Upload folder using huggingface_hub

38/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

38/adapter_config.json

@@ -0,0 +1,39 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "ibm-granite/granite-3.3-8b-base",
+  "bias": "none",
+  "corda_config": null,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 256,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 128,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "q_proj",
+    "o_proj",
+    "gate_proj",
+    "up_proj",
+    "down_proj",
+    "k_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

38/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:892109cabed38f92e6841c0e985798c6cf33c49714d4379123679998cd740cd1
+size 791751704

38/latest

@@ -0,0 +1 @@
+global_step966

38/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:42038d63df0c35501d497eef87341bea733726fc5700a2a08917d3d8e9c437fa
+size 15920

38/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4e34b790f00c4ab5c747d86f070b9e3f1a7b15e3527452838f44da6954982c23
+size 15920

38/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6c4072dd2071d87e7d3342d993b5365d9f3d5546d9ee034a8e206b881b3081c0
+size 15920

38/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:43bdb17046edc4ae1dad104457b49dc057599b61a748edd1571bee3019ef5868
+size 15920

38/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a85db63d4a11accd4c4eecf0d3bc7c87e71260d9d00ff03154d6215f68e3717
+size 15920

38/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ea8fd9a22b62aca3cabcd93de40387a760ca7dff5c8e9fd2fb68550a0b6fc499
+size 15920

38/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:556defb47820a1247fd5960a9247f10ae09075943ff1be5a0b42dd7c00c85be0
+size 15920

38/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d38c5e503594b7e9d47123fd7ad0a3e7a5b70f285e77b4c726c03f43f40b219a
+size 15920

38/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2affc0e2016014fefaec458d8e1d2fb0850f347d73aa965992254d16943b1d7e
+size 1064

38/special_tokens_map.json

@@ -0,0 +1,51 @@
+{
+  "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": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

38/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": "<|endoftext|>",
+  "padding_side": "left",
+  "tokenizer_class": "GPT2Tokenizer",
+  "unk_token": "<|endoftext|>",
+  "vocab_size": 49152
+}

