Upload folder using huggingface_hub

base/10_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/10_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": [
+    "q_proj",
+    "v_proj",
+    "o_proj",
+    "k_proj",
+    "up_proj",
+    "gate_proj",
+    "down_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

base/10_128_e3_3e-5/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4735d7be4ae7764a72c445500ee7b106195cf425486cbb093cebb41514f27ff5
+size 791751704

base/10_128_e3_3e-5/latest

@@ -0,0 +1 @@
+global_step1041

base/10_128_e3_3e-5/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e0758ad39fd8eeea7cf112233832c0f129ac0d33b608d64e59fd8792fe5dd2c4
+size 16389

base/10_128_e3_3e-5/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6febda9e0cb5ba7f2431eaeed7bbed270e1c8d7f6896ca1967e241eb0226e58b
+size 16389

base/10_128_e3_3e-5/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c72f16bd562150cd67cf2414e9975fe7d60e8b8d791efe0d3b0005ffc8f1e57
+size 16389

base/10_128_e3_3e-5/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4c22ddbd5c663888d7468018d66b5a6968ec9ab81243c070b452055c3172a4a2
+size 16389

base/10_128_e3_3e-5/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e142c91d52f075f662d22263abfc1970fb2eada9b86d9f0eaec7f3898599aafe
+size 16389

base/10_128_e3_3e-5/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f18e3148a0547fa09d502b2dba1b6b684e0ccf5abc8562d1bfe308e6228b581a
+size 16389

base/10_128_e3_3e-5/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4574251a83f783884f9af25eac726851a0023d57a2c9150caec6494b12b2c805
+size 16389

base/10_128_e3_3e-5/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ff57f9674dd6ee9a4c0815cb1b0feaa56a35a76f2c122b6821d56767914d9a74
+size 16389

base/10_128_e3_3e-5/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b0c794195822b65d01876dcce65339a7fc8b865e74a0b67a51912b08b325618e
+size 1401

base/10_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/10_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/10_128_e3_3e-5/trainer_state.json

