Upload folder using huggingface_hub

base/11_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/11_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": [
+    "k_proj",
+    "down_proj",
+    "gate_proj",
+    "o_proj",
+    "q_proj",
+    "v_proj",
+    "up_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

base/11_128_e3_3e-5/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d120dda6ca79ade1f3c3c7696175237da26824ab446ac51271bbc1c50e1b9cd5
+size 791751704

base/11_128_e3_3e-5/latest

@@ -0,0 +1 @@
+global_step948

base/11_128_e3_3e-5/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:abc7b4722da009108a40f14b15b3822d0ccb2955f9e9fd1b3e51a4444085a961
+size 16389

base/11_128_e3_3e-5/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a48695b3478b3f106d6518a11eb17f5fa5d8ed41a44056767141b38f762c3192
+size 16389

base/11_128_e3_3e-5/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bed335fa8acbeae6a4e970eb522bb0f63577e9fa12bdd6eaa35fb8f69e47b06b
+size 16389

base/11_128_e3_3e-5/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:74d391a388bb204e975d7f2a36522a2104bf538a02f32c49eb0260ae62cfb121
+size 16389

base/11_128_e3_3e-5/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:37d8355cce07d16897ad0aa1bba11f34682dcae6047d5683e72efa082d8fd1a6
+size 16389

base/11_128_e3_3e-5/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:90f6c5c17ba9e3a5643c3ab0d12b9b96bd89a9d6cb5dd9cb5850709ee90d6fcb
+size 16389

base/11_128_e3_3e-5/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9929fd0178066537ea48692463076d0213810988a8170e1cea3b6f4deea26094
+size 16389

base/11_128_e3_3e-5/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8931d1dad4e62b66417935ad642d97ddfd607695fa2b80b212828d19881772e0
+size 16389

base/11_128_e3_3e-5/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5ed9ef9e5cded2fc79a5cbc3fe7ad81ce93cbd970cedccd58ae5996fbee0517e
+size 1401

base/11_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/11_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/11_128_e3_3e-5/trainer_state.json