38/trainer_state.json

@@ -0,0 +1,1385 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 966,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.010351966873706004,
+      "grad_norm": 1.1085225343704224,
+      "learning_rate": 9.917355371900827e-07,
+      "loss": 1.2488,
+      "step": 5
+    },
+    {
+      "epoch": 0.020703933747412008,
+      "grad_norm": 1.2809991836547852,
+      "learning_rate": 2.231404958677686e-06,
+      "loss": 1.3141,
+      "step": 10
+    },
+    {
+      "epoch": 0.031055900621118012,
+      "grad_norm": 0.7257665395736694,
+      "learning_rate": 3.4710743801652895e-06,
+      "loss": 1.3325,
+      "step": 15
+    },
+    {
+      "epoch": 0.041407867494824016,
+      "grad_norm": 0.57087641954422,
+      "learning_rate": 4.710743801652893e-06,
+      "loss": 1.2479,
+      "step": 20
+    },
+    {
+      "epoch": 0.051759834368530024,
+      "grad_norm": 0.4889119863510132,
+      "learning_rate": 5.9504132231404965e-06,
+      "loss": 1.257,
+      "step": 25
+    },
+    {
+      "epoch": 0.062111801242236024,
+      "grad_norm": 0.546435534954071,
+      "learning_rate": 7.1900826446281e-06,
+      "loss": 1.2388,
+      "step": 30
+    },
+    {
+      "epoch": 0.07246376811594203,
+      "grad_norm": 0.5399731993675232,
+      "learning_rate": 8.429752066115703e-06,
+      "loss": 1.2067,
+      "step": 35
+    },
+    {
+      "epoch": 0.08281573498964803,
+      "grad_norm": 0.4598652422428131,
+      "learning_rate": 9.669421487603305e-06,
+      "loss": 1.1797,
+      "step": 40
+    },
+    {
+      "epoch": 0.09316770186335403,
+      "grad_norm": 0.5164291858673096,
+      "learning_rate": 1.0909090909090909e-05,
+      "loss": 1.2163,
+      "step": 45
+    },
+    {
+      "epoch": 0.10351966873706005,
+      "grad_norm": 0.5329040288925171,
+      "learning_rate": 1.2148760330578513e-05,
+      "loss": 1.1808,
+      "step": 50
+    },
+    {
+      "epoch": 0.11387163561076605,
+      "grad_norm": 0.48895642161369324,
+      "learning_rate": 1.3388429752066117e-05,
+      "loss": 1.1609,
+      "step": 55
+    },
+    {
+      "epoch": 0.12422360248447205,
+      "grad_norm": 0.4498719274997711,
+      "learning_rate": 1.4628099173553719e-05,
+      "loss": 1.2152,
+      "step": 60
+    },
+    {
+      "epoch": 0.13457556935817805,
+      "grad_norm": 0.49965572357177734,
+      "learning_rate": 1.5867768595041323e-05,
+      "loss": 1.1666,
+      "step": 65
+    },
+    {
+      "epoch": 0.14492753623188406,
+      "grad_norm": 0.40445607900619507,
+      "learning_rate": 1.7107438016528925e-05,
+      "loss": 1.1829,
+      "step": 70
+    },
+    {
+      "epoch": 0.15527950310559005,
+      "grad_norm": 0.4943860173225403,
+      "learning_rate": 1.8347107438016527e-05,
+      "loss": 1.1806,
+      "step": 75
+    },
+    {
+      "epoch": 0.16563146997929606,
+      "grad_norm": 0.45567524433135986,
+      "learning_rate": 1.9586776859504133e-05,
+      "loss": 1.1481,
+      "step": 80
+    },
+    {
+      "epoch": 0.17598343685300208,
+      "grad_norm": 0.43203210830688477,
+      "learning_rate": 2.0826446280991735e-05,
+      "loss": 1.1783,
+      "step": 85
+    },
+    {
+      "epoch": 0.18633540372670807,
+      "grad_norm": 0.46975600719451904,
+      "learning_rate": 2.2066115702479338e-05,
+      "loss": 1.1676,
+      "step": 90
+    },
+    {
+      "epoch": 0.19668737060041408,
+      "grad_norm": 0.5382651686668396,
+      "learning_rate": 2.3305785123966943e-05,
+      "loss": 1.1316,
+      "step": 95
+    },
+    {
+      "epoch": 0.2070393374741201,
+      "grad_norm": 0.467735230922699,
+      "learning_rate": 2.454545454545455e-05,
+      "loss": 1.1187,
+      "step": 100
+    },
+    {
+      "epoch": 0.21739130434782608,
+      "grad_norm": 0.5939744114875793,
+      "learning_rate": 2.578512396694215e-05,
+      "loss": 1.0799,
+      "step": 105
+    },
+    {
+      "epoch": 0.2277432712215321,
+      "grad_norm": 0.49806952476501465,
+      "learning_rate": 2.7024793388429753e-05,
+      "loss": 1.1269,
+      "step": 110
+    },
+    {