@@ -0,0 +1,1490 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 1041,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01440922190201729,
+      "grad_norm": 0.8141383528709412,
+      "learning_rate": 2.2641509433962262e-06,
+      "loss": 1.2863,
+      "step": 5
+    },
+    {
+      "epoch": 0.02881844380403458,
+      "grad_norm": 0.6066226363182068,
+      "learning_rate": 5.094339622641509e-06,
+      "loss": 1.2324,
+      "step": 10
+    },
+    {
+      "epoch": 0.043227665706051875,
+      "grad_norm": 0.5625881552696228,
+      "learning_rate": 7.924528301886793e-06,
+      "loss": 1.2923,
+      "step": 15
+    },
+    {
+      "epoch": 0.05763688760806916,
+      "grad_norm": 0.4588092267513275,
+      "learning_rate": 1.0754716981132076e-05,
+      "loss": 1.1993,
+      "step": 20
+    },
+    {
+      "epoch": 0.07204610951008646,
+      "grad_norm": 0.4452914893627167,
+      "learning_rate": 1.358490566037736e-05,
+      "loss": 1.2431,
+      "step": 25
+    },
+    {
+      "epoch": 0.08645533141210375,
+      "grad_norm": 0.45694565773010254,
+      "learning_rate": 1.6415094339622643e-05,
+      "loss": 1.223,
+      "step": 30
+    },
+    {
+      "epoch": 0.10086455331412104,
+      "grad_norm": 0.3892032504081726,
+      "learning_rate": 1.9245283018867924e-05,
+      "loss": 1.1694,
+      "step": 35
+    },
+    {
+      "epoch": 0.11527377521613832,
+      "grad_norm": 0.40726372599601746,
+      "learning_rate": 2.2075471698113208e-05,
+      "loss": 1.1648,
+      "step": 40
+    },
+    {
+      "epoch": 0.12968299711815562,
+      "grad_norm": 0.4167666435241699,
+      "learning_rate": 2.4905660377358492e-05,
+      "loss": 1.1734,
+      "step": 45
+    },
+    {
+      "epoch": 0.1440922190201729,
+      "grad_norm": 0.4388352036476135,
+      "learning_rate": 2.7735849056603773e-05,
+      "loss": 1.1426,
+      "step": 50
+    },
+    {
+      "epoch": 0.1585014409221902,
+      "grad_norm": 0.4387800395488739,
+      "learning_rate": 2.9999924169005146e-05,
+      "loss": 1.1193,
+      "step": 55
+    },
+    {
+      "epoch": 0.1729106628242075,
+      "grad_norm": 0.47398778796195984,
+      "learning_rate": 2.9997270164689188e-05,
+      "loss": 1.0857,
+      "step": 60
+    },
+    {
+      "epoch": 0.1873198847262248,
+      "grad_norm": 0.5207752585411072,
+      "learning_rate": 2.999082537730771e-05,
+      "loss": 1.0514,
+      "step": 65
+    },
+    {
+      "epoch": 0.2017291066282421,
+      "grad_norm": 0.5195696949958801,
+      "learning_rate": 2.998059143587657e-05,
+      "loss": 1.0871,
+      "step": 70
+    },
+    {
+      "epoch": 0.21613832853025935,
+      "grad_norm": 0.5634521245956421,
+      "learning_rate": 2.9966570927176653e-05,
+      "loss": 1.0172,
+      "step": 75
+    },
+    {
+      "epoch": 0.23054755043227665,
+      "grad_norm": 0.5398810505867004,
+      "learning_rate": 2.994876739510005e-05,
+      "loss": 1.0156,
+      "step": 80
+    },
+    {
+      "epoch": 0.24495677233429394,
+      "grad_norm": 0.5460149049758911,
+      "learning_rate": 2.9927185339754245e-05,
+      "loss": 0.9616,
+      "step": 85
+    },
+    {
+      "epoch": 0.25936599423631124,
+      "grad_norm": 0.6154254078865051,
+      "learning_rate": 2.9901830216324694e-05,
+      "loss": 1.0023,
+      "step": 90
+    },
+    {
+      "epoch": 0.2737752161383285,
+      "grad_norm": 0.7125405669212341,
+      "learning_rate": 2.9872708433695907e-05,
+      "loss": 0.9718,
+      "step": 95
+    },
+    {
+      "epoch": 0.2881844380403458,
+      "grad_norm": 0.6797090768814087,
+      "learning_rate": 2.9839827352831522e-05,
+      "loss": 0.8927,
+      "step": 100
+    },
+    {
+      "epoch": 0.3025936599423631,
+      "grad_norm": 0.6263461112976074,
+      "learning_rate": 2.980319528491373e-05,
+      "loss": 0.8913,
+      "step": 105
+    },
+    {
+      "epoch": 0.3170028818443804,
+      "grad_norm": 0.8376904726028442,
+      "learning_rate": 2.976282148924246e-05,
+      "loss": 0.8914,
+      "step": 110
+    },
+    {
+      "epoch": 0.3314121037463977,
+      "grad_norm": 0.7999120950698853,
+      "learning_rate": 2.9718716170894987e-05,
+      "loss": 0.865,
+      "step": 115
+    },
+    {
+      "epoch": 0.345821325648415,
+      "grad_norm": 0.7597854137420654,
+      "learning_rate": 2.967089047814643e-05,
+      "loss": 0.8383,
+      "step": 120
+    },
+    {
+      "epoch": 0.36023054755043227,
+      "grad_norm": 0.7182387113571167,
+      "learning_rate": 2.961935649965188e-05,
+      "loss": 0.8858,
+      "step": 125
+    },
+    {
+      "epoch": 0.3746397694524496,
+      "grad_norm": 0.7395715117454529,
+      "learning_rate": 2.956412726139078e-05,
+      "loss": 0.8279,
+      "step": 130
+    },
+    {
+      "epoch": 0.38904899135446686,
+      "grad_norm": 0.9215261340141296,
+      "learning_rate": 2.9505216723374442e-05,
+      "loss": 0.8062,
+      "step": 135
+    },
+    {
+      "epoch": 0.4034582132564842,
+      "grad_norm": 0.9007133841514587,
+      "learning_rate": 2.9442639776117436e-05,
+      "loss": 0.7809,
+      "step": 140
+    },
+    {
+      "epoch": 0.41786743515850144,
+      "grad_norm": 0.7355058789253235,
+      "learning_rate": 2.9376412236873792e-05,
+      "loss": 0.7754,
+      "step": 145
+    },
+    {
+      "epoch": 0.4322766570605187,
+      "grad_norm": 0.9349695444107056,
