946236559139785,
+      "grad_norm": 1.2069525718688965,
+      "learning_rate": 2.3995634467055882e-05,
+      "loss": 0.3887,
+      "step": 370
+    },
+    {
+      "epoch": 1.0080645161290323,
+      "grad_norm": 1.4079045057296753,
+      "learning_rate": 2.3816778784387097e-05,
+      "loss": 0.3542,
+      "step": 375
+    },
+    {
+      "epoch": 1.021505376344086,
+      "grad_norm": 1.1576917171478271,
+      "learning_rate": 2.363598699175281e-05,
+      "loss": 0.3247,
+      "step": 380
+    },
+    {
+      "epoch": 1.0349462365591398,
+      "grad_norm": 1.0292298793792725,
+      "learning_rate": 2.3453298789909935e-05,
+      "loss": 0.2872,
+      "step": 385
+    },
+    {
+      "epoch": 1.0483870967741935,
+      "grad_norm": 1.1727144718170166,
+      "learning_rate": 2.3268754296055122e-05,
+      "loss": 0.2748,
+      "step": 390
+    },
+    {
+      "epoch": 1.0618279569892473,
+      "grad_norm": 1.1642152070999146,
+      "learning_rate": 2.3082394035015212e-05,
+      "loss": 0.3109,
+      "step": 395
+    },
+    {
+      "epoch": 1.075268817204301,
+      "grad_norm": 1.1566616296768188,
+      "learning_rate": 2.2894258930348284e-05,
+      "loss": 0.3382,
+      "step": 400
+    },
+    {
+      "epoch": 1.0887096774193548,
+      "grad_norm": 1.1437203884124756,
+      "learning_rate": 2.2704390295357056e-05,
+      "loss": 0.2718,
+      "step": 405
+    },
+    {
+      "epoch": 1.1021505376344085,
+      "grad_norm": 1.2010996341705322,
+      "learning_rate": 2.2512829824016773e-05,
+      "loss": 0.2892,
+      "step": 410
+    },
+    {
+      "epoch": 1.1155913978494623,
+      "grad_norm": 1.1548149585723877,
+      "learning_rate": 2.2319619581819458e-05,
+      "loss": 0.2969,
+      "step": 415
+    },
+    {
+      "epoch": 1.129032258064516,
+      "grad_norm": 1.3920344114303589,
+      "learning_rate": 2.2124801996536575e-05,
+      "loss": 0.3025,
+      "step": 420
+    },
+    {
+      "epoch": 1.14247311827957,
+      "grad_norm": 0.9887757897377014,
+      "learning_rate": 2.1928419848902163e-05,
+      "loss": 0.2795,
+      "step": 425
+    },
+    {
+      "epoch": 1.1559139784946237,
+      "grad_norm": 1.1330392360687256,
+      "learning_rate": 2.1730516263218472e-05,
+      "loss": 0.2554,
+      "step": 430
+    },
+    {
+      "epoch": 1.1693548387096775,
+      "grad_norm": 1.2178608179092407,
+      "learning_rate": 2.1531134697886103e-05,
+      "loss": 0.2622,
+      "step": 435
+    },
+    {
+      "epoch": 1.1827956989247312,
+      "grad_norm": 1.1591136455535889,
+      "learning_rate": 2.133031893586083e-05,
+      "loss": 0.2488,
+      "step": 440
+    },
+    {
+      "epoch": 1.196236559139785,
+      "grad_norm": 1.3371623754501343,
+      "learning_rate": 2.1128113075039124e-05,
+      "loss": 0.2546,
+      "step": 445
+    },
+    {
+      "epoch": 1.2096774193548387,
+      "grad_norm": 1.237210988998413,
+      "learning_rate": 2.092456151857455e-05,
+      "loss": 0.2554,
+      "step": 450
+    },
+    {
+      "epoch": 1.2231182795698925,
+      "grad_norm": 1.468112826347351,
+      "learning_rate": 2.0719708965127073e-05,
+      "loss": 0.2905,
+      "step": 455
+    },
+    {
+      "epoch": 1.2365591397849462,
+      "grad_norm": 1.249820590019226,
+      "learning_rate": 2.0513600399047545e-05,
+      "loss": 0.2456,
+      "step": 460