+      "epoch": 0.23809523809523808,
+      "grad_norm": 0.4695712924003601,
+      "learning_rate": 2.8264462809917356e-05,
+      "loss": 1.0667,
+      "step": 115
+    },
+    {
+      "epoch": 0.2484472049689441,
+      "grad_norm": 0.5918298959732056,
+      "learning_rate": 2.950413223140496e-05,
+      "loss": 1.0781,
+      "step": 120
+    },
+    {
+      "epoch": 0.2587991718426501,
+      "grad_norm": 0.47227102518081665,
+      "learning_rate": 2.999987340513785e-05,
+      "loss": 1.1187,
+      "step": 125
+    },
+    {
+      "epoch": 0.2691511387163561,
+      "grad_norm": 0.5420899987220764,
+      "learning_rate": 2.9999099777607477e-05,
+      "loss": 1.0795,
+      "step": 130
+    },
+    {
+      "epoch": 0.2795031055900621,
+      "grad_norm": 0.5801032185554504,
+      "learning_rate": 2.9997622889254703e-05,
+      "loss": 1.0691,
+      "step": 135
+    },
+    {
+      "epoch": 0.2898550724637681,
+      "grad_norm": 0.5146046876907349,
+      "learning_rate": 2.9995442809326197e-05,
+      "loss": 1.0526,
+      "step": 140
+    },
+    {
+      "epoch": 0.3002070393374741,
+      "grad_norm": 0.6078941822052002,
+      "learning_rate": 2.999255964003909e-05,
+      "loss": 1.0572,
+      "step": 145
+    },
+    {
+      "epoch": 0.3105590062111801,
+      "grad_norm": 0.6194486021995544,
+      "learning_rate": 2.998897351657615e-05,
+      "loss": 0.9861,
+      "step": 150
+    },
+    {
+      "epoch": 0.32091097308488614,
+      "grad_norm": 0.6928229331970215,
+      "learning_rate": 2.9984684607079488e-05,
+      "loss": 1.0391,
+      "step": 155
+    },
+    {
+      "epoch": 0.33126293995859213,
+      "grad_norm": 0.6550882458686829,
+      "learning_rate": 2.997969311264263e-05,
+      "loss": 1.016,
+      "step": 160
+    },
+    {
+      "epoch": 0.3416149068322981,
+      "grad_norm": 0.662232518196106,
+      "learning_rate": 2.997399926730113e-05,
+      "loss": 1.0466,
+      "step": 165
+    },
+    {
+      "epoch": 0.35196687370600416,
+      "grad_norm": 0.5352460145950317,
+      "learning_rate": 2.996760333802156e-05,
+      "loss": 1.0018,
+      "step": 170
+    },
+    {
+      "epoch": 0.36231884057971014,
+      "grad_norm": 0.6422825455665588,
+      "learning_rate": 2.9960505624689024e-05,
+      "loss": 0.9842,
+      "step": 175
+    },
+    {
+      "epoch": 0.37267080745341613,
+      "grad_norm": 0.6380168199539185,
+      "learning_rate": 2.9952706460093073e-05,
+      "loss": 1.0388,
+      "step": 180
+    },
+    {
+      "epoch": 0.3830227743271222,
+      "grad_norm": 0.648824155330658,
+      "learning_rate": 2.99442062099121e-05,
+      "loss": 0.9385,
+      "step": 185
+    },
+    {
+      "epoch": 0.39337474120082816,
+      "grad_norm": 0.6799726486206055,
+      "learning_rate": 2.993500527269624e-05,
+      "loss": 0.992,
+      "step": 190
+    },
+    {
+      "epoch": 0.40372670807453415,
+      "grad_norm": 0.6741364598274231,
+      "learning_rate": 2.9925104079848617e-05,
+      "loss": 0.9377,
+      "step": 195
+    },
+    {
+      "epoch": 0.4140786749482402,
+      "grad_norm": 0.5715411901473999,
+      "learning_rate": 2.9914503095605166e-05,
+      "loss": 0.975,
+      "step": 200
+    },
+    {
+      "epoch": 0.4244306418219462,
+      "grad_norm": 0.6836850643157959,
+      "learning_rate": 2.9903202817012837e-05,
+      "loss": 0.9512,
+      "step": 205
+    },
+    {
+      "epoch": 0.43478260869565216,
+      "grad_norm": 0.6438029408454895,
+      "learning_rate": 2.9891203773906314e-05,
+      "loss": 0.9681,
+      "step": 210
+    },
+    {
+      "epoch": 0.4451345755693582,
+      "grad_norm": 0.6084693670272827,
+      "learning_rate": 2.9878506528883152e-05,
+      "loss": 0.9366,
+      "step": 215
+    },
+    {
+      "epoch": 0.4554865424430642,
+      "grad_norm": 0.7058281898498535,
+      "learning_rate": 2.9865111677277417e-05,
+      "loss": 0.923,
+      "step": 220
+    },
+    {
+      "epoch": 0.4658385093167702,
+      "grad_norm": 0.690096914768219,
+      "learning_rate": 2.9851019847131757e-05,
+      "loss": 0.9176,
+      "step": 225