@@ -0,0 +1,1357 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 948,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01584786053882726,
+      "grad_norm": 1.1275420188903809,
+      "learning_rate": 2.4999999999999998e-06,
+      "loss": 1.2428,
+      "step": 5
+    },
+    {
+      "epoch": 0.03169572107765452,
+      "grad_norm": 0.7254608869552612,
+      "learning_rate": 5.625e-06,
+      "loss": 1.2411,
+      "step": 10
+    },
+    {
+      "epoch": 0.04754358161648178,
+      "grad_norm": 0.47944149374961853,
+      "learning_rate": 8.750000000000001e-06,
+      "loss": 1.2607,
+      "step": 15
+    },
+    {
+      "epoch": 0.06339144215530904,
+      "grad_norm": 0.49795594811439514,
+      "learning_rate": 1.1874999999999999e-05,
+      "loss": 1.2782,
+      "step": 20
+    },
+    {
+      "epoch": 0.07923930269413629,
+      "grad_norm": 0.41080501675605774,
+      "learning_rate": 1.5e-05,
+      "loss": 1.1966,
+      "step": 25
+    },
+    {
+      "epoch": 0.09508716323296355,
+      "grad_norm": 0.48956918716430664,
+      "learning_rate": 1.8125e-05,
+      "loss": 1.2098,
+      "step": 30
+    },
+    {
+      "epoch": 0.1109350237717908,
+      "grad_norm": 0.3955107629299164,
+      "learning_rate": 2.125e-05,
+      "loss": 1.1822,
+      "step": 35
+    },
+    {
+      "epoch": 0.12678288431061807,
+      "grad_norm": 0.4201875329017639,
+      "learning_rate": 2.4375e-05,
+      "loss": 1.1583,
+      "step": 40
+    },
+    {
+      "epoch": 0.14263074484944532,
+      "grad_norm": 0.4281456172466278,
+      "learning_rate": 2.75e-05,
+      "loss": 1.1244,
+      "step": 45
+    },
+    {
+      "epoch": 0.15847860538827258,
+      "grad_norm": 0.4577862024307251,
+      "learning_rate": 2.9999908614866857e-05,
+      "loss": 1.1278,
+      "step": 50
+    },
+    {
+      "epoch": 0.17432646592709986,
+      "grad_norm": 0.5039284825325012,
+      "learning_rate": 2.9996710252122685e-05,
+      "loss": 1.1096,
+      "step": 55
+    },
+    {
+      "epoch": 0.1901743264659271,
+      "grad_norm": 0.5241454839706421,
+      "learning_rate": 2.9988943746159317e-05,
+      "loss": 1.0751,
+      "step": 60
+    },
+    {
+      "epoch": 0.20602218700475436,
+      "grad_norm": 0.48489686846733093,
+      "learning_rate": 2.9976611462729715e-05,
+      "loss": 1.0639,
+      "step": 65
+    },
+    {
+      "epoch": 0.2218700475435816,
+      "grad_norm": 0.5610966682434082,
+      "learning_rate": 2.995971715836687e-05,
+      "loss": 1.0625,
+      "step": 70
+    },
+    {
+      "epoch": 0.23771790808240886,
+      "grad_norm": 0.6409398317337036,
+      "learning_rate": 2.9938265979239552e-05,
+      "loss": 1.0013,
+      "step": 75
+    },
+    {
+      "epoch": 0.25356576862123614,
+      "grad_norm": 0.5345151424407959,
+      "learning_rate": 2.9912264459584732e-05,
+      "loss": 0.984,
+      "step": 80
+    },
+    {
+      "epoch": 0.2694136291600634,
+      "grad_norm": 0.5739941596984863,
+      "learning_rate": 2.988172051971717e-05,
+      "loss": 1.0007,
+      "step": 85
+    },
+    {
+      "epoch": 0.28526148969889065,
+      "grad_norm": 0.5841180682182312,
+      "learning_rate": 2.9846643463616834e-05,
+      "loss": 0.9449,
+      "step": 90
+    },
+    {
+      "epoch": 0.3011093502377179,
+      "grad_norm": 0.581023633480072,
+      "learning_rate": 2.980704397609482e-05,
+      "loss": 0.9434,
+      "step": 95
+    },
+    {
+      "epoch": 0.31695721077654515,
+      "grad_norm": 0.5966278314590454,
+      "learning_rate": 2.9762934119538628e-05,
+      "loss": 0.9134,
+      "step": 100
+    },
+    {
+      "epoch": 0.3328050713153724,
+      "grad_norm": 0.7325734496116638,
+      "learning_rate": 2.9714327330237873e-05,
+      "loss": 0.906,
+      "step": 105
+    },
+    {
+      "epoch": 0.3486529318541997,
+      "grad_norm": 0.6706508994102478,
+      "learning_rate": 2.966123841429144e-05,
+      "loss": 0.8891,
+      "step": 110
+    },
+    {
+      "epoch": 0.36450079239302696,
+      "grad_norm": 0.6813719272613525,
+      "learning_rate": 2.9603683543097406e-05,
+      "loss": 0.8627,
+      "step": 115
+    },
+    {
+      "epoch": 0.3803486529318542,
+      "grad_norm": 0.7000885009765625,
+      "learning_rate": 2.9541680248427087e-05,
+      "loss": 0.7888,
+      "step": 120
+    },
+    {
+      "epoch": 0.39619651347068147,
+      "grad_norm": 0.8879828453063965,
+      "learning_rate": 2.9475247417084672e-05,
+      "loss": 0.7877,
+      "step": 125
+    },
+    {
+      "epoch": 0.4120443740095087,
+      "grad_norm": 0.9493227005004883,
+      "learning_rate": 2.9404405285154146e-05,
+      "loss": 0.8158,
+      "step": 130
+    },
+    {
+      "epoch": 0.42789223454833597,
+      "grad_norm": 0.9030868411064148,
+      "learning_rate": 2.9329175431835152e-05,
+      "loss": 0.8041,
+      "step": 135
+    },
+    {