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.289322018623352,
+      "learning_rate": 2.0306281080499413e-05,
+      "loss": 0.26,
+      "step": 465
+    },
+    {
+      "epoch": 1.2634408602150538,
+      "grad_norm": 1.1969671249389648,
+      "learning_rate": 2.009779653551983e-05,
+      "loss": 0.214,
+      "step": 470
+    },
+    {
+      "epoch": 1.2768817204301075,
+      "grad_norm": 1.141692042350769,
+      "learning_rate": 1.9888192546022488e-05,
+      "loss": 0.2424,
+      "step": 475
+    },
+    {
+      "epoch": 1.2903225806451613,
+      "grad_norm": 1.0056333541870117,
+      "learning_rate": 1.9677515139744126e-05,
+      "loss": 0.2209,
+      "step": 480
+    },
+    {
+      "epoch": 1.303763440860215,
+      "grad_norm": 1.447640061378479,
+      "learning_rate": 1.946581058013717e-05,
+      "loss": 0.2355,
+      "step": 485
+    },
+    {
+      "epoch": 1.3172043010752688,
+      "grad_norm": 1.144538164138794,
+      "learning_rate": 1.9253125356210547e-05,
+      "loss": 0.2039,
+      "step": 490
+    },
+    {
+      "epoch": 1.3306451612903225,
+      "grad_norm": 1.264729619026184,
+      "learning_rate": 1.903950617232098e-05,
+      "loss": 0.2286,
+      "step": 495
+    },
+    {
+      "epoch": 1.3440860215053765,
+      "grad_norm": 1.1473528146743774,
+      "learning_rate": 1.8824999937917025e-05,
+      "loss": 0.2195,
+      "step": 500
+    },
+    {
+      "epoch": 1.35752688172043,
+      "grad_norm": 1.1178231239318848,
+      "learning_rate": 1.8609653757238026e-05,
+      "loss": 0.1954,
+      "step": 505
+    },
+    {
+      "epoch": 1.370967741935484,
+      "grad_norm": 1.1528292894363403,
+      "learning_rate": 1.8393514918970315e-05,
+      "loss": 0.2005,
+      "step": 510
+    },
+    {
+      "epoch": 1.3844086021505375,
+      "grad_norm": 1.1556999683380127,
+      "learning_rate": 1.8176630885862927e-05,
+      "loss": 0.212,
+      "step": 515
+    },
+    {
+      "epoch": 1.3978494623655915,
+      "grad_norm": 1.1602588891983032,
+      "learning_rate": 1.7959049284305056e-05,
+      "loss": 0.2268,
+      "step": 520
+    },
+    {
+      "epoch": 1.4112903225806452,
+      "grad_norm": 1.2353614568710327,
+      "learning_rate": 1.7740817893867613e-05,
+      "loss": 0.2299,
+      "step": 525
+    },
+    {
+      "epoch": 1.424731182795699,
+      "grad_norm": 1.1625109910964966,
+      "learning_rate": 1.752198463681111e-05,
+      "loss": 0.2052,
+      "step": 530
+    },
+    {
+      "epoch": 1.4381720430107527,
+      "grad_norm": 1.1569108963012695,
+      "learning_rate": 1.7302597567562237e-05,
+      "loss": 0.1909,
+      "step": 535
+    },
+    {
+      "epoch": 1.4516129032258065,
+      "grad_norm": 1.1060644388198853,
+      "learning_rate": 1.7082704862161422e-05,
+      "loss": 0.2188,
+      "step": 540
+    },
+    {
+      "epoch": 1.4650537634408602,
+      "grad_norm": 1.0977604389190674,
+      "learning_rate": 1.6862354807683643e-05,
+      "loss": 0.1966,
+      "step": 545
+    },
+    {
+      "epoch": 1.478494623655914,
+      "grad_norm": 1.2112979888916016,
+      "learning_rate": 1.66415957916349e-05,
+      "loss": 0.1922,
+      "step": 550
+    },
+    {
+      "epoch": 1.4919354838709677,
+      "grad_norm": 1.2029993534088135,
+      "learning_rate": 1.642047629132663e-05,
+      "loss": 0.2023,
+      "step": 555
+    },
+    {