+    },
+    {
+      "epoch": 0.47619047619047616,
+      "grad_norm": 0.7328043580055237,
+      "learning_rate": 2.9836231699167956e-05,
+      "loss": 0.9718,
+      "step": 230
+    },
+    {
+      "epoch": 0.4865424430641822,
+      "grad_norm": 0.7374163866043091,
+      "learning_rate": 2.9820747926755975e-05,
+      "loss": 0.9398,
+      "step": 235
+    },
+    {
+      "epoch": 0.4968944099378882,
+      "grad_norm": 0.6593307256698608,
+      "learning_rate": 2.980456925588141e-05,
+      "loss": 0.8945,
+      "step": 240
+    },
+    {
+      "epoch": 0.5072463768115942,
+      "grad_norm": 0.7611751556396484,
+      "learning_rate": 2.9787696445111486e-05,
+      "loss": 0.9214,
+      "step": 245
+    },
+    {
+      "epoch": 0.5175983436853002,
+      "grad_norm": 0.6727657914161682,
+      "learning_rate": 2.9770130285559462e-05,
+      "loss": 0.8661,
+      "step": 250
+    },
+    {
+      "epoch": 0.5279503105590062,
+      "grad_norm": 0.8605769276618958,
+      "learning_rate": 2.9751871600847557e-05,
+      "loss": 0.9108,
+      "step": 255
+    },
+    {
+      "epoch": 0.5383022774327122,
+      "grad_norm": 0.7828345894813538,
+      "learning_rate": 2.973292124706833e-05,
+      "loss": 0.8643,
+      "step": 260
+    },
+    {
+      "epoch": 0.5486542443064182,
+      "grad_norm": 0.7450469136238098,
+      "learning_rate": 2.9713280112744518e-05,
+      "loss": 0.9014,
+      "step": 265
+    },
+    {
+      "epoch": 0.5590062111801242,
+      "grad_norm": 0.7156685590744019,
+      "learning_rate": 2.969294911878742e-05,
+      "loss": 0.8531,
+      "step": 270
+    },
+    {
+      "epoch": 0.5693581780538303,
+      "grad_norm": 0.8362778425216675,
+      "learning_rate": 2.9671929218453672e-05,
+      "loss": 0.8251,
+      "step": 275
+    },
+    {
+      "epoch": 0.5797101449275363,
+      "grad_norm": 0.7471181750297546,
+      "learning_rate": 2.9650221397300584e-05,
+      "loss": 0.8198,
+      "step": 280
+    },
+    {
+      "epoch": 0.5900621118012422,
+      "grad_norm": 0.7804244756698608,
+      "learning_rate": 2.9627826673139925e-05,
+      "loss": 0.9199,
+      "step": 285
+    },
+    {
+      "epoch": 0.6004140786749482,
+      "grad_norm": 0.7540392875671387,
+      "learning_rate": 2.9604746095990182e-05,
+      "loss": 0.8299,
+      "step": 290
+    },
+    {
+      "epoch": 0.6107660455486542,
+      "grad_norm": 0.7669722437858582,
+      "learning_rate": 2.9580980748027337e-05,
+      "loss": 0.8949,
+      "step": 295
+    },
+    {
+      "epoch": 0.6211180124223602,
+      "grad_norm": 0.7878426909446716,
+      "learning_rate": 2.9556531743534153e-05,
+      "loss": 0.8905,
+      "step": 300
+    },
+    {
+      "epoch": 0.6314699792960663,
+      "grad_norm": 0.7589171528816223,
+      "learning_rate": 2.953140022884789e-05,
+      "loss": 0.8481,
+      "step": 305
+    },
+    {
+      "epoch": 0.6418219461697723,
+      "grad_norm": 0.7870969176292419,
+      "learning_rate": 2.950558738230657e-05,
+      "loss": 0.809,
+      "step": 310
+    },
+    {
+      "epoch": 0.6521739130434783,
+      "grad_norm": 0.7980630397796631,
+      "learning_rate": 2.947909441419376e-05,
+      "loss": 0.8635,
+      "step": 315
+    },
+    {
+      "epoch": 0.6625258799171843,
+      "grad_norm": 0.7470607757568359,
+      "learning_rate": 2.9451922566681757e-05,
+      "loss": 0.8426,
+      "step": 320
+    },
+    {
+      "epoch": 0.6728778467908902,
+      "grad_norm": 0.7770550847053528,
+      "learning_rate": 2.9424073113773423e-05,
+      "loss": 0.8376,
+      "step": 325
+    },
+    {
+      "epoch": 0.6832298136645962,
+      "grad_norm": 0.7686053514480591,
+      "learning_rate": 2.9395547361242396e-05,
+      "loss": 0.8111,
+      "step": 330
+    },
+    {
+      "epoch": 0.6935817805383023,
+      "grad_norm": 0.8475646376609802,
+      "learning_rate": 2.9366346646571887e-05,
+      "loss": 0.7939,
+      "step": 335
+    },
+    {
+      "epoch": 0.7039337474120083,
+      "grad_norm": 0.8654535412788391,
+      "learning_rate": 2.9336472338891976e-05,