+      "epoch": 0.4437400950871632,
+      "grad_norm": 0.84649658203125,
+      "learning_rate": 2.9249580772869786e-05,
+      "loss": 0.7954,
+      "step": 140
+    },
+    {
+      "epoch": 0.4595879556259905,
+      "grad_norm": 1.0014506578445435,
+      "learning_rate": 2.9165645553562215e-05,
+      "loss": 0.6614,
+      "step": 145
+    },
+    {
+      "epoch": 0.4754358161648177,
+      "grad_norm": 1.2417408227920532,
+      "learning_rate": 2.907739534139334e-05,
+      "loss": 0.7516,
+      "step": 150
+    },
+    {
+      "epoch": 0.49128367670364503,
+      "grad_norm": 0.8059664368629456,
+      "learning_rate": 2.8984857018232682e-05,
+      "loss": 0.7241,
+      "step": 155
+    },
+    {
+      "epoch": 0.5071315372424723,
+      "grad_norm": 1.1359765529632568,
+      "learning_rate": 2.8888058772149923e-05,
+      "loss": 0.6788,
+      "step": 160
+    },
+    {
+      "epoch": 0.5229793977812995,
+      "grad_norm": 1.0021790266036987,
+      "learning_rate": 2.8787030088828517e-05,
+      "loss": 0.6568,
+      "step": 165
+    },
+    {
+      "epoch": 0.5388272583201268,
+      "grad_norm": 0.9580931663513184,
+      "learning_rate": 2.8681801742584095e-05,
+      "loss": 0.6719,
+      "step": 170
+    },
+    {
+      "epoch": 0.554675118858954,
+      "grad_norm": 1.1519267559051514,
+      "learning_rate": 2.8572405786990293e-05,
+      "loss": 0.668,
+      "step": 175
+    },
+    {
+      "epoch": 0.5705229793977813,
+      "grad_norm": 1.0274407863616943,
+      "learning_rate": 2.8458875545114926e-05,
+      "loss": 0.624,
+      "step": 180
+    },
+    {
+      "epoch": 0.5863708399366085,
+      "grad_norm": 0.9643006920814514,
+      "learning_rate": 2.8341245599369464e-05,
+      "loss": 0.6883,
+      "step": 185
+    },
+    {
+      "epoch": 0.6022187004754358,
+      "grad_norm": 0.9000951051712036,
+      "learning_rate": 2.821955178097488e-05,
+      "loss": 0.6288,
+      "step": 190
+    },
+    {
+      "epoch": 0.618066561014263,
+      "grad_norm": 0.8870799541473389,
+      "learning_rate": 2.8093831159047146e-05,
+      "loss": 0.6188,
+      "step": 195
+    },
+    {
+      "epoch": 0.6339144215530903,
+      "grad_norm": 1.0168464183807373,
+      "learning_rate": 2.7964122029305574e-05,
+      "loss": 0.6573,
+      "step": 200
+    },
+    {
+      "epoch": 0.6497622820919176,
+      "grad_norm": 1.1650798320770264,
+      "learning_rate": 2.78304639024076e-05,
+      "loss": 0.6164,
+      "step": 205
+    },
+    {
+      "epoch": 0.6656101426307448,
+      "grad_norm": 0.998997151851654,
+      "learning_rate": 2.769289749191346e-05,
+      "loss": 0.615,
+      "step": 210
+    },
+    {
+      "epoch": 0.6814580031695721,
+      "grad_norm": 0.9582847356796265,
+      "learning_rate": 2.7551464701884426e-05,
+      "loss": 0.5603,
+      "step": 215
+    },
+    {
+      "epoch": 0.6973058637083994,
+      "grad_norm": 1.0360305309295654,
+      "learning_rate": 2.7406208614118427e-05,
+      "loss": 0.5724,
+      "step": 220
+    },
+    {
+      "epoch": 0.7131537242472267,
+      "grad_norm": 1.3253408670425415,
+      "learning_rate": 2.7257173475026926e-05,
+      "loss": 0.5875,
+      "step": 225
+    },
+    {
+      "epoch": 0.7290015847860539,
+      "grad_norm": 1.0502142906188965,
+      "learning_rate": 2.710440468215703e-05,
+      "loss": 0.5565,
+      "step": 230
+    },
+    {
+      "epoch": 0.7448494453248812,
+      "grad_norm": 1.0638993978500366,
+      "learning_rate": 2.6947948770362945e-05,
+      "loss": 0.5126,
+      "step": 235
+    },
+    {
+      "epoch": 0.7606973058637084,
+      "grad_norm": 1.0780391693115234,
+      "learning_rate": 2.678785339763103e-05,
+      "loss": 0.5206,
+      "step": 240
+    },
+    {
+      "epoch": 0.7765451664025357,
+      "grad_norm": 1.1873353719711304,
+      "learning_rate": 2.6624167330562697e-05,
+      "loss": 0.5242,
+      "step": 245
+    },
+    {
+      "epoch": 0.7923930269413629,
+      "grad_norm": 1.0513725280761719,
+      "learning_rate": 2.6456940429519637e-05,
+      "loss": 0.495,
+      "step": 250
+    },
+    {
+      "epoch": 0.8082408874801902,
+      "grad_norm": 1.1819757223129272,
+      "learning_rate": 2.6286223633435865e-05,
+      "loss": 0.4321,
+      "step": 255
+    },
+    {
+      "epoch": 0.8240887480190174,
+      "grad_norm": 1.0600351095199585,
+      "learning_rate": 2.611206894430125e-05,
+      "loss": 0.4842,
+      "step": 260
+    },
+    {
+      "epoch": 0.8399366085578447,
+      "grad_norm": 1.044066309928894,
+      "learning_rate": 2.5934529411321174e-05,
+      "loss": 0.4666,
+      "step": 265
+    },
+    {
+      "epoch": 0.8557844690966719,
+      "grad_norm": 1.1660919189453125,
+      "learning_rate": 2.575365911475724e-05,
+      "loss": 0.4321,
+      "step": 270
+    },
+    {
+      "epoch": 0.8716323296354992,
+      "grad_norm": 1.1836363077163696,