+      "learning_rate": 2.9306550845638953e-05,
+      "loss": 0.7773,
+      "step": 150
+    },
+    {
+      "epoch": 0.44668587896253603,
+      "grad_norm": 0.9250203371047974,
+      "learning_rate": 2.9233073260918497e-05,
+      "loss": 0.7484,
+      "step": 155
+    },
+    {
+      "epoch": 0.4610951008645533,
+      "grad_norm": 1.0328516960144043,
+      "learning_rate": 2.9155998055264676e-05,
+      "loss": 0.7429,
+      "step": 160
+    },
+    {
+      "epoch": 0.4755043227665706,
+      "grad_norm": 0.9785757064819336,
+      "learning_rate": 2.907534471058195e-05,
+      "loss": 0.6954,
+      "step": 165
+    },
+    {
+      "epoch": 0.4899135446685879,
+      "grad_norm": 0.8597846627235413,
+      "learning_rate": 2.8991133613202615e-05,
+      "loss": 0.7068,
+      "step": 170
+    },
+    {
+      "epoch": 0.5043227665706052,
+      "grad_norm": 0.8984968662261963,
+      "learning_rate": 2.890338604873387e-05,
+      "loss": 0.6963,
+      "step": 175
+    },
+    {
+      "epoch": 0.5187319884726225,
+      "grad_norm": 1.0543302297592163,
+      "learning_rate": 2.8812124196677585e-05,
+      "loss": 0.6718,
+      "step": 180
+    },
+    {
+      "epoch": 0.5331412103746398,
+      "grad_norm": 1.158306360244751,
+      "learning_rate": 2.871737112482405e-05,
+      "loss": 0.6358,
+      "step": 185
+    },
+    {
+      "epoch": 0.547550432276657,
+      "grad_norm": 0.9802110195159912,
+      "learning_rate": 2.8619150783421303e-05,
+      "loss": 0.6967,
+      "step": 190
+    },
+    {
+      "epoch": 0.5619596541786743,
+      "grad_norm": 0.9862010478973389,
+      "learning_rate": 2.851748799912131e-05,
+      "loss": 0.6437,
+      "step": 195
+    },
+    {
+      "epoch": 0.5763688760806917,
+      "grad_norm": 0.9609131813049316,
+      "learning_rate": 2.8412408468704673e-05,
+      "loss": 0.6722,
+      "step": 200
+    },
+    {
+      "epoch": 0.590778097982709,
+      "grad_norm": 0.8917747735977173,
+      "learning_rate": 2.83039387525854e-05,
+      "loss": 0.6452,
+      "step": 205
+    },
+    {
+      "epoch": 0.6051873198847262,
+      "grad_norm": 1.2040942907333374,
+      "learning_rate": 2.8192106268097336e-05,
+      "loss": 0.6394,
+      "step": 210
+    },
+    {
+      "epoch": 0.6195965417867435,
+      "grad_norm": 1.0708210468292236,
+      "learning_rate": 2.8076939282564054e-05,
+      "loss": 0.5967,
+      "step": 215
+    },
+    {
+      "epoch": 0.6340057636887608,
+      "grad_norm": 1.1458494663238525,
+      "learning_rate": 2.795846690615385e-05,
+      "loss": 0.5424,
+      "step": 220
+    },
+    {
+      "epoch": 0.6484149855907781,
+      "grad_norm": 1.4293025732040405,
+      "learning_rate": 2.7836719084521714e-05,
+      "loss": 0.5777,
+      "step": 225
+    },
+    {
+      "epoch": 0.6628242074927954,
+      "grad_norm": 1.0692564249038696,
+      "learning_rate": 2.7711726591240133e-05,
+      "loss": 0.5182,
+      "step": 230
+    },
+    {
+      "epoch": 0.6772334293948127,
+      "grad_norm": 1.153468132019043,
+      "learning_rate": 2.7583521020020615e-05,
+      "loss": 0.5078,
+      "step": 235
+    },
+    {
+      "epoch": 0.69164265129683,
+      "grad_norm": 1.3504886627197266,
+      "learning_rate": 2.7452134776727875e-05,
+      "loss": 0.5297,
+      "step": 240
+    },
+    {
+      "epoch": 0.7060518731988472,
+      "grad_norm": 1.0638216733932495,
+      "learning_rate": 2.7317601071188823e-05,
+      "loss": 0.5645,
+      "step": 245
+    },
+    {
+      "epoch": 0.7204610951008645,
+      "grad_norm": 1.0662178993225098,
+      "learning_rate": 2.7179953908798246e-05,
+      "loss": 0.5328,
+      "step": 250
+    },
+    {
+      "epoch": 0.7348703170028819,
+      "grad_norm": 1.306769847869873,
+      "learning_rate": 2.7039228081923448e-05,
+      "loss": 0.5356,
+      "step": 255
+    },
+    {
+      "epoch": 0.7492795389048992,
+      "grad_norm": 1.3149913549423218,
+      "learning_rate": 2.6895459161109978e-05,
+      "loss": 0.5303,
+      "step": 260
+    },
+    {
+      "epoch": 0.7636887608069164,
+      "grad_norm": 1.0441232919692993,
+      "learning_rate": 2.6748683486090616e-05,
+      "loss": 0.4887,
+      "step": 265
+    },
+    {
+      "epoch": 0.7780979827089337,
+      "grad_norm": 1.1239460706710815,
+      "learning_rate": 2.6598938156600005e-05,
+      "loss": 0.4516,
+      "step": 270
+    },
+    {
+      "epoch": 0.792507204610951,
+      "grad_norm": 1.1201856136322021,
+      "learning_rate": 2.6446261022997098e-05,
+      "loss": 0.5082,
+      "step": 275
+    },
+    {
+      "epoch": 0.8069164265129684,
+      "grad_norm": 1.248568058013916,
+      "learning_rate": 2.629069067669795e-05,
+      "loss": 0.4358,
+      "step": 280
+    },
+    {
+      "epoch": 0.8213256484149856,
+      "grad_norm": 1.203162670135498,
+      "learning_rate": 2.613226644042114e-05,
+      "loss": 0.4095,
+      "step": 285
+    },
+    {
+      "epoch": 0.8357348703170029,
+      "grad_norm": 1.2084534168243408,
+      "learning_rate": 2.5971028358248396e-05,
+      "loss": 0.462,
+      "step": 290
+    },
+    {
+      "epoch": 0.8501440922190202,
+      "grad_norm": 1.1092439889907837,
+      "learning_rate": 2.5807017185502833e-05,
+      "loss": 0.4447,
+      "step": 295
+    },
+    {
+      "epoch": 0.8645533141210374,
+      "grad_norm": 1.0245718955993652,