+      "loss": 0.8479,
+      "step": 340
+    },
+    {
+      "epoch": 0.7142857142857143,
+      "grad_norm": 0.8916240930557251,
+      "learning_rate": 2.9305925838915405e-05,
+      "loss": 0.7793,
+      "step": 345
+    },
+    {
+      "epoch": 0.7246376811594203,
+      "grad_norm": 0.8215227127075195,
+      "learning_rate": 2.9274708578871913e-05,
+      "loss": 0.8094,
+      "step": 350
+    },
+    {
+      "epoch": 0.7349896480331263,
+      "grad_norm": 0.9009755849838257,
+      "learning_rate": 2.924282202244106e-05,
+      "loss": 0.792,
+      "step": 355
+    },
+    {
+      "epoch": 0.7453416149068323,
+      "grad_norm": 0.9841690063476562,
+      "learning_rate": 2.9210267664683647e-05,
+      "loss": 0.7932,
+      "step": 360
+    },
+    {
+      "epoch": 0.7556935817805382,
+      "grad_norm": 0.8054786324501038,
+      "learning_rate": 2.9177047031971567e-05,
+      "loss": 0.7797,
+      "step": 365
+    },
+    {
+      "epoch": 0.7660455486542443,
+      "grad_norm": 0.9446910619735718,
+      "learning_rate": 2.9143161681916264e-05,
+      "loss": 0.7841,
+      "step": 370
+    },
+    {
+      "epoch": 0.7763975155279503,
+      "grad_norm": 0.99616938829422,
+      "learning_rate": 2.91086132032957e-05,
+      "loss": 0.8371,
+      "step": 375
+    },
+    {
+      "epoch": 0.7867494824016563,
+      "grad_norm": 0.8234410881996155,
+      "learning_rate": 2.9073403215979856e-05,
+      "loss": 0.766,
+      "step": 380
+    },
+    {
+      "epoch": 0.7971014492753623,
+      "grad_norm": 1.5421980619430542,
+      "learning_rate": 2.9037533370854783e-05,
+      "loss": 0.713,
+      "step": 385
+    },
+    {
+      "epoch": 0.8074534161490683,
+      "grad_norm": 0.8199692368507385,
+      "learning_rate": 2.9001005349745206e-05,
+      "loss": 0.8018,
+      "step": 390
+    },
+    {
+      "epoch": 0.8178053830227743,
+      "grad_norm": 0.9766334891319275,
+      "learning_rate": 2.8963820865335652e-05,
+      "loss": 0.7383,
+      "step": 395
+    },
+    {
+      "epoch": 0.8281573498964804,
+      "grad_norm": 0.8232831954956055,
+      "learning_rate": 2.8925981661090162e-05,
+      "loss": 0.7658,
+      "step": 400
+    },
+    {
+      "epoch": 0.8385093167701864,
+      "grad_norm": 0.8931108117103577,
+      "learning_rate": 2.8887489511170534e-05,
+      "loss": 0.6995,
+      "step": 405
+    },
+    {
+      "epoch": 0.8488612836438924,
+      "grad_norm": 0.8014031052589417,
+      "learning_rate": 2.884834622035316e-05,
+      "loss": 0.7635,
+      "step": 410
+    },
+    {
+      "epoch": 0.8592132505175983,
+      "grad_norm": 0.8512042760848999,
+      "learning_rate": 2.8808553623944366e-05,
+      "loss": 0.6997,
+      "step": 415
+    },
+    {
+      "epoch": 0.8695652173913043,
+      "grad_norm": 0.810701310634613,
+      "learning_rate": 2.8768113587694405e-05,
+      "loss": 0.7593,
+      "step": 420
+    },
+    {
+      "epoch": 0.8799171842650103,
+      "grad_norm": 0.8463972210884094,
+      "learning_rate": 2.8727028007709946e-05,
+      "loss": 0.7262,
+      "step": 425
+    },
+    {
+      "epoch": 0.8902691511387164,
+      "grad_norm": 0.8761401772499084,
+      "learning_rate": 2.868529881036518e-05,
+      "loss": 0.6839,
+      "step": 430
+    },
+    {
+      "epoch": 0.9006211180124224,
+      "grad_norm": 0.8870850801467896,
+      "learning_rate": 2.864292795221151e-05,
+      "loss": 0.6602,
+      "step": 435
+    },
+    {
+      "epoch": 0.9109730848861284,
+      "grad_norm": 0.9297683835029602,
+      "learning_rate": 2.8599917419885803e-05,
+      "loss": 0.6789,
+      "step": 440
+    },
+    {
+      "epoch": 0.9213250517598344,
+      "grad_norm": 0.9556940197944641,
+      "learning_rate": 2.855626923001723e-05,
+      "loss": 0.7168,
+      "step": 445
+    },
+    {
+      "epoch": 0.9316770186335404,
+      "grad_norm": 0.9701102375984192,
+      "learning_rate": 2.8511985429132752e-05,
+      "loss": 0.6728,
+      "step": 450
+    },
+    {
+      "epoch": 0.9420289855072463,
+      "grad_norm": 0.8694581389427185,