+      "learning_rate": 2.556951314945392e-05,
+      "loss": 0.4812,
+      "step": 275
+    },
+    {
+      "epoch": 0.8874801901743264,
+      "grad_norm": 1.0469483137130737,
+      "learning_rate": 2.5382147608056104e-05,
+      "loss": 0.4788,
+      "step": 280
+    },
+    {
+      "epoch": 0.9033280507131537,
+      "grad_norm": 1.1866803169250488,
+      "learning_rate": 2.519161956392275e-05,
+      "loss": 0.4715,
+      "step": 285
+    },
+    {
+      "epoch": 0.919175911251981,
+      "grad_norm": 1.1852424144744873,
+      "learning_rate": 2.4997987053741785e-05,
+      "loss": 0.483,
+      "step": 290
+    },
+    {
+      "epoch": 0.9350237717908082,
+      "grad_norm": 1.1492961645126343,
+      "learning_rate": 2.4801309059851586e-05,
+      "loss": 0.4774,
+      "step": 295
+    },
+    {
+      "epoch": 0.9508716323296355,
+      "grad_norm": 1.1926027536392212,
+      "learning_rate": 2.4601645492274334e-05,
+      "loss": 0.4087,
+      "step": 300
+    },
+    {
+      "epoch": 0.9667194928684627,
+      "grad_norm": 1.2466610670089722,
+      "learning_rate": 2.439905717046691e-05,
+      "loss": 0.4191,
+      "step": 305
+    },
+    {
+      "epoch": 0.9825673534072901,
+      "grad_norm": 1.1822142601013184,
+      "learning_rate": 2.419360580479465e-05,
+      "loss": 0.4061,
+      "step": 310
+    },
+    {
+      "epoch": 0.9984152139461173,
+      "grad_norm": 1.1111178398132324,
+      "learning_rate": 2.3985353977733787e-05,
+      "loss": 0.4188,
+      "step": 315
+    },
+    {
+      "epoch": 1.012678288431062,
+      "grad_norm": 1.2942121028900146,
+      "learning_rate": 2.3774365124808236e-05,
+      "loss": 0.3105,
+      "step": 320
+    },
+    {
+      "epoch": 1.0285261489698891,
+      "grad_norm": 1.2817046642303467,
+      "learning_rate": 2.356070351526648e-05,
+      "loss": 0.3652,
+      "step": 325
+    },
+    {
+      "epoch": 1.0443740095087164,
+      "grad_norm": 1.3061411380767822,
+      "learning_rate": 2.3344434232504573e-05,
+      "loss": 0.2954,
+      "step": 330
+    },
+    {
+      "epoch": 1.0602218700475436,
+      "grad_norm": 1.200972080230713,
+      "learning_rate": 2.31256231542411e-05,
+      "loss": 0.3797,
+      "step": 335
+    },
+    {
+      "epoch": 1.076069730586371,
+      "grad_norm": 1.169838786125183,
+      "learning_rate": 2.2904336932450167e-05,
+      "loss": 0.2999,
+      "step": 340
+    },
+    {
+      "epoch": 1.0919175911251982,
+      "grad_norm": 1.2752426862716675,
+      "learning_rate": 2.2680642973058574e-05,
+      "loss": 0.285,
+      "step": 345
+    },
+    {
+      "epoch": 1.1077654516640254,
+      "grad_norm": 1.0887740850448608,
+      "learning_rate": 2.245460941541333e-05,
+      "loss": 0.3287,
+      "step": 350
+    },
+    {
+      "epoch": 1.1236133122028527,
+      "grad_norm": 1.4116795063018799,
+      "learning_rate": 2.222630511152573e-05,
+      "loss": 0.3612,
+      "step": 355
+    },
+    {
+      "epoch": 1.13946117274168,
+      "grad_norm": 1.1738829612731934,
+      "learning_rate": 2.199579960509837e-05,
+      "loss": 0.3088,
+      "step": 360
+    },
+    {
+      "epoch": 1.1553090332805072,
+      "grad_norm": 1.1801519393920898,
+      "learning_rate": 2.176316311034146e-05,
+      "loss": 0.2924,
+      "step": 365
+    },
+    {
+      "epoch": 1.1711568938193344,
+      "grad_norm": 1.577104926109314,
+      "learning_rate": 2.1528466490584914e-05,
+      "loss": 0.2999,
+      "step": 370
+    },
+    {
+      "epoch": 1.1870047543581617,
+      "grad_norm": 1.2856918573379517,
+      "learning_rate": 2.1291781236692657e-05,
+      "loss": 0.3066,
+      "step": 375
+    },
+    {
+      "epoch": 1.202852614896989,
+      "grad_norm": 1.1310962438583374,
+      "learning_rate": 2.1053179445285853e-05,
+      "loss": 0.3198,
+      "step": 380
+    },
+    {
+      "epoch": 1.2187004754358162,
+      "grad_norm": 1.1905473470687866,
+      "learning_rate": 2.0812733796781544e-05,
+      "loss": 0.3052,
+      "step": 385
+    },
+    {
+      "epoch": 1.2345483359746434,
+      "grad_norm": 1.2639427185058594,
+      "learning_rate": 2.0570517533253524e-05,
+      "loss": 0.2971,
+      "step": 390
+    },
+    {
+      "epoch": 1.2503961965134707,
+      "grad_norm": 1.182596206665039,
+      "learning_rate": 2.0326604436122056e-05,
+      "loss": 0.2831,
+      "step": 395
+    },
+    {
+      "epoch": 1.266244057052298,
+      "grad_norm": 1.3307101726531982,
+      "learning_rate": 2.0081068803679374e-05,
+      "loss": 0.2811,
+      "step": 400
+    },
+    {
+      "epoch": 1.2820919175911252,
+      "grad_norm": 1.131519079208374,
+      "learning_rate": 1.983398542845767e-05,
+      "loss": 0.2801,
+      "step": 405
+    },
+    {
+      "epoch": 1.2979397781299524,
+      "grad_norm": 1.3476864099502563,
+      "learning_rate": 1.9585429574446588e-05,