+      "learning_rate": 2.5640274378447444e-05,
+      "loss": 0.4718,
+      "step": 300
+    },
+    {
+      "epoch": 0.8789625360230547,
+      "grad_norm": 1.1717734336853027,
+      "learning_rate": 2.5470842083806424e-05,
+      "loss": 0.4796,
+      "step": 305
+    },
+    {
+      "epoch": 0.8933717579250721,
+      "grad_norm": 1.1770378351211548,
+      "learning_rate": 2.5298763128111956e-05,
+      "loss": 0.4011,
+      "step": 310
+    },
+    {
+      "epoch": 0.9077809798270894,
+      "grad_norm": 1.159165859222412,
+      "learning_rate": 2.5124081006879148e-05,
+      "loss": 0.3933,
+      "step": 315
+    },
+    {
+      "epoch": 0.9221902017291066,
+      "grad_norm": 1.2271698713302612,
+      "learning_rate": 2.494683987361193e-05,
+      "loss": 0.4314,
+      "step": 320
+    },
+    {
+      "epoch": 0.9365994236311239,
+      "grad_norm": 1.294250726699829,
+      "learning_rate": 2.4767084528642564e-05,
+      "loss": 0.4143,
+      "step": 325
+    },
+    {
+      "epoch": 0.9510086455331412,
+      "grad_norm": 1.1222279071807861,
+      "learning_rate": 2.458486040780772e-05,
+      "loss": 0.3741,
+      "step": 330
+    },
+    {
+      "epoch": 0.9654178674351584,
+      "grad_norm": 1.171310544013977,
+      "learning_rate": 2.440021357096388e-05,
+      "loss": 0.4214,
+      "step": 335
+    },
+    {
+      "epoch": 0.9798270893371758,
+      "grad_norm": 1.5770543813705444,
+      "learning_rate": 2.4213190690345018e-05,
+      "loss": 0.4145,
+      "step": 340
+    },
+    {
+      "epoch": 0.9942363112391931,
+      "grad_norm": 1.364061713218689,
+      "learning_rate": 2.4023839038765525e-05,
+      "loss": 0.3458,
+      "step": 345
+    },
+    {
+      "epoch": 1.0086455331412103,
+      "grad_norm": 1.1885762214660645,
+      "learning_rate": 2.383220647767127e-05,
+      "loss": 0.3343,
+      "step": 350
+    },
+    {
+      "epoch": 1.0230547550432276,
+      "grad_norm": 1.1836655139923096,
+      "learning_rate": 2.363834144504192e-05,
+      "loss": 0.3435,
+      "step": 355
+    },
+    {
+      "epoch": 1.037463976945245,
+      "grad_norm": 1.0806833505630493,
+      "learning_rate": 2.3442292943147543e-05,
+      "loss": 0.331,
+      "step": 360
+    },
+    {
+      "epoch": 1.0518731988472623,
+      "grad_norm": 1.117882490158081,
+      "learning_rate": 2.324411052616251e-05,
+      "loss": 0.3071,
+      "step": 365
+    },
+    {
+      "epoch": 1.0662824207492796,
+      "grad_norm": 1.2636076211929321,
+      "learning_rate": 2.304384428763998e-05,
+      "loss": 0.3311,
+      "step": 370
+    },
+    {
+      "epoch": 1.080691642651297,
+      "grad_norm": 1.1904758214950562,
+      "learning_rate": 2.2841544847849994e-05,
+      "loss": 0.3004,
+      "step": 375
+    },
+    {
+      "epoch": 1.0951008645533142,
+      "grad_norm": 1.235081434249878,
+      "learning_rate": 2.2637263340984446e-05,
+      "loss": 0.3293,
+      "step": 380
+    },
+    {
+      "epoch": 1.1095100864553313,
+      "grad_norm": 1.1662256717681885,
+      "learning_rate": 2.2431051402232164e-05,
+      "loss": 0.2871,
+      "step": 385
+    },
+    {
+      "epoch": 1.1239193083573487,
+      "grad_norm": 1.1332361698150635,
+      "learning_rate": 2.2222961154727346e-05,
+      "loss": 0.2578,
+      "step": 390
+    },
+    {
+      "epoch": 1.138328530259366,
+      "grad_norm": 1.2158137559890747,
+      "learning_rate": 2.2013045196374645e-05,
+      "loss": 0.3073,
+      "step": 395
+    },
+    {
+      "epoch": 1.1527377521613833,
+      "grad_norm": 1.365811824798584,
+      "learning_rate": 2.1801356586554298e-05,
+      "loss": 0.2758,
+      "step": 400
+    },
+    {
+      "epoch": 1.1671469740634006,
+      "grad_norm": 1.2800344228744507,
+      "learning_rate": 2.1587948832710557e-05,
+      "loss": 0.333,
+      "step": 405
+    },
+    {
+      "epoch": 1.181556195965418,
+      "grad_norm": 1.3023836612701416,
+      "learning_rate": 2.1372875876826892e-05,
+      "loss": 0.2779,
+      "step": 410
+    },
+    {
+      "epoch": 1.195965417867435,
+      "grad_norm": 1.1153329610824585,
+      "learning_rate": 2.1156192081791355e-05,
+      "loss": 0.2883,
+      "step": 415
+    },
+    {
+      "epoch": 1.2103746397694524,
+      "grad_norm": 1.6149669885635376,
+      "learning_rate": 2.093795221765554e-05,
+      "loss": 0.2846,
+      "step": 420
+    },
+    {
+      "epoch": 1.2247838616714697,
+      "grad_norm": 1.1790399551391602,
+      "learning_rate": 2.071821144779066e-05,
+      "loss": 0.2762,
+      "step": 425
+    },
+    {
+      "epoch": 1.239193083573487,
+      "grad_norm": 1.1916753053665161,
+      "learning_rate": 2.049702531494417e-05,
+      "loss": 0.2713,
+      "step": 430
+    },
+    {
+      "epoch": 1.2536023054755043,
+      "grad_norm": 1.1540919542312622,
+      "learning_rate": 2.0274449727200497e-05,
+      "loss": 0.2395,
+      "step": 435
+    },
+    {
+      "epoch": 1.2680115273775217,
+      "grad_norm": 1.1812299489974976,
+      "learning_rate": 2.0050540943849477e-05,
+      "loss": 0.285,
+      "step": 440
+    },
+    {
+      "epoch": 1.282420749279539,
+      "grad_norm": 1.3528060913085938,
+      "learning_rate": 1.9825355561165953e-05,
+      "loss": 0.2876,
+      "step": 445
+    },
+    {
+      "epoch": 1.2968299711815563,
+      "grad_norm": 1.1958235502243042,
+      "learning_rate": 1.959895049810423e-05,