+      "epoch": 1.5053763440860215,
+      "grad_norm": 1.3151119947433472,
+      "learning_rate": 1.619904486323037e-05,
+      "loss": 0.1957,
+      "step": 560
+    },
+    {
+      "epoch": 1.5188172043010753,
+      "grad_norm": 1.200359582901001,
+      "learning_rate": 1.597735013231507e-05,
+      "loss": 0.1927,
+      "step": 565
+    },
+    {
+      "epoch": 1.532258064516129,
+      "grad_norm": 1.5202263593673706,
+      "learning_rate": 1.5755440781369345e-05,
+      "loss": 0.207,
+      "step": 570
+    },
+    {
+      "epoch": 1.5456989247311828,
+      "grad_norm": 1.3761674165725708,
+      "learning_rate": 1.5533365540311038e-05,
+      "loss": 0.1895,
+      "step": 575
+    },
+    {
+      "epoch": 1.5591397849462365,
+      "grad_norm": 1.169979214668274,
+      "learning_rate": 1.531117317548643e-05,
+      "loss": 0.18,
+      "step": 580
+    },
+    {
+      "epoch": 1.5725806451612905,
+      "grad_norm": 1.1688958406448364,
+      "learning_rate": 1.5088912478961458e-05,
+      "loss": 0.1994,
+      "step": 585
+    },
+    {
+      "epoch": 1.586021505376344,
+      "grad_norm": 1.1405879259109497,
+      "learning_rate": 1.4866632257807278e-05,
+      "loss": 0.1942,
+      "step": 590
+    },
+    {
+      "epoch": 1.599462365591398,
+      "grad_norm": 1.206282615661621,
+      "learning_rate": 1.4644381323382539e-05,
+      "loss": 0.1643,
+      "step": 595
+    },
+    {
+      "epoch": 1.6129032258064515,
+      "grad_norm": 0.9810739755630493,
+      "learning_rate": 1.4422208480614684e-05,
+      "loss": 0.165,
+      "step": 600
+    },
+    {
+      "epoch": 1.6263440860215055,
+      "grad_norm": 1.1487164497375488,
+      "learning_rate": 1.4200162517282736e-05,
+      "loss": 0.1586,
+      "step": 605
+    },
+    {
+      "epoch": 1.639784946236559,
+      "grad_norm": 1.1038219928741455,
+      "learning_rate": 1.3978292193303768e-05,
+      "loss": 0.167,
+      "step": 610
+    },
+    {
+      "epoch": 1.653225806451613,
+      "grad_norm": 1.291054129600525,
+      "learning_rate": 1.3756646230025555e-05,
+      "loss": 0.1495,
+      "step": 615
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 1.1429893970489502,
+      "learning_rate": 1.3535273299527622e-05,
+      "loss": 0.1597,
+      "step": 620
+    },
+    {
+      "epoch": 1.6801075268817205,
+      "grad_norm": 1.0427857637405396,
+      "learning_rate": 1.3314222013933218e-05,
+      "loss": 0.1498,
+      "step": 625
+    },
+    {
+      "epoch": 1.6935483870967742,
+      "grad_norm": 1.3520914316177368,
+      "learning_rate": 1.3093540914734351e-05,
+      "loss": 0.167,
+      "step": 630
+    },
+    {
+      "epoch": 1.706989247311828,
+      "grad_norm": 1.138988971710205,
+      "learning_rate": 1.2873278462132401e-05,
+      "loss": 0.1626,
+      "step": 635
+    },
+    {
+      "epoch": 1.7204301075268817,
+      "grad_norm": 1.2722725868225098,
+      "learning_rate": 1.2653483024396535e-05,
+      "loss": 0.1517,
+      "step": 640
+    },
+    {
+      "epoch": 1.7338709677419355,
+      "grad_norm": 1.4539331197738647,
+      "learning_rate": 1.2434202867242372e-05,
+      "loss": 0.1547,
+      "step": 645
+    },
+    {
+      "epoch": 1.7473118279569892,
+      "grad_norm": 0.9995672702789307,
+      "learning_rate": 1.221548614323308e-05,
+      "loss": 0.1497,
+      "step": 650
+    },
+    {
+      "epoch": 1.760752688172043,