+      "learning_rate": 2.8467068093561125e-05,
+      "loss": 0.7127,
+      "step": 455
+    },
+    {
+      "epoch": 0.9523809523809523,
+      "grad_norm": 0.9791345596313477,
+      "learning_rate": 2.8421519329335562e-05,
+      "loss": 0.6749,
+      "step": 460
+    },
+    {
+      "epoch": 0.9627329192546584,
+      "grad_norm": 1.0260063409805298,
+      "learning_rate": 2.8375341272095004e-05,
+      "loss": 0.6664,
+      "step": 465
+    },
+    {
+      "epoch": 0.9730848861283644,
+      "grad_norm": 0.8952248096466064,
+      "learning_rate": 2.832853608698394e-05,
+      "loss": 0.6665,
+      "step": 470
+    },
+    {
+      "epoch": 0.9834368530020704,
+      "grad_norm": 0.8835335969924927,
+      "learning_rate": 2.8281105968550957e-05,
+      "loss": 0.6851,
+      "step": 475
+    },
+    {
+      "epoch": 0.9937888198757764,
+      "grad_norm": 0.9015234708786011,
+      "learning_rate": 2.8233053140645786e-05,
+      "loss": 0.6547,
+      "step": 480
+    },
+    {
+      "epoch": 1.0041407867494825,
+      "grad_norm": 0.9358381032943726,
+      "learning_rate": 2.818437985631506e-05,
+      "loss": 0.6393,
+      "step": 485
+    },
+    {
+      "epoch": 1.0144927536231885,
+      "grad_norm": 1.0095746517181396,
+      "learning_rate": 2.8135088397696675e-05,
+      "loss": 0.5986,
+      "step": 490
+    },
+    {
+      "epoch": 1.0248447204968945,
+      "grad_norm": 1.0317535400390625,
+      "learning_rate": 2.8085181075912775e-05,
+      "loss": 0.5928,
+      "step": 495
+    },
+    {
+      "epoch": 1.0351966873706004,
+      "grad_norm": 1.0995941162109375,
+      "learning_rate": 2.8034660230961414e-05,
+      "loss": 0.5715,
+      "step": 500
+    },
+    {
+      "epoch": 1.0455486542443064,
+      "grad_norm": 0.952195405960083,
+      "learning_rate": 2.798352823160681e-05,
+      "loss": 0.59,
+      "step": 505
+    },
+    {
+      "epoch": 1.0559006211180124,
+      "grad_norm": 1.0069035291671753,
+      "learning_rate": 2.7931787475268313e-05,
+      "loss": 0.6122,
+      "step": 510
+    },
+    {
+      "epoch": 1.0662525879917184,
+      "grad_norm": 0.9740335941314697,
+      "learning_rate": 2.787944038790797e-05,
+      "loss": 0.5487,
+      "step": 515
+    },
+    {
+      "epoch": 1.0766045548654244,
+      "grad_norm": 1.0057287216186523,
+      "learning_rate": 2.7826489423916787e-05,
+      "loss": 0.5653,
+      "step": 520
+    },
+    {
+      "epoch": 1.0869565217391304,
+      "grad_norm": 0.9624221324920654,
+      "learning_rate": 2.777293706599967e-05,
+      "loss": 0.5737,
+      "step": 525
+    },
+    {
+      "epoch": 1.0973084886128364,
+      "grad_norm": 1.0218902826309204,
+      "learning_rate": 2.7718785825058997e-05,
+      "loss": 0.536,
+      "step": 530
+    },
+    {
+      "epoch": 1.1076604554865424,
+      "grad_norm": 0.9604259133338928,
+      "learning_rate": 2.7664038240076888e-05,
+      "loss": 0.5557,
+      "step": 535
+    },
+    {
+      "epoch": 1.1180124223602483,
+      "grad_norm": 0.9701653122901917,
+      "learning_rate": 2.7608696877996173e-05,
+      "loss": 0.5881,
+      "step": 540
+    },
+    {
+      "epoch": 1.1283643892339545,
+      "grad_norm": 0.954048216342926,
+      "learning_rate": 2.7552764333600036e-05,
+      "loss": 0.4989,
+      "step": 545
+    },
+    {
+      "epoch": 1.1387163561076605,
+      "grad_norm": 1.0160104036331177,
+      "learning_rate": 2.7496243229390346e-05,
+      "loss": 0.5612,
+      "step": 550
+    },
+    {
+      "epoch": 1.1490683229813665,
+      "grad_norm": 0.8945682048797607,
+      "learning_rate": 2.7439136215464692e-05,
+      "loss": 0.5848,
+      "step": 555
+    },
+    {
+      "epoch": 1.1594202898550725,
+      "grad_norm": 0.9766803979873657,
+      "learning_rate": 2.7381445969392137e-05,
+      "loss": 0.5666,
+      "step": 560
+    },
+    {
+      "epoch": 1.1697722567287785,
+      "grad_norm": 0.916334867477417,
+      "learning_rate": 2.7323175196087685e-05,
+      "loss": 0.5701,
+      "step": 565
+    },
+    {
+      "epoch": 1.1801242236024845,
+      "grad_norm": 0.8937172293663025,