+      "loss": 0.277,
+      "step": 410
+    },
+    {
+      "epoch": 1.3137876386687797,
+      "grad_norm": 1.1894779205322266,
+      "learning_rate": 1.9335476954167073e-05,
+      "loss": 0.2834,
+      "step": 415
+    },
+    {
+      "epoch": 1.329635499207607,
+      "grad_norm": 1.2662113904953003,
+      "learning_rate": 1.9084203705608614e-05,
+      "loss": 0.2949,
+      "step": 420
+    },
+    {
+      "epoch": 1.3454833597464342,
+      "grad_norm": 1.2718911170959473,
+      "learning_rate": 1.883168636903686e-05,
+      "loss": 0.2643,
+      "step": 425
+    },
+    {
+      "epoch": 1.3613312202852614,
+      "grad_norm": 1.3229607343673706,
+      "learning_rate": 1.8578001863678713e-05,
+      "loss": 0.2576,
+      "step": 430
+    },
+    {
+      "epoch": 1.3771790808240887,
+      "grad_norm": 1.1633561849594116,
+      "learning_rate": 1.8323227464292014e-05,
+      "loss": 0.2213,
+      "step": 435
+    },
+    {
+      "epoch": 1.393026941362916,
+      "grad_norm": 1.3053103685379028,
+      "learning_rate": 1.8067440777626853e-05,
+      "loss": 0.2285,
+      "step": 440
+    },
+    {
+      "epoch": 1.4088748019017432,
+      "grad_norm": 1.3128119707107544,
+      "learning_rate": 1.781071971878587e-05,
+      "loss": 0.2447,
+      "step": 445
+    },
+    {
+      "epoch": 1.4247226624405704,
+      "grad_norm": 1.1208722591400146,
+      "learning_rate": 1.7553142487490488e-05,
+      "loss": 0.2251,
+      "step": 450
+    },
+    {
+      "epoch": 1.4405705229793977,
+      "grad_norm": 1.2184315919876099,
+      "learning_rate": 1.7294787544260573e-05,
+      "loss": 0.2513,
+      "step": 455
+    },
+    {
+      "epoch": 1.456418383518225,
+      "grad_norm": 1.3051326274871826,
+      "learning_rate": 1.7035733586514568e-05,
+      "loss": 0.1944,
+      "step": 460
+    },
+    {
+      "epoch": 1.4722662440570522,
+      "grad_norm": 1.1214866638183594,
+      "learning_rate": 1.6776059524597518e-05,
+      "loss": 0.2358,
+      "step": 465
+    },
+    {
+      "epoch": 1.4881141045958794,
+      "grad_norm": 1.1321771144866943,
+      "learning_rate": 1.6515844457744193e-05,
+      "loss": 0.1935,
+      "step": 470
+    },
+    {
+      "epoch": 1.5039619651347067,
+      "grad_norm": 1.0923840999603271,
+      "learning_rate": 1.6255167649984738e-05,
+      "loss": 0.1823,
+      "step": 475
+    },
+    {
+      "epoch": 1.519809825673534,
+      "grad_norm": 1.161553978919983,
+      "learning_rate": 1.5994108506000005e-05,
+      "loss": 0.2032,
+      "step": 480
+    },
+    {
+      "epoch": 1.5356576862123612,
+      "grad_norm": 1.0820062160491943,
+      "learning_rate": 1.57327465469342e-05,
+      "loss": 0.2575,
+      "step": 485
+    },
+    {
+      "epoch": 1.5515055467511885,
+      "grad_norm": 1.3692166805267334,
+      "learning_rate": 1.5471161386171925e-05,
+      "loss": 0.2103,
+      "step": 490
+    },
+    {
+      "epoch": 1.5673534072900157,
+      "grad_norm": 1.4225720167160034,
+      "learning_rate": 1.5209432705087183e-05,
+      "loss": 0.2195,
+      "step": 495
+    },
+    {
+      "epoch": 1.583201267828843,
+      "grad_norm": 1.2195379734039307,
+      "learning_rate": 1.494764022877165e-05,
+      "loss": 0.1747,
+      "step": 500
+    },
+    {
+      "epoch": 1.5990491283676702,
+      "grad_norm": 1.0686793327331543,
+      "learning_rate": 1.4685863701749648e-05,
+      "loss": 0.1823,
+      "step": 505
+    },
+    {
+      "epoch": 1.6148969889064975,
+      "grad_norm": 1.1644694805145264,
+      "learning_rate": 1.4424182863687201e-05,
+      "loss": 0.1743,
+      "step": 510
+    },
+    {
+      "epoch": 1.6307448494453247,
+      "grad_norm": 1.1313848495483398,
+      "learning_rate": 1.4162677425102542e-05,
+      "loss": 0.191,
+      "step": 515
+    },
+    {
+      "epoch": 1.6465927099841522,
+      "grad_norm": 1.136939525604248,
+      "learning_rate": 1.3901427043085528e-05,
+      "loss": 0.1786,
+      "step": 520
+    },
+    {
+      "epoch": 1.6624405705229794,
+      "grad_norm": 1.1411586999893188,
+      "learning_rate": 1.36405112970333e-05,
+      "loss": 0.1538,
+      "step": 525
+    },
+    {
+      "epoch": 1.6782884310618067,
+      "grad_norm": 1.3606233596801758,
+      "learning_rate": 1.3380009664409656e-05,
+      "loss": 0.2164,
+      "step": 530
+    },
+    {
+      "epoch": 1.694136291600634,
+      "grad_norm": 1.1211293935775757,
+      "learning_rate": 1.3120001496535434e-05,
+      "loss": 0.1749,
+      "step": 535
+    },
+    {
+      "epoch": 1.7099841521394612,
+      "grad_norm": 1.9247572422027588,
+      "learning_rate": 1.2860565994417327e-05,
+      "loss": 0.1681,
+      "step": 540
+    },
+    {
+      "epoch": 1.7258320126782885,
+      "grad_norm": 1.3400368690490723,
+      "learning_rate": 1.2601782184622479e-05,
+      "loss": 0.1738,
+      "step": 545
+    },