+      "grad_norm": 1.0385746955871582,
+      "learning_rate": 1.1997380881205431e-05,
+      "loss": 0.14,
+      "step": 655
+    },
+    {
+      "epoch": 1.7741935483870968,
+      "grad_norm": 1.258118987083435,
+      "learning_rate": 1.1779934975722919e-05,
+      "loss": 0.1594,
+      "step": 660
+    },
+    {
+      "epoch": 1.7876344086021505,
+      "grad_norm": 1.154404640197754,
+      "learning_rate": 1.1563196176558436e-05,
+      "loss": 0.1662,
+      "step": 665
+    },
+    {
+      "epoch": 1.8010752688172043,
+      "grad_norm": 1.165967583656311,
+      "learning_rate": 1.13472120782087e-05,
+      "loss": 0.1699,
+      "step": 670
+    },
+    {
+      "epoch": 1.814516129032258,
+      "grad_norm": 1.0650913715362549,
+      "learning_rate": 1.1132030109442823e-05,
+      "loss": 0.1542,
+      "step": 675
+    },
+    {
+      "epoch": 1.827956989247312,
+      "grad_norm": 0.9981470704078674,
+      "learning_rate": 1.0917697522887217e-05,
+      "loss": 0.1187,
+      "step": 680
+    },
+    {
+      "epoch": 1.8413978494623655,
+      "grad_norm": 0.9565959572792053,
+      "learning_rate": 1.0704261384649242e-05,
+      "loss": 0.1182,
+      "step": 685
+    },
+    {
+      "epoch": 1.8548387096774195,
+      "grad_norm": 1.1371008157730103,
+      "learning_rate": 1.0491768563981747e-05,
+      "loss": 0.1477,
+      "step": 690
+    },
+    {
+      "epoch": 1.868279569892473,
+      "grad_norm": 1.0350284576416016,
+      "learning_rate": 1.0280265722990908e-05,
+      "loss": 0.1165,
+      "step": 695
+    },
+    {
+      "epoch": 1.881720430107527,
+      "grad_norm": 1.03700852394104,
+      "learning_rate": 1.0069799306389485e-05,
+      "loss": 0.1443,
+      "step": 700
+    },
+    {
+      "epoch": 1.8951612903225805,
+      "grad_norm": 0.9594007730484009,
+      "learning_rate": 9.860415531297881e-06,
+      "loss": 0.1198,
+      "step": 705
+    },
+    {
+      "epoch": 1.9086021505376345,
+      "grad_norm": 1.2112687826156616,
+      "learning_rate": 9.652160377095124e-06,
+      "loss": 0.1396,
+      "step": 710
+    },
+    {
+      "epoch": 1.922043010752688,
+      "grad_norm": 1.053672194480896,
+      "learning_rate": 9.445079575322101e-06,
+      "loss": 0.128,
+      "step": 715
+    },
+    {
+      "epoch": 1.935483870967742,
+      "grad_norm": 1.1258035898208618,
+      "learning_rate": 9.239218599639171e-06,
+      "loss": 0.1009,
+      "step": 720
+    },
+    {
+      "epoch": 1.9489247311827957,
+      "grad_norm": 0.8760501742362976,
+      "learning_rate": 9.03462265584046e-06,
+      "loss": 0.1033,
+      "step": 725
+    },
+    {
+      "epoch": 1.9623655913978495,
+      "grad_norm": 0.8852777481079102,
+      "learning_rate": 8.831336671926924e-06,
+      "loss": 0.1292,
+      "step": 730
+    },
+    {
+      "epoch": 1.9758064516129032,
+      "grad_norm": 1.2534979581832886,
+      "learning_rate": 8.629405288240461e-06,
+      "loss": 0.1085,
+      "step": 735
+    },
+    {
+      "epoch": 1.989247311827957,
+      "grad_norm": 0.9125783443450928,
+      "learning_rate": 8.428872847661139e-06,
+      "loss": 0.1426,
+      "step": 740
+    },
+    {
+      "epoch": 2.002688172043011,
+      "grad_norm": 0.921627938747406,
+      "learning_rate": 8.229783385869807e-06,
+      "loss": 0.1206,
+      "step": 745
+    },
+    {
+      "epoch": 2.0161290322580645,