+      "loss": 0.2568,
+      "step": 450
+    },
+    {
+      "epoch": 1.3112391930835736,
+      "grad_norm": 1.162650227546692,
+      "learning_rate": 1.937138298191098e-05,
+      "loss": 0.2804,
+      "step": 455
+    },
+    {
+      "epoch": 1.3256484149855907,
+      "grad_norm": 1.1569205522537231,
+      "learning_rate": 1.914271053366018e-05,
+      "loss": 0.2486,
+      "step": 460
+    },
+    {
+      "epoch": 1.340057636887608,
+      "grad_norm": 1.2222715616226196,
+      "learning_rate": 1.8912990953713812e-05,
+      "loss": 0.2078,
+      "step": 465
+    },
+    {
+      "epoch": 1.3544668587896254,
+      "grad_norm": 1.439195990562439,
+      "learning_rate": 1.8682282307111988e-05,
+      "loss": 0.25,
+      "step": 470
+    },
+    {
+      "epoch": 1.3688760806916427,
+      "grad_norm": 1.2527015209197998,
+      "learning_rate": 1.8450642908896104e-05,
+      "loss": 0.2356,
+      "step": 475
+    },
+    {
+      "epoch": 1.38328530259366,
+      "grad_norm": 1.1780779361724854,
+      "learning_rate": 1.8218131309368876e-05,
+      "loss": 0.2286,
+      "step": 480
+    },
+    {
+      "epoch": 1.397694524495677,
+      "grad_norm": 1.0420624017715454,
+      "learning_rate": 1.798480627929488e-05,
+      "loss": 0.2139,
+      "step": 485
+    },
+    {
+      "epoch": 1.4121037463976944,
+      "grad_norm": 1.4442424774169922,
+      "learning_rate": 1.7750726795045345e-05,
+      "loss": 0.2383,
+      "step": 490
+    },
+    {
+      "epoch": 1.4265129682997117,
+      "grad_norm": 1.294108271598816,
+      "learning_rate": 1.7515952023691022e-05,
+      "loss": 0.2387,
+      "step": 495
+    },
+    {
+      "epoch": 1.440922190201729,
+      "grad_norm": 1.2167388200759888,
+      "learning_rate": 1.728054130804681e-05,
+      "loss": 0.2243,
+      "step": 500
+    },
+    {
+      "epoch": 1.4553314121037464,
+      "grad_norm": 1.1080995798110962,
+      "learning_rate": 1.7044554151672003e-05,
+      "loss": 0.2058,
+      "step": 505
+    },
+    {
+      "epoch": 1.4697406340057637,
+      "grad_norm": 1.2727315425872803,
+      "learning_rate": 1.6808050203829845e-05,
+      "loss": 0.2052,
+      "step": 510
+    },
+    {
+      "epoch": 1.484149855907781,
+      "grad_norm": 1.185162901878357,
+      "learning_rate": 1.657108924441031e-05,
+      "loss": 0.2022,
+      "step": 515
+    },
+    {
+      "epoch": 1.4985590778097984,
+      "grad_norm": 1.1185624599456787,
+      "learning_rate": 1.6333731168819854e-05,
+      "loss": 0.2367,
+      "step": 520
+    },
+    {
+      "epoch": 1.5129682997118157,
+      "grad_norm": 1.4358898401260376,
+      "learning_rate": 1.6096035972841937e-05,
+      "loss": 0.2123,
+      "step": 525
+    },
+    {
+      "epoch": 1.527377521613833,
+      "grad_norm": 1.1585294008255005,
+      "learning_rate": 1.5858063737472222e-05,
+      "loss": 0.2377,
+      "step": 530
+    },
+    {
+      "epoch": 1.54178674351585,
+      "grad_norm": 1.2158011198043823,
+      "learning_rate": 1.5619874613732198e-05,
+      "loss": 0.1965,
+      "step": 535
+    },
+    {
+      "epoch": 1.5561959654178674,
+      "grad_norm": 1.150206208229065,
+      "learning_rate": 1.5381528807465113e-05,
+      "loss": 0.2284,
+      "step": 540
+    },
+    {
+      "epoch": 1.5706051873198847,
+      "grad_norm": 1.1202812194824219,
+      "learning_rate": 1.5143086564118042e-05,
+      "loss": 0.1896,
+      "step": 545
+    },
+    {
+      "epoch": 1.585014409221902,
+      "grad_norm": 1.1779747009277344,
+      "learning_rate": 1.4904608153513986e-05,
+      "loss": 0.1767,
+      "step": 550
+    },
+    {
+      "epoch": 1.5994236311239192,
+      "grad_norm": 1.2527257204055786,
+      "learning_rate": 1.466615385461774e-05,
+      "loss": 0.1853,
+      "step": 555
+    },
+    {
+      "epoch": 1.6138328530259365,
+      "grad_norm": 1.2691919803619385,
+      "learning_rate": 1.4427783940299526e-05,
+      "loss": 0.176,
+      "step": 560
+    },
+    {
+      "epoch": 1.6282420749279538,
+      "grad_norm": 1.1541917324066162,
+      "learning_rate": 1.4189558662100094e-05,
+      "loss": 0.1985,
+      "step": 565
+    },
+    {
+      "epoch": 1.6426512968299711,
+      "grad_norm": 1.1894545555114746,
+      "learning_rate": 1.3951538235001262e-05,
+      "loss": 0.1763,
+      "step": 570
+    },
+    {
+      "epoch": 1.6570605187319885,
+      "grad_norm": 1.088067889213562,
+      "learning_rate": 1.3713782822205703e-05,
+      "loss": 0.2044,
+      "step": 575
+    },
+    {
+      "epoch": 1.6714697406340058,
+      "grad_norm": 1.2625999450683594,
+      "learning_rate": 1.3476352519929766e-05,
+      "loss": 0.1564,
+      "step": 580
+    },
+    {
+      "epoch": 1.685878962536023,
+      "grad_norm": 1.2265729904174805,
+      "learning_rate": 1.3239307342213282e-05,
+      "loss": 0.1646,
+      "step": 585
+    },
+    {
+      "epoch": 1.7002881844380404,
+      "grad_norm": 1.4542834758758545,
+      "learning_rate": 1.3002707205750142e-05,
+      "loss": 0.1975,
+      "step": 590
+    },
+    {
+      "epoch": 1.7146974063400577,
+      "grad_norm": 1.090027093887329,
+      "learning_rate": 1.2766611914743415e-05,
+      "loss": 0.1643,
+      "step": 595
+    },
+    {
+      "epoch": 1.729106628242075,
+      "grad_norm": 1.2392020225524902,
+      "learning_rate": 1.2531081145788989e-05,