+    {
+      "epoch": 1.7416798732171157,
+      "grad_norm": 1.2520912885665894,
+      "learning_rate": 1.2343728895206252e-05,
+      "loss": 0.1671,
+      "step": 550
+    },
+    {
+      "epoch": 1.757527733755943,
+      "grad_norm": 1.1265698671340942,
+      "learning_rate": 1.2086484731700424e-05,
+      "loss": 0.1742,
+      "step": 555
+    },
+    {
+      "epoch": 1.7733755942947702,
+      "grad_norm": 1.234559416770935,
+      "learning_rate": 1.1830128053169172e-05,
+      "loss": 0.1693,
+      "step": 560
+    },
+    {
+      "epoch": 1.7892234548335975,
+      "grad_norm": 1.2183938026428223,
+      "learning_rate": 1.1574736948340163e-05,
+      "loss": 0.1677,
+      "step": 565
+    },
+    {
+      "epoch": 1.8050713153724247,
+      "grad_norm": 1.262617826461792,
+      "learning_rate": 1.132038921181796e-05,
+      "loss": 0.1573,
+      "step": 570
+    },
+    {
+      "epoch": 1.820919175911252,
+      "grad_norm": 1.1336777210235596,
+      "learning_rate": 1.1067162320387032e-05,
+      "loss": 0.145,
+      "step": 575
+    },
+    {
+      "epoch": 1.8367670364500792,
+      "grad_norm": 1.1317390203475952,
+      "learning_rate": 1.0815133409411564e-05,
+      "loss": 0.1343,
+      "step": 580
+    },
+    {
+      "epoch": 1.8526148969889065,
+      "grad_norm": 1.1058349609375,
+      "learning_rate": 1.0564379249339306e-05,
+      "loss": 0.1547,
+      "step": 585
+    },
+    {
+      "epoch": 1.8684627575277337,
+      "grad_norm": 1.0509099960327148,
+      "learning_rate": 1.031497622231651e-05,
+      "loss": 0.1513,
+      "step": 590
+    },
+    {
+      "epoch": 1.884310618066561,
+      "grad_norm": 1.1539928913116455,
+      "learning_rate": 1.0067000298921251e-05,
+      "loss": 0.1743,
+      "step": 595
+    },
+    {
+      "epoch": 1.9001584786053882,
+      "grad_norm": 0.8569218516349792,
+      "learning_rate": 9.820527015021981e-06,
+      "loss": 0.1337,
+      "step": 600
+    },
+    {
+      "epoch": 1.9160063391442155,
+      "grad_norm": 1.0981194972991943,
+      "learning_rate": 9.575631448768618e-06,
+      "loss": 0.1416,
+      "step": 605
+    },
+    {
+      "epoch": 1.931854199683043,
+      "grad_norm": 1.0505129098892212,
+      "learning_rate": 9.332388197722995e-06,
+      "loss": 0.1244,
+      "step": 610
+    },
+    {
+      "epoch": 1.9477020602218702,
+      "grad_norm": 0.9961683750152588,
+      "learning_rate": 9.090871356135733e-06,
+      "loss": 0.1245,
+      "step": 615
+    },
+    {
+      "epoch": 1.9635499207606975,
+      "grad_norm": 1.1788432598114014,
+      "learning_rate": 8.851154492376408e-06,
+      "loss": 0.139,
+      "step": 620
+    },
+    {
+      "epoch": 1.9793977812995247,
+      "grad_norm": 1.2589232921600342,
+      "learning_rate": 8.61331062652391e-06,
+      "loss": 0.1574,
+      "step": 625
+    },
+    {
+      "epoch": 1.995245641838352,
+      "grad_norm": 1.131536841392517,
+      "learning_rate": 8.377412208123823e-06,
+      "loss": 0.1409,
+      "step": 630
+    },
+    {
+      "epoch": 2.0095087163232965,
+      "grad_norm": 1.0718803405761719,
+      "learning_rate": 8.143531094119591e-06,
+      "loss": 0.1141,
+      "step": 635
+    },
+    {
+      "epoch": 2.025356576862124,
+      "grad_norm": 0.9724932312965393,
+      "learning_rate": 7.911738526964192e-06,
+      "loss": 0.106,
+      "step": 640
+    },
+    {
+      "epoch": 2.041204437400951,
+      "grad_norm": 1.1464614868164062,
+      "learning_rate": 7.682105112919007e-06,
+      "loss": 0.0931,
+      "step": 645
+    },
+    {
+      "epoch": 2.0570522979397783,
+      "grad_norm": 0.8791486024856567,
+      "learning_rate": 7.454700800546474e-06,
+      "loss": 0.1126,
+      "step": 650
+    },
+    {
+      "epoch": 2.0729001584786055,
+      "grad_norm": 1.1549429893493652,
+      "learning_rate": 7.229594859403049e-06,
+      "loss": 0.1124,
+      "step": 655
+    },
+    {
+      "epoch": 2.088748019017433,
+      "grad_norm": 1.127228856086731,
+      "learning_rate": 7.006855858939049e-06,
+      "loss": 0.1058,
+      "step": 660
+    },
+    {
+      "epoch": 2.10459587955626,
+      "grad_norm": 0.9851636290550232,
+      "learning_rate": 6.7865516476117475e-06,
+      "loss": 0.091,
+      "step": 665
+    },
+    {
+      "epoch": 2.1204437400950873,
+      "grad_norm": 1.2654474973678589,
+      "learning_rate": 6.568749332218045e-06,
+      "loss": 0.1147,
+      "step": 670
+    },
+    {
+      "epoch": 2.1362916006339145,
+      "grad_norm": 1.0832056999206543,
+      "learning_rate": 6.3535152574531025e-06,
+      "loss": 0.0953,
+      "step": 675
+    },
+    {
+      "epoch": 2.152139461172742,
+      "grad_norm": 1.170467495918274,
+      "learning_rate": 6.1409149857011175e-06,
+      "loss": 0.0995,
+      "step": 680
+    },
+    {
+      "epoch": 2.167987321711569,
+      "grad_norm": 0.8074630498886108,