+      "grad_norm": 1.1075901985168457,
+      "learning_rate": 8.03218062167811e-06,
+      "loss": 0.0912,
+      "step": 750
+    },
+    {
+      "epoch": 2.0295698924731185,
+      "grad_norm": 0.9789226651191711,
+      "learning_rate": 7.836107947428115e-06,
+      "loss": 0.0925,
+      "step": 755
+    },
+    {
+      "epoch": 2.043010752688172,
+      "grad_norm": 1.0046566724777222,
+      "learning_rate": 7.641608419463621e-06,
+      "loss": 0.0882,
+      "step": 760
+    },
+    {
+      "epoch": 2.056451612903226,
+      "grad_norm": 1.0565558671951294,
+      "learning_rate": 7.448724748675273e-06,
+      "loss": 0.0831,
+      "step": 765
+    },
+    {
+      "epoch": 2.0698924731182795,
+      "grad_norm": 0.9930645823478699,
+      "learning_rate": 7.257499291121473e-06,
+      "loss": 0.1022,
+      "step": 770
+    },
+    {
+      "epoch": 2.0833333333333335,
+      "grad_norm": 0.9431966543197632,
+      "learning_rate": 7.0679740387272896e-06,
+      "loss": 0.0901,
+      "step": 775
+    },
+    {
+      "epoch": 2.096774193548387,
+      "grad_norm": 0.7657452821731567,
+      "learning_rate": 6.880190610063272e-06,
+      "loss": 0.0848,
+      "step": 780
+    },
+    {
+      "epoch": 2.110215053763441,
+      "grad_norm": 0.7795389294624329,
+      "learning_rate": 6.694190241206277e-06,
+      "loss": 0.0852,
+      "step": 785
+    },
+    {
+      "epoch": 2.1236559139784945,
+      "grad_norm": 0.8091223239898682,
+      "learning_rate": 6.510013776684281e-06,
+      "loss": 0.1062,
+      "step": 790
+    },
+    {
+      "epoch": 2.1370967741935485,
+      "grad_norm": 0.8571935296058655,
+      "learning_rate": 6.327701660507191e-06,
+      "loss": 0.0751,
+      "step": 795
+    },
+    {
+      "epoch": 2.150537634408602,
+      "grad_norm": 0.9531056880950928,
+      "learning_rate": 6.147293927285537e-06,
+      "loss": 0.072,
+      "step": 800
+    },
+    {
+      "epoch": 2.163978494623656,
+      "grad_norm": 1.0143887996673584,
+      "learning_rate": 5.968830193439195e-06,
+      "loss": 0.0739,
+      "step": 805
+    },
+    {
+      "epoch": 2.1774193548387095,
+      "grad_norm": 0.9456474781036377,
+      "learning_rate": 5.7923496484978115e-06,
+      "loss": 0.064,
+      "step": 810
+    },
+    {
+      "epoch": 2.1908602150537635,
+      "grad_norm": 0.8176582455635071,
+      "learning_rate": 5.617891046495046e-06,
+      "loss": 0.0832,
+      "step": 815
+    },
+    {
+      "epoch": 2.204301075268817,
+      "grad_norm": 0.9019757509231567,
+      "learning_rate": 5.44549269745842e-06,
+      "loss": 0.0703,
+      "step": 820
+    },
+    {
+      "epoch": 2.217741935483871,
+      "grad_norm": 1.0114434957504272,
+      "learning_rate": 5.275192458996682e-06,
+      "loss": 0.0782,
+      "step": 825
+    },
+    {
+      "epoch": 2.2311827956989245,
+      "grad_norm": 0.7793991565704346,
+      "learning_rate": 5.1070277279864926e-06,
+      "loss": 0.0686,
+      "step": 830
+    },
+    {
+      "epoch": 2.2446236559139785,
+      "grad_norm": 0.7833619117736816,
+      "learning_rate": 4.941035432360333e-06,
+      "loss": 0.0692,
+      "step": 835
+    },
+    {
+      "epoch": 2.258064516129032,
+      "grad_norm": 0.7958183884620667,
+      "learning_rate": 4.777252022997327e-06,
+      "loss": 0.0818,
+      "step": 840
+    },
+    {
+      "epoch": 2.271505376344086,