+      "learning_rate": 2.726432662768542e-05,
+      "loss": 0.5383,
+      "step": 570
+    },
+    {
+      "epoch": 1.1904761904761905,
+      "grad_norm": 1.0442960262298584,
+      "learning_rate": 2.7204903023410462e-05,
+      "loss": 0.5424,
+      "step": 575
+    },
+    {
+      "epoch": 1.2008281573498965,
+      "grad_norm": 0.9355764389038086,
+      "learning_rate": 2.7144907169449535e-05,
+      "loss": 0.5487,
+      "step": 580
+    },
+    {
+      "epoch": 1.2111801242236024,
+      "grad_norm": 0.9582256078720093,
+      "learning_rate": 2.708434187882037e-05,
+      "loss": 0.5767,
+      "step": 585
+    },
+    {
+      "epoch": 1.2215320910973084,
+      "grad_norm": 1.1042433977127075,
+      "learning_rate": 2.7023209991239792e-05,
+      "loss": 0.516,
+      "step": 590
+    },
+    {
+      "epoch": 1.2318840579710144,
+      "grad_norm": 1.0506311655044556,
+      "learning_rate": 2.69615143729906e-05,
+      "loss": 0.5389,
+      "step": 595
+    },
+    {
+      "epoch": 1.2422360248447206,
+      "grad_norm": 0.9193505644798279,
+      "learning_rate": 2.6899257916787145e-05,
+      "loss": 0.5342,
+      "step": 600
+    },
+    {
+      "epoch": 1.2525879917184266,
+      "grad_norm": 0.9831416606903076,
+      "learning_rate": 2.6836443541639704e-05,
+      "loss": 0.5156,
+      "step": 605
+    },
+    {
+      "epoch": 1.2629399585921326,
+      "grad_norm": 1.0520617961883545,
+      "learning_rate": 2.677307419271766e-05,
+      "loss": 0.5562,
+      "step": 610
+    },
+    {
+      "epoch": 1.2732919254658386,
+      "grad_norm": 0.9378306865692139,
+      "learning_rate": 2.6709152841211348e-05,
+      "loss": 0.4997,
+      "step": 615
+    },
+    {
+      "epoch": 1.2836438923395446,
+      "grad_norm": 1.0318716764450073,
+      "learning_rate": 2.6644682484192784e-05,
+      "loss": 0.4962,
+      "step": 620
+    },
+    {
+      "epoch": 1.2939958592132506,
+      "grad_norm": 0.9622194766998291,
+      "learning_rate": 2.6579666144475136e-05,
+      "loss": 0.525,
+      "step": 625
+    },
+    {
+      "epoch": 1.3043478260869565,
+      "grad_norm": 0.9609589576721191,
+      "learning_rate": 2.651410687047099e-05,
+      "loss": 0.4871,
+      "step": 630
+    },
+    {
+      "epoch": 1.3146997929606625,
+      "grad_norm": 1.0801514387130737,
+      "learning_rate": 2.644800773604942e-05,
+      "loss": 0.51,
+      "step": 635
+    },
+    {
+      "epoch": 1.3250517598343685,
+      "grad_norm": 0.9182636141777039,
+      "learning_rate": 2.6381371840391862e-05,
+      "loss": 0.5397,
+      "step": 640
+    },
+    {
+      "epoch": 1.3354037267080745,
+      "grad_norm": 0.9630659222602844,
+      "learning_rate": 2.6314202307846815e-05,
+      "loss": 0.504,
+      "step": 645
+    },
+    {
+      "epoch": 1.3457556935817805,
+      "grad_norm": 0.9499583840370178,
+      "learning_rate": 2.6246502287783332e-05,
+      "loss": 0.4934,
+      "step": 650
+    },
+    {
+      "epoch": 1.3561076604554865,
+      "grad_norm": 0.9716949462890625,
+      "learning_rate": 2.6178274954443368e-05,
+      "loss": 0.4573,
+      "step": 655
+    },
+    {
+      "epoch": 1.3664596273291925,
+      "grad_norm": 0.9908297657966614,
+      "learning_rate": 2.6109523506792946e-05,
+      "loss": 0.5283,
+      "step": 660
+    },
+    {
+      "epoch": 1.3768115942028984,
+      "grad_norm": 1.0123237371444702,
+      "learning_rate": 2.604025116837216e-05,
+      "loss": 0.5034,
+      "step": 665
+    },
+    {
+      "epoch": 1.3871635610766044,
+      "grad_norm": 0.9543603658676147,
+      "learning_rate": 2.597046118714406e-05,
+      "loss": 0.4865,
+      "step": 670
+    },
+    {
+      "epoch": 1.3975155279503104,
+      "grad_norm": 1.111745834350586,
+      "learning_rate": 2.590015683534232e-05,
+      "loss": 0.5022,
+      "step": 675
+    },
+    {
+      "epoch": 1.4078674948240166,
+      "grad_norm": 1.0601131916046143,
+      "learning_rate": 2.5829341409317866e-05,
+      "loss": 0.4725,
+      "step": 680
+    },
+    {
+      "epoch": 1.4182194616977226,