+      "loss": 0.1572,
+      "step": 600
+    },
+    {
+      "epoch": 1.7435158501440924,
+      "grad_norm": 1.204676866531372,
+      "learning_rate": 1.2296174432791415e-05,
+      "loss": 0.1618,
+      "step": 605
+    },
+    {
+      "epoch": 1.7579250720461095,
+      "grad_norm": 1.176655650138855,
+      "learning_rate": 1.20619511519158e-05,
+      "loss": 0.143,
+      "step": 610
+    },
+    {
+      "epoch": 1.7723342939481268,
+      "grad_norm": 1.1337347030639648,
+      "learning_rate": 1.1828470506579631e-05,
+      "loss": 0.1502,
+      "step": 615
+    },
+    {
+      "epoch": 1.7867435158501441,
+      "grad_norm": 1.1548479795455933,
+      "learning_rate": 1.1595791512488213e-05,
+      "loss": 0.144,
+      "step": 620
+    },
+    {
+      "epoch": 1.8011527377521612,
+      "grad_norm": 1.1713483333587646,
+      "learning_rate": 1.1363972982717588e-05,
+      "loss": 0.162,
+      "step": 625
+    },
+    {
+      "epoch": 1.8155619596541785,
+      "grad_norm": 1.3299273252487183,
+      "learning_rate": 1.1133073512848635e-05,
+      "loss": 0.1452,
+      "step": 630
+    },
+    {
+      "epoch": 1.8299711815561959,
+      "grad_norm": 1.2308896780014038,
+      "learning_rate": 1.090315146615617e-05,
+      "loss": 0.1629,
+      "step": 635
+    },
+    {
+      "epoch": 1.8443804034582132,
+      "grad_norm": 1.2195062637329102,
+      "learning_rate": 1.0674264958856779e-05,
+      "loss": 0.149,
+      "step": 640
+    },
+    {
+      "epoch": 1.8587896253602305,
+      "grad_norm": 1.3013137578964233,
+      "learning_rate": 1.0446471845419063e-05,
+      "loss": 0.1764,
+      "step": 645
+    },
+    {
+      "epoch": 1.8731988472622478,
+      "grad_norm": 1.0573104619979858,
+      "learning_rate": 1.0219829703940047e-05,
+      "loss": 0.1566,
+      "step": 650
+    },
+    {
+      "epoch": 1.8876080691642652,
+      "grad_norm": 1.0930781364440918,
+      "learning_rate": 9.994395821591501e-06,
+      "loss": 0.1486,
+      "step": 655
+    },
+    {
+      "epoch": 1.9020172910662825,
+      "grad_norm": 1.0063505172729492,
+      "learning_rate": 9.770227180139727e-06,
+      "loss": 0.1047,
+      "step": 660
+    },
+    {
+      "epoch": 1.9164265129682998,
+      "grad_norm": 1.0165241956710815,
+      "learning_rate": 9.54738044154255e-06,
+      "loss": 0.133,
+      "step": 665
+    },
+    {
+      "epoch": 1.9308357348703171,
+      "grad_norm": 1.1598025560379028,
+      "learning_rate": 9.325911933627228e-06,
+      "loss": 0.1312,
+      "step": 670
+    },
+    {
+      "epoch": 1.9452449567723344,
+      "grad_norm": 1.3237215280532837,
+      "learning_rate": 9.10587763585269e-06,
+      "loss": 0.134,
+      "step": 675
+    },
+    {
+      "epoch": 1.9596541786743515,
+      "grad_norm": 1.1646838188171387,
+      "learning_rate": 8.887333165159921e-06,
+      "loss": 0.1261,
+      "step": 680
+    },
+    {
+      "epoch": 1.9740634005763689,
+      "grad_norm": 1.1378811597824097,
+      "learning_rate": 8.67033376191398e-06,
+      "loss": 0.1254,
+      "step": 685
+    },
+    {
+      "epoch": 1.9884726224783862,
+      "grad_norm": 1.0642471313476562,
+      "learning_rate": 8.454934275941129e-06,
+      "loss": 0.1334,
+      "step": 690
+    },
+    {
+      "epoch": 2.0028818443804033,
+      "grad_norm": 0.8658217787742615,
+      "learning_rate": 8.241189152664756e-06,
+      "loss": 0.107,
+      "step": 695
+    },
+    {
+      "epoch": 2.0172910662824206,
+      "grad_norm": 1.0285111665725708,
+      "learning_rate": 8.029152419343472e-06,
+      "loss": 0.1047,
+      "step": 700
+    },
+    {
+      "epoch": 2.031700288184438,
+      "grad_norm": 0.870439887046814,
+      "learning_rate": 7.81887767141492e-06,
+      "loss": 0.0972,
+      "step": 705
+    },
+    {
+      "epoch": 2.0461095100864553,
+      "grad_norm": 0.7768731117248535,
+      "learning_rate": 7.6104180589487354e-06,
+      "loss": 0.0838,
+      "step": 710
+    },
+    {
+      "epoch": 2.0605187319884726,
+      "grad_norm": 0.8978638648986816,
+      "learning_rate": 7.403826273212066e-06,
+      "loss": 0.1069,
+      "step": 715
+    },
+    {
+      "epoch": 2.07492795389049,
+      "grad_norm": 0.9977701306343079,
+      "learning_rate": 7.199154533351086e-06,
+      "loss": 0.0823,
+      "step": 720
+    },
+    {
+      "epoch": 2.089337175792507,
+      "grad_norm": 0.9625939130783081,
+      "learning_rate": 6.996454573191799e-06,
+      "loss": 0.0954,
+      "step": 725
+    },
+    {
+      "epoch": 2.1037463976945245,
+      "grad_norm": 1.164396047592163,
+      "learning_rate": 6.795777628163599e-06,
+      "loss": 0.0768,
+      "step": 730
+    },
+    {
+      "epoch": 2.118155619596542,
+      "grad_norm": 0.9446038603782654,
+      "learning_rate": 6.59717442234869e-06,
+      "loss": 0.1094,
+      "step": 735
+    },
+    {
+      "epoch": 2.132564841498559,
+      "grad_norm": 1.1718053817749023,
+      "learning_rate": 6.400695155660866e-06,
+      "loss": 0.1002,
+      "step": 740
+    },
+    {
+      "epoch": 2.1469740634005765,
+      "grad_norm": 1.3492405414581299,
+      "learning_rate": 6.2063894911567185e-06,
+      "loss": 0.0832,
+      "step": 745
+    },
+    {
+      "epoch": 2.161383285302594,
+      "grad_norm": 1.107492446899414,
+      "learning_rate": 6.0143065424825585e-06,
+      "loss": 0.0922,
+      "step": 750