+      "grad_norm": 0.8023877143859863,
+      "learning_rate": 4.615713465718867e-06,
+      "loss": 0.0824,
+      "step": 845
+    },
+    {
+      "epoch": 2.28494623655914,
+      "grad_norm": 0.9039289951324463,
+      "learning_rate": 4.456455233390751e-06,
+      "loss": 0.0618,
+      "step": 850
+    },
+    {
+      "epoch": 2.2983870967741935,
+      "grad_norm": 0.7927225828170776,
+      "learning_rate": 4.299512298133546e-06,
+      "loss": 0.0695,
+      "step": 855
+    },
+    {
+      "epoch": 2.3118279569892475,
+      "grad_norm": 0.6995267868041992,
+      "learning_rate": 4.144919123642947e-06,
+      "loss": 0.0719,
+      "step": 860
+    },
+    {
+      "epoch": 2.325268817204301,
+      "grad_norm": 0.8228567242622375,
+      "learning_rate": 3.992709657621739e-06,
+      "loss": 0.0757,
+      "step": 865
+    },
+    {
+      "epoch": 2.338709677419355,
+      "grad_norm": 0.7862951159477234,
+      "learning_rate": 3.842917324325107e-06,
+      "loss": 0.0765,
+      "step": 870
+    },
+    {
+      "epoch": 2.3521505376344085,
+      "grad_norm": 0.8511716723442078,
+      "learning_rate": 3.6955750172208763e-06,
+      "loss": 0.0751,
+      "step": 875
+    },
+    {
+      "epoch": 2.3655913978494625,
+      "grad_norm": 0.8172338008880615,
+      "learning_rate": 3.550715091766307e-06,
+      "loss": 0.0721,
+      "step": 880
+    },
+    {
+      "epoch": 2.379032258064516,
+      "grad_norm": 0.7338234782218933,
+      "learning_rate": 3.4083693583030306e-06,
+      "loss": 0.0786,
+      "step": 885
+    },
+    {
+      "epoch": 2.39247311827957,
+      "grad_norm": 0.8766265511512756,
+      "learning_rate": 3.268569075071722e-06,
+      "loss": 0.0654,
+      "step": 890
+    },
+    {
+      "epoch": 2.4059139784946235,
+      "grad_norm": 0.8047353029251099,
+      "learning_rate": 3.131344941347937e-06,
+      "loss": 0.0898,
+      "step": 895
+    },
+    {
+      "epoch": 2.4193548387096775,
+      "grad_norm": 1.4056237936019897,
+      "learning_rate": 2.996727090700794e-06,
+      "loss": 0.068,
+      "step": 900
+    },
+    {
+      "epoch": 2.432795698924731,
+      "grad_norm": 0.8166691660881042,
+      "learning_rate": 2.86474508437579e-06,
+      "loss": 0.0661,
+      "step": 905
+    },
+    {
+      "epoch": 2.446236559139785,
+      "grad_norm": 0.8785800337791443,
+      "learning_rate": 2.7354279048033524e-06,
+      "loss": 0.0705,
+      "step": 910
+    },
+    {
+      "epoch": 2.4596774193548385,
+      "grad_norm": 0.7274101972579956,
+      "learning_rate": 2.6088039492344707e-06,
+      "loss": 0.0718,
+      "step": 915
+    },
+    {
+      "epoch": 2.4731182795698925,
+      "grad_norm": 0.8320068717002869,
+      "learning_rate": 2.4849010235048513e-06,
+      "loss": 0.071,
+      "step": 920
+    },
+    {
+      "epoch": 2.486559139784946,
+      "grad_norm": 0.8361717462539673,
+      "learning_rate": 2.3637463359288914e-06,
+      "loss": 0.0706,
+      "step": 925
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.7936782240867615,
+      "learning_rate": 2.2453664913249505e-06,
+      "loss": 0.0725,
+      "step": 930
+    },
+    {
+      "epoch": 2.513440860215054,
+      "grad_norm": 0.7411880493164062,
+      "learning_rate": 2.1297874851730575e-06,
+      "loss": 0.0663,
+      "step": 935
+    },
+    {
+      "epoch": 2.5268817204301075,