+      "learning_rate": 5.931013277064377e-06,
+      "loss": 0.0951,
+      "step": 685
+    },
+    {
+      "epoch": 2.1838351822503963,
+      "grad_norm": 1.0604583024978638,
+      "learning_rate": 5.72387406963669e-06,
+      "loss": 0.0955,
+      "step": 690
+    },
+    {
+      "epoch": 2.1996830427892236,
+      "grad_norm": 0.9850785136222839,
+      "learning_rate": 5.519560460027236e-06,
+      "loss": 0.1051,
+      "step": 695
+    },
+    {
+      "epoch": 2.215530903328051,
+      "grad_norm": 1.1232699155807495,
+      "learning_rate": 5.318134684140739e-06,
+      "loss": 0.1152,
+      "step": 700
+    },
+    {
+      "epoch": 2.231378763866878,
+      "grad_norm": 1.0174535512924194,
+      "learning_rate": 5.119658098219791e-06,
+      "loss": 0.0962,
+      "step": 705
+    },
+    {
+      "epoch": 2.2472266244057053,
+      "grad_norm": 0.9607807397842407,
+      "learning_rate": 4.92419116015515e-06,
+      "loss": 0.0832,
+      "step": 710
+    },
+    {
+      "epoch": 2.2630744849445326,
+      "grad_norm": 1.0852659940719604,
+      "learning_rate": 4.731793411069669e-06,
+      "loss": 0.0943,
+      "step": 715
+    },
+    {
+      "epoch": 2.27892234548336,
+      "grad_norm": 0.9273315668106079,
+      "learning_rate": 4.5425234571815335e-06,
+      "loss": 0.0767,
+      "step": 720
+    },
+    {
+      "epoch": 2.294770206022187,
+      "grad_norm": 0.8090162873268127,
+      "learning_rate": 4.356438951952189e-06,
+      "loss": 0.0865,
+      "step": 725
+    },
+    {
+      "epoch": 2.3106180665610143,
+      "grad_norm": 0.9195489883422852,
+      "learning_rate": 4.173596578524568e-06,
+      "loss": 0.0668,
+      "step": 730
+    },
+    {
+      "epoch": 2.3264659270998416,
+      "grad_norm": 0.9405513405799866,
+      "learning_rate": 3.994052032456853e-06,
+      "loss": 0.0814,
+      "step": 735
+    },
+    {
+      "epoch": 2.342313787638669,
+      "grad_norm": 0.8571817874908447,
+      "learning_rate": 3.8178600047570675e-06,
+      "loss": 0.0988,
+      "step": 740
+    },
+    {
+      "epoch": 2.358161648177496,
+      "grad_norm": 0.8187187910079956,
+      "learning_rate": 3.645074165223656e-06,
+      "loss": 0.0855,
+      "step": 745
+    },
+    {
+      "epoch": 2.3740095087163233,
+      "grad_norm": 1.2618416547775269,
+      "learning_rate": 3.475747146097153e-06,
+      "loss": 0.076,
+      "step": 750
+    },
+    {
+      "epoch": 2.3898573692551506,
+      "grad_norm": 0.9226362109184265,
+      "learning_rate": 3.309930526027885e-06,
+      "loss": 0.0894,
+      "step": 755
+    },
+    {
+      "epoch": 2.405705229793978,
+      "grad_norm": 1.063637137413025,
+      "learning_rate": 3.1476748143646437e-06,
+      "loss": 0.0694,
+      "step": 760
+    },
+    {
+      "epoch": 2.421553090332805,
+      "grad_norm": 0.9497388601303101,
+      "learning_rate": 2.9890294357689994e-06,
+      "loss": 0.0719,
+      "step": 765
+    },
+    {
+      "epoch": 2.4374009508716323,
+      "grad_norm": 0.8773331642150879,
+      "learning_rate": 2.8340427151601036e-06,
+      "loss": 0.0674,
+      "step": 770
+    },
+    {
+      "epoch": 2.4532488114104596,
+      "grad_norm": 0.7115756273269653,
+      "learning_rate": 2.6827618629944394e-06,
+      "loss": 0.0703,
+      "step": 775
+    },
+    {
+      "epoch": 2.469096671949287,
+      "grad_norm": 0.8550273776054382,
+      "learning_rate": 2.5352329608850783e-06,
+      "loss": 0.0707,
+      "step": 780
+    },
+    {
+      "epoch": 2.484944532488114,
+      "grad_norm": 0.8449233770370483,
+      "learning_rate": 2.3915009475647364e-06,
+      "loss": 0.0784,
+      "step": 785
+    },
+    {
+      "epoch": 2.5007923930269413,
+      "grad_norm": 0.8761749267578125,
+      "learning_rate": 2.2516096051970438e-06,
+      "loss": 0.069,
+      "step": 790
+    },
+    {
+      "epoch": 2.5166402535657686,
+      "grad_norm": 0.8362906575202942,
+      "learning_rate": 2.1156015460400333e-06,
+      "loss": 0.0769,
+      "step": 795
+    },
+    {
+      "epoch": 2.532488114104596,
+      "grad_norm": 0.6792451739311218,
+      "learning_rate": 1.9835181994660754e-06,
+      "loss": 0.0685,
+      "step": 800
+    },
+    {
+      "epoch": 2.548335974643423,
+      "grad_norm": 0.7980534434318542,
+      "learning_rate": 1.8553997993420495e-06,
+      "loss": 0.0748,
+      "step": 805
+    },
+    {
+      "epoch": 2.5641838351822503,
+      "grad_norm": 0.8626962304115295,
+      "learning_rate": 1.731285371773741e-06,
+      "loss": 0.0738,
+      "step": 810
+    },
+    {
+      "epoch": 2.5800316957210776,
+      "grad_norm": 0.8830509185791016,
+      "learning_rate": 1.6112127232181163e-06,
+      "loss": 0.0692,
+      "step": 815
+    },
+    {
+      "epoch": 2.595879556259905,
+      "grad_norm": 0.704272449016571,
+      "learning_rate": 1.4952184289670974e-06,
+      "loss": 0.0618,