+    },
+    {
+      "epoch": 2.175792507204611,
+      "grad_norm": 0.856367290019989,
+      "learning_rate": 5.824494861460226e-06,
+      "loss": 0.1009,
+      "step": 755
+    },
+    {
+      "epoch": 2.1902017291066285,
+      "grad_norm": 1.1320542097091675,
+      "learning_rate": 5.6370024258148595e-06,
+      "loss": 0.0932,
+      "step": 760
+    },
+    {
+      "epoch": 2.2046109510086453,
+      "grad_norm": 0.8428774476051331,
+      "learning_rate": 5.451876627047873e-06,
+      "loss": 0.0791,
+      "step": 765
+    },
+    {
+      "epoch": 2.2190201729106627,
+      "grad_norm": 0.9998539090156555,
+      "learning_rate": 5.269164258457997e-06,
+      "loss": 0.0916,
+      "step": 770
+    },
+    {
+      "epoch": 2.23342939481268,
+      "grad_norm": 0.816993772983551,
+      "learning_rate": 5.088911503313577e-06,
+      "loss": 0.0852,
+      "step": 775
+    },
+    {
+      "epoch": 2.2478386167146973,
+      "grad_norm": 0.9639632701873779,
+      "learning_rate": 4.91116392317912e-06,
+      "loss": 0.0914,
+      "step": 780
+    },
+    {
+      "epoch": 2.2622478386167146,
+      "grad_norm": 0.9253155589103699,
+      "learning_rate": 4.735966446398854e-06,
+      "loss": 0.0804,
+      "step": 785
+    },
+    {
+      "epoch": 2.276657060518732,
+      "grad_norm": 1.1870715618133545,
+      "learning_rate": 4.563363356740486e-06,
+      "loss": 0.0834,
+      "step": 790
+    },
+    {
+      "epoch": 2.2910662824207493,
+      "grad_norm": 1.0312397480010986,
+      "learning_rate": 4.393398282201788e-06,
+      "loss": 0.0914,
+      "step": 795
+    },
+    {
+      "epoch": 2.3054755043227666,
+      "grad_norm": 1.0728243589401245,
+      "learning_rate": 4.22611418398298e-06,
+      "loss": 0.0953,
+      "step": 800
+    },
+    {
+      "epoch": 2.319884726224784,
+      "grad_norm": 1.1212364435195923,
+      "learning_rate": 4.0615533456276445e-06,
+      "loss": 0.0879,
+      "step": 805
+    },
+    {
+      "epoch": 2.3342939481268012,
+      "grad_norm": 0.8394720554351807,
+      "learning_rate": 3.8997573623349385e-06,
+      "loss": 0.0869,
+      "step": 810
+    },
+    {
+      "epoch": 2.3487031700288186,
+      "grad_norm": 0.8967013955116272,
+      "learning_rate": 3.7407671304457865e-06,
+      "loss": 0.0939,
+      "step": 815
+    },
+    {
+      "epoch": 2.363112391930836,
+      "grad_norm": 0.8728684782981873,
+      "learning_rate": 3.584622837105702e-06,
+      "loss": 0.0816,
+      "step": 820
+    },
+    {
+      "epoch": 2.377521613832853,
+      "grad_norm": 0.745019793510437,
+      "learning_rate": 3.4313639501069423e-06,
+      "loss": 0.0594,
+      "step": 825
+    },
+    {
+      "epoch": 2.39193083573487,
+      "grad_norm": 0.9238844513893127,
+      "learning_rate": 3.281029207912364e-06,
+      "loss": 0.0724,
+      "step": 830
+    },
+    {
+      "epoch": 2.4063400576368874,
+      "grad_norm": 0.9909102916717529,
+      "learning_rate": 3.1336566098637553e-06,
+      "loss": 0.0831,
+      "step": 835
+    },
+    {
+      "epoch": 2.4207492795389047,
+      "grad_norm": 0.8444496989250183,
+      "learning_rate": 2.989283406576932e-06,
+      "loss": 0.0789,
+      "step": 840
+    },
+    {
+      "epoch": 2.435158501440922,
+      "grad_norm": 0.9357560873031616,
+      "learning_rate": 2.847946090526056e-06,
+      "loss": 0.0762,
+      "step": 845
+    },
+    {
+      "epoch": 2.4495677233429394,
+      "grad_norm": 0.813606321811676,
+      "learning_rate": 2.7096803868196546e-06,
+      "loss": 0.0784,
+      "step": 850
+    },
+    {
+      "epoch": 2.4639769452449567,
+      "grad_norm": 0.8887344002723694,
+      "learning_rate": 2.574521244170554e-06,
+      "loss": 0.0707,
+      "step": 855
+    },
+    {
+      "epoch": 2.478386167146974,
+      "grad_norm": 0.7541307210922241,
+      "learning_rate": 2.442502826062072e-06,
+      "loss": 0.0682,
+      "step": 860
+    },
+    {
+      "epoch": 2.4927953890489913,
+      "grad_norm": 0.6023633480072021,
+      "learning_rate": 2.3136585021126965e-06,
+      "loss": 0.0741,
+      "step": 865
+    },
+    {
+      "epoch": 2.5072046109510087,
+      "grad_norm": 0.7302165031433105,
+      "learning_rate": 2.1880208396413996e-06,
+      "loss": 0.0655,
+      "step": 870
+    },
+    {
+      "epoch": 2.521613832853026,
+      "grad_norm": 0.6301232576370239,
+      "learning_rate": 2.0656215954358025e-06,
+      "loss": 0.0641,
+      "step": 875
+    },
+    {
+      "epoch": 2.5360230547550433,
+      "grad_norm": 0.8045523166656494,
+      "learning_rate": 1.946491707725122e-06,
+      "loss": 0.0755,
+      "step": 880
+    },
+    {
+      "epoch": 2.5504322766570606,
+      "grad_norm": 0.7054874897003174,
+      "learning_rate": 1.8306612883601193e-06,
+      "loss": 0.0783,
+      "step": 885
+    },
+    {
+      "epoch": 2.564841498559078,
+      "grad_norm": 0.7850629687309265,
+      "learning_rate": 1.718159615201853e-06,
+      "loss": 0.078,
+      "step": 890
+    },
+    {
+      "epoch": 2.5792507204610953,
+      "grad_norm": 0.9525399804115295,
+      "learning_rate": 1.6090151247212814e-06,
+      "loss": 0.0872,
+      "step": 895
+    },
+    {
+      "epoch": 2.5936599423631126,
+      "grad_norm": 0.8838265538215637,
+      "learning_rate": 1.503255404811511e-06,
+      "loss": 0.0769,
+      "step": 900
+    },
+    {