+      "grad_norm": 0.7881729602813721,
+      "learning_rate": 2.017034697906489e-06,
+      "loss": 0.066,
+      "step": 940
+    },
+    {
+      "epoch": 2.540322580645161,
+      "grad_norm": 0.7956838011741638,
+      "learning_rate": 1.9071328893383667e-06,
+      "loss": 0.0754,
+      "step": 945
+    },
+    {
+      "epoch": 2.553763440860215,
+      "grad_norm": 0.7309661507606506,
+      "learning_rate": 1.8001061932245654e-06,
+      "loss": 0.07,
+      "step": 950
+    },
+    {
+      "epoch": 2.567204301075269,
+      "grad_norm": 0.7204325795173645,
+      "learning_rate": 1.6959781119640894e-06,
+      "loss": 0.0768,
+      "step": 955
+    },
+    {
+      "epoch": 2.5806451612903225,
+      "grad_norm": 0.8123862147331238,
+      "learning_rate": 1.594771511438095e-06,
+      "loss": 0.0668,
+      "step": 960
+    },
+    {
+      "epoch": 2.5940860215053765,
+      "grad_norm": 0.7669591903686523,
+      "learning_rate": 1.4965086159886694e-06,
+      "loss": 0.0497,
+      "step": 965
+    },
+    {
+      "epoch": 2.60752688172043,
+      "grad_norm": 0.6710484027862549,
+      "learning_rate": 1.401211003538519e-06,
+      "loss": 0.0744,
+      "step": 970
+    },
+    {
+      "epoch": 2.620967741935484,
+      "grad_norm": 0.586686372756958,
+      "learning_rate": 1.308899600852585e-06,
+      "loss": 0.0588,
+      "step": 975
+    },
+    {
+      "epoch": 2.6344086021505375,
+      "grad_norm": 0.6301101446151733,
+      "learning_rate": 1.2195946789426531e-06,
+      "loss": 0.0624,
+      "step": 980
+    },
+    {
+      "epoch": 2.6478494623655915,
+      "grad_norm": 0.8452237844467163,
+      "learning_rate": 1.13331584861597e-06,
+      "loss": 0.0722,
+      "step": 985
+    },
+    {
+      "epoch": 2.661290322580645,
+      "grad_norm": 0.9079617857933044,
+      "learning_rate": 1.0500820561688374e-06,
+      "loss": 0.0678,
+      "step": 990
+    },
+    {
+      "epoch": 2.674731182795699,
+      "grad_norm": 0.7542258501052856,
+      "learning_rate": 9.699115792260953e-07,
+      "loss": 0.0629,
+      "step": 995
+    },
+    {
+      "epoch": 2.688172043010753,
+      "grad_norm": 0.6149124503135681,
+      "learning_rate": 8.928220227275086e-07,
+      "loss": 0.0606,
+      "step": 1000
+    },
+    {
+      "epoch": 2.7016129032258065,
+      "grad_norm": 0.6679168939590454,
+      "learning_rate": 8.18830315061801e-07,
+      "loss": 0.0695,
+      "step": 1005
+    },
+    {
+      "epoch": 2.71505376344086,
+      "grad_norm": 0.6144979000091553,
+      "learning_rate": 7.479527043492984e-07,
+      "loss": 0.0593,
+      "step": 1010
+    },
+    {
+      "epoch": 2.728494623655914,
+      "grad_norm": 0.5943490266799927,
+      "learning_rate": 6.802047548739409e-07,
+      "loss": 0.0614,
+      "step": 1015
+    },
+    {
+      "epoch": 2.741935483870968,
+      "grad_norm": 0.5119675993919373,
+      "learning_rate": 6.156013436654617e-07,
+      "loss": 0.0532,
+      "step": 1020
+    },
+    {
+      "epoch": 2.7553763440860215,
+      "grad_norm": 0.740615725517273,
+      "learning_rate": 5.541566572324786e-07,
+      "loss": 0.0605,
+      "step": 1025
+    },
+    {
+      "epoch": 2.768817204301075,
+      "grad_norm": 0.7387497425079346,
+      "learning_rate": 4.958841884472409e-07,
+      "loss": 0.0708,
+      "step": 1030
+    },
+    {
+      "epoch": 2.782258064516129,