+      "step": 820
+    },
+    {
+      "epoch": 2.611727416798732,
+      "grad_norm": 0.6761613488197327,
+      "learning_rate": 1.3833378220063713e-06,
+      "loss": 0.073,
+      "step": 825
+    },
+    {
+      "epoch": 2.6275752773375594,
+      "grad_norm": 0.810330331325531,
+      "learning_rate": 1.2756049822526288e-06,
+      "loss": 0.0678,
+      "step": 830
+    },
+    {
+      "epoch": 2.6434231378763866,
+      "grad_norm": 0.8599700927734375,
+      "learning_rate": 1.172052726172494e-06,
+      "loss": 0.0866,
+      "step": 835
+    },
+    {
+      "epoch": 2.659270998415214,
+      "grad_norm": 0.7906323075294495,
+      "learning_rate": 1.0727125967862972e-06,
+      "loss": 0.072,
+      "step": 840
+    },
+    {
+      "epoch": 2.675118858954041,
+      "grad_norm": 0.7346165180206299,
+      "learning_rate": 9.776148540597834e-07,
+      "loss": 0.0704,
+      "step": 845
+    },
+    {
+      "epoch": 2.6909667194928684,
+      "grad_norm": 0.7947611808776855,
+      "learning_rate": 8.867884656866181e-07,
+      "loss": 0.0694,
+      "step": 850
+    },
+    {
+      "epoch": 2.7068145800316956,
+      "grad_norm": 0.7332427501678467,
+      "learning_rate": 8.002610982645558e-07,
+      "loss": 0.0616,
+      "step": 855
+    },
+    {
+      "epoch": 2.722662440570523,
+      "grad_norm": 0.9564700126647949,
+      "learning_rate": 7.180591088679212e-07,
+      "loss": 0.069,
+      "step": 860
+    },
+    {
+      "epoch": 2.73851030110935,
+      "grad_norm": 0.6761453747749329,
+      "learning_rate": 6.402075370189914e-07,
+      "loss": 0.0703,
+      "step": 865
+    },
+    {
+      "epoch": 2.7543581616481774,
+      "grad_norm": 0.8223546743392944,
+      "learning_rate": 5.667300970607192e-07,
+      "loss": 0.0848,
+      "step": 870
+    },
+    {
+      "epoch": 2.7702060221870046,
+      "grad_norm": 0.7804328203201294,
+      "learning_rate": 4.976491709331094e-07,
+      "loss": 0.0663,
+      "step": 875
+    },
+    {
+      "epoch": 2.786053882725832,
+      "grad_norm": 0.694176435470581,
+      "learning_rate": 4.3298580135546053e-07,
+      "loss": 0.066,
+      "step": 880
+    },
+    {
+      "epoch": 2.801901743264659,
+      "grad_norm": 0.8208397030830383,
+      "learning_rate": 3.7275968541655104e-07,
+      "loss": 0.0744,
+      "step": 885
+    },
+    {
+      "epoch": 2.8177496038034864,
+      "grad_norm": 0.6727555394172668,
+      "learning_rate": 3.16989168574725e-07,
+      "loss": 0.0689,
+      "step": 890
+    },
+    {
+      "epoch": 2.8335974643423136,
+      "grad_norm": 1.0596997737884521,
+      "learning_rate": 2.6569123906967083e-07,
+      "loss": 0.0815,
+      "step": 895
+    },
+    {
+      "epoch": 2.849445324881141,
+      "grad_norm": 0.7818054556846619,
+      "learning_rate": 2.1888152274764872e-07,
+      "loss": 0.0656,
+      "step": 900
+    },
+    {
+      "epoch": 2.8652931854199686,
+      "grad_norm": 0.6917042136192322,
+      "learning_rate": 1.7657427830170824e-07,
+      "loss": 0.0793,
+      "step": 905
+    },
+    {
+      "epoch": 2.8811410459587954,
+      "grad_norm": 0.6354867219924927,
+      "learning_rate": 1.3878239292834604e-07,
+      "loss": 0.0726,
+      "step": 910
+    },
+    {
+      "epoch": 2.896988906497623,
+      "grad_norm": 0.6291057467460632,
+      "learning_rate": 1.0551737840194587e-07,
+      "loss": 0.0581,
+      "step": 915
+    },
+    {
+      "epoch": 2.91283676703645,
+      "grad_norm": 0.767151951789856,
+      "learning_rate": 7.67893675681769e-08,
+      "loss": 0.0644,
+      "step": 920
+    },
+    {
+      "epoch": 2.9286846275752776,
+      "grad_norm": 0.6630634665489197,
+      "learning_rate": 5.260711125743445e-08,
+      "loss": 0.0756,
+      "step": 925
+    },
+    {
+      "epoch": 2.9445324881141044,
+      "grad_norm": 0.6594980359077454,
+      "learning_rate": 3.2977975619250536e-08,
+      "loss": 0.0658,
+      "step": 930
+    },
+    {
+      "epoch": 2.960380348652932,
+      "grad_norm": 0.5983220934867859,
+      "learning_rate": 1.7907939878490376e-08,
+      "loss": 0.0719,
+      "step": 935
+    },
+    {
+      "epoch": 2.976228209191759,
+      "grad_norm": 0.6329766511917114,
+      "learning_rate": 7.401594514026e-09,
+      "loss": 0.0735,
+      "step": 940
+    },
+    {
+      "epoch": 2.9920760697305866,
+      "grad_norm": 0.7154616117477417,
+      "learning_rate": 1.4621398604364179e-09,
+      "loss": 0.0647,
+      "step": 945
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 948,
+  "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": 1.624269140179026e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

base/11_128_e3_3e-5/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:50e7cc30afb4cc92c237a02a271cd1b3b4c0dc4a783e9f729cfc17f8f2dd28e0
+size 8273

base/11_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)