+      "epoch": 2.60806916426513,
+      "grad_norm": 0.7777808904647827,
+      "learning_rate": 1.4009071878145502e-06,
+      "loss": 0.0738,
+      "step": 905
+    },
+    {
+      "epoch": 2.6224783861671472,
+      "grad_norm": 0.6544051766395569,
+      "learning_rate": 1.301996343764319e-06,
+      "loss": 0.0752,
+      "step": 910
+    },
+    {
+      "epoch": 2.636887608069164,
+      "grad_norm": 0.8318099975585938,
+      "learning_rate": 1.2065478738475883e-06,
+      "loss": 0.0606,
+      "step": 915
+    },
+    {
+      "epoch": 2.6512968299711814,
+      "grad_norm": 0.7002658247947693,
+      "learning_rate": 1.1145859040846012e-06,
+      "loss": 0.0705,
+      "step": 920
+    },
+    {
+      "epoch": 2.6657060518731988,
+      "grad_norm": 0.7776904702186584,
+      "learning_rate": 1.0261336792308168e-06,
+      "loss": 0.0547,
+      "step": 925
+    },
+    {
+      "epoch": 2.680115273775216,
+      "grad_norm": 0.7371366024017334,
+      "learning_rate": 9.412135569014807e-07,
+      "loss": 0.064,
+      "step": 930
+    },
+    {
+      "epoch": 2.6945244956772334,
+      "grad_norm": 0.9154393672943115,
+      "learning_rate": 8.598470019204047e-07,
+      "loss": 0.0805,
+      "step": 935
+    },
+    {
+      "epoch": 2.7089337175792507,
+      "grad_norm": 0.6822687387466431,
+      "learning_rate": 7.820545808943947e-07,
+      "loss": 0.0691,
+      "step": 940
+    },
+    {
+      "epoch": 2.723342939481268,
+      "grad_norm": 0.6636258363723755,
+      "learning_rate": 7.078559570147542e-07,
+      "loss": 0.0566,
+      "step": 945
+    },
+    {
+      "epoch": 2.7377521613832854,
+      "grad_norm": 0.6640030145645142,
+      "learning_rate": 6.372698850871101e-07,
+      "loss": 0.075,
+      "step": 950
+    },
+    {
+      "epoch": 2.7521613832853027,
+      "grad_norm": 0.5294077396392822,
+      "learning_rate": 5.703142067908613e-07,
+      "loss": 0.0644,
+      "step": 955
+    },
+    {
+      "epoch": 2.76657060518732,
+      "grad_norm": 0.5495630502700806,
+      "learning_rate": 5.070058461694261e-07,
+      "loss": 0.0537,
+      "step": 960
+    },
+    {
+      "epoch": 2.7809798270893373,
+      "grad_norm": 0.5607128739356995,
+      "learning_rate": 4.4736080535244084e-07,
+      "loss": 0.055,
+      "step": 965
+    },
+    {
+      "epoch": 2.795389048991354,
+      "grad_norm": 0.6360466480255127,
+      "learning_rate": 3.9139416051098053e-07,
+      "loss": 0.0493,
+      "step": 970
+    },
+    {
+      "epoch": 2.8097982708933715,
+      "grad_norm": 0.9226232767105103,
+      "learning_rate": 3.391200580468318e-07,
+      "loss": 0.0758,
+      "step": 975
+    },
+    {
+      "epoch": 2.824207492795389,
+      "grad_norm": 0.5957112908363342,
+      "learning_rate": 2.905517110167899e-07,
+      "loss": 0.0628,
+      "step": 980
+    },
+    {
+      "epoch": 2.838616714697406,
+      "grad_norm": 0.7306458353996277,
+      "learning_rate": 2.457013957928472e-07,
+      "loss": 0.0685,
+      "step": 985
+    },
+    {
+      "epoch": 2.8530259365994235,
+      "grad_norm": 0.6501022577285767,
+      "learning_rate": 2.0458044895916516e-07,
+      "loss": 0.0569,
+      "step": 990
+    },
+    {
+      "epoch": 2.867435158501441,
+      "grad_norm": 0.6212237477302551,
+      "learning_rate": 1.6719926444658472e-07,
+      "loss": 0.0768,
+      "step": 995
+    },
+    {
+      "epoch": 2.881844380403458,
+      "grad_norm": 0.7248502373695374,
+      "learning_rate": 1.335672909054081e-07,
+      "loss": 0.0658,
+      "step": 1000
+    },
+    {
+      "epoch": 2.8962536023054755,
+      "grad_norm": 0.7521823644638062,
+      "learning_rate": 1.0369302931710067e-07,
+      "loss": 0.0718,
+      "step": 1005
+    },
+    {
+      "epoch": 2.910662824207493,
+      "grad_norm": 0.5145280957221985,
+      "learning_rate": 7.758403084555499e-08,
+      "loss": 0.0636,
+      "step": 1010
+    },
+    {
+      "epoch": 2.92507204610951,
+      "grad_norm": 0.7226182222366333,
+      "learning_rate": 5.5246894928412396e-08,
+      "loss": 0.066,
+      "step": 1015
+    },
+    {
+      "epoch": 2.9394812680115274,
+      "grad_norm": 0.7902790904045105,
+      "learning_rate": 3.668726760896246e-08,
+      "loss": 0.074,
+      "step": 1020
+    },
+    {
+      "epoch": 2.9538904899135447,
+      "grad_norm": 0.7196719646453857,
+      "learning_rate": 2.190984010901953e-08,
+      "loss": 0.0626,
+      "step": 1025
+    },
+    {
+      "epoch": 2.968299711815562,
+      "grad_norm": 0.757164716720581,
+      "learning_rate": 1.0918347643146254e-08,
+      "loss": 0.0707,
+      "step": 1030
+    },
+    {
+      "epoch": 2.9827089337175794,
+      "grad_norm": 0.8064667582511902,
+      "learning_rate": 3.715568474522146e-09,
+      "loss": 0.0723,
+      "step": 1035
+    },
+    {
+      "epoch": 2.9971181556195967,
+      "grad_norm": 0.6263481974601746,
+      "learning_rate": 3.0332321269865847e-10,
+      "loss": 0.0689,
+      "step": 1040
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1041,
+  "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": 1.7685010145974682e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

base/10_128_e3_3e-5/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:878a9062c8bfd9b3efd7201d29d533e18f79a920fffebf4d3cfb183ba79c6846
+size 8273

base/10_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)