+      "grad_norm": 0.5583337545394897,
+      "learning_rate": 4.407967335826585e-07,
+      "loss": 0.058,
+      "step": 1035
+    },
+    {
+      "epoch": 2.795698924731183,
+      "grad_norm": 0.5696523785591125,
+      "learning_rate": 3.889063895023287e-07,
+      "loss": 0.0559,
+      "step": 1040
+    },
+    {
+      "epoch": 2.8091397849462365,
+      "grad_norm": 0.6738840937614441,
+      "learning_rate": 3.4022455100414184e-07,
+      "loss": 0.0757,
+      "step": 1045
+    },
+    {
+      "epoch": 2.8225806451612905,
+      "grad_norm": 0.6328443288803101,
+      "learning_rate": 2.947619083180525e-07,
+      "loss": 0.0639,
+      "step": 1050
+    },
+    {
+      "epoch": 2.836021505376344,
+      "grad_norm": 0.5463178157806396,
+      "learning_rate": 2.5252844475856906e-07,
+      "loss": 0.052,
+      "step": 1055
+    },
+    {
+      "epoch": 2.849462365591398,
+      "grad_norm": 0.744583010673523,
+      "learning_rate": 2.1353343453248408e-07,
+      "loss": 0.0592,
+      "step": 1060
+    },
+    {
+      "epoch": 2.8629032258064515,
+      "grad_norm": 0.5999979972839355,
+      "learning_rate": 1.777854407023083e-07,
+      "loss": 0.0562,
+      "step": 1065
+    },
+    {
+      "epoch": 2.8763440860215055,
+      "grad_norm": 0.6866050362586975,
+      "learning_rate": 1.4529231330588644e-07,
+      "loss": 0.0683,
+      "step": 1070
+    },
+    {
+      "epoch": 2.889784946236559,
+      "grad_norm": 0.594747006893158,
+      "learning_rate": 1.160611876325679e-07,
+      "loss": 0.0619,
+      "step": 1075
+    },
+    {
+      "epoch": 2.903225806451613,
+      "grad_norm": 0.6902169585227966,
+      "learning_rate": 9.009848265634669e-08,
+      "loss": 0.0618,
+      "step": 1080
+    },
+    {
+      "epoch": 2.9166666666666665,
+      "grad_norm": 0.5708094239234924,
+      "learning_rate": 6.740989962628896e-08,
+      "loss": 0.0575,
+      "step": 1085
+    },
+    {
+      "epoch": 2.9301075268817205,
+      "grad_norm": 0.5824809074401855,
+      "learning_rate": 4.800042081458456e-08,
+      "loss": 0.0541,
+      "step": 1090
+    },
+    {
+      "epoch": 2.943548387096774,
+      "grad_norm": 0.6768313050270081,
+      "learning_rate": 3.187430842245709e-08,
+      "loss": 0.0636,
+      "step": 1095
+    },
+    {
+      "epoch": 2.956989247311828,
+      "grad_norm": 0.6510891318321228,
+      "learning_rate": 1.9035103644222675e-08,
+      "loss": 0.0638,
+      "step": 1100
+    },
+    {
+      "epoch": 2.970430107526882,
+      "grad_norm": 0.6160299181938171,
+      "learning_rate": 9.485625889660288e-09,
+      "loss": 0.0607,
+      "step": 1105
+    },
+    {
+      "epoch": 2.9838709677419355,
+      "grad_norm": 0.68213951587677,
+      "learning_rate": 3.2279721648870162e-09,
+      "loss": 0.0555,
+      "step": 1110
+    },
+    {
+      "epoch": 2.997311827956989,
+      "grad_norm": 0.521416962146759,
+      "learning_rate": 2.6351661186974074e-10,
+      "loss": 0.0611,
+      "step": 1115
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1116,
+  "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": 1.922984504847237e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

base/22_128_e3_3e-5/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:270ed060df7eb8cfc78675c3fd5404d60eaae1ee91e46031f1a9a6e543e70bcd
+size 8273

base/22_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)