+      "grad_norm": 0.9767875075340271,
+      "learning_rate": 2.5758018229384283e-05,
+      "loss": 0.4895,
+      "step": 685
+    },
+    {
+      "epoch": 1.4285714285714286,
+      "grad_norm": 1.0154308080673218,
+      "learning_rate": 2.568619063966214e-05,
+      "loss": 0.4662,
+      "step": 690
+    },
+    {
+      "epoch": 1.4389233954451346,
+      "grad_norm": 1.1045129299163818,
+      "learning_rate": 2.5613862007922206e-05,
+      "loss": 0.4932,
+      "step": 695
+    },
+    {
+      "epoch": 1.4492753623188406,
+      "grad_norm": 0.9831157922744751,
+      "learning_rate": 2.554103572542755e-05,
+      "loss": 0.5033,
+      "step": 700
+    },
+    {
+      "epoch": 1.4596273291925466,
+      "grad_norm": 1.0847877264022827,
+      "learning_rate": 2.5467715206774516e-05,
+      "loss": 0.4887,
+      "step": 705
+    },
+    {
+      "epoch": 1.4699792960662525,
+      "grad_norm": 1.0042952299118042,
+      "learning_rate": 2.5393903889732643e-05,
+      "loss": 0.4662,
+      "step": 710
+    },
+    {
+      "epoch": 1.4803312629399585,
+      "grad_norm": 0.9293481707572937,
+      "learning_rate": 2.5319605235083455e-05,
+      "loss": 0.5089,
+      "step": 715
+    },
+    {
+      "epoch": 1.4906832298136645,
+      "grad_norm": 1.011654257774353,
+      "learning_rate": 2.5244822726458227e-05,
+      "loss": 0.5135,
+      "step": 720
+    },
+    {
+      "epoch": 1.5010351966873707,
+      "grad_norm": 1.0469202995300293,
+      "learning_rate": 2.516955987017462e-05,
+      "loss": 0.4516,
+      "step": 725
+    },
+    {
+      "epoch": 1.5113871635610767,
+      "grad_norm": 0.9834375381469727,
+      "learning_rate": 2.5093820195072303e-05,
+      "loss": 0.4691,
+      "step": 730
+    },
+    {
+      "epoch": 1.5217391304347827,
+      "grad_norm": 1.0728363990783691,
+      "learning_rate": 2.501760725234747e-05,
+      "loss": 0.4594,
+      "step": 735
+    },
+    {
+      "epoch": 1.5320910973084887,
+      "grad_norm": 1.0717525482177734,
+      "learning_rate": 2.4940924615386373e-05,
+      "loss": 0.4324,
+      "step": 740
+    },
+    {
+      "epoch": 1.5424430641821947,
+      "grad_norm": 1.1047194004058838,
+      "learning_rate": 2.4863775879597738e-05,
+      "loss": 0.4389,
+      "step": 745
+    },
+    {
+      "epoch": 1.5527950310559007,
+      "grad_norm": 1.0790616273880005,
+      "learning_rate": 2.4786164662244214e-05,
+      "loss": 0.4565,
+      "step": 750
+    },
+    {
+      "epoch": 1.5631469979296067,
+      "grad_norm": 1.113751769065857,
+      "learning_rate": 2.4708094602272774e-05,
+      "loss": 0.4425,
+      "step": 755
+    },
+    {
+      "epoch": 1.5734989648033126,
+      "grad_norm": 1.010530948638916,
+      "learning_rate": 2.462956936014405e-05,
+      "loss": 0.4341,
+      "step": 760
+    },
+    {
+      "epoch": 1.5838509316770186,
+      "grad_norm": 0.9428200125694275,
+      "learning_rate": 2.455059261766078e-05,
+      "loss": 0.4221,
+      "step": 765
+    },
+    {
+      "epoch": 1.5942028985507246,
+      "grad_norm": 1.254734992980957,
+      "learning_rate": 2.447116807779511e-05,
+      "loss": 0.4183,
+      "step": 770
+    },
+    {
+      "epoch": 1.6045548654244306,
+      "grad_norm": 1.0950216054916382,
+      "learning_rate": 2.4391299464515025e-05,
+      "loss": 0.4262,
+      "step": 775
+    },
+    {
+      "epoch": 1.6149068322981366,
+      "grad_norm": 1.0931682586669922,
+      "learning_rate": 2.43109905226097e-05,
+      "loss": 0.4191,
+      "step": 780
+    },
+    {
+      "epoch": 1.6252587991718426,
+      "grad_norm": 1.0379023551940918,
+      "learning_rate": 2.423024501751397e-05,
+      "loss": 0.4404,
+      "step": 785
+    },
+    {
+      "epoch": 1.6356107660455486,
+      "grad_norm": 1.0413020849227905,
+      "learning_rate": 2.4149066735131712e-05,
+      "loss": 0.4415,
+      "step": 790
+    },
+    {
+      "epoch": 1.6459627329192545,
+      "grad_norm": 1.1122803688049316,
+      "learning_rate": 2.4067459481658408e-05,
+      "loss": 0.4452,
+      "step": 795
+    },
+    {
+      "epoch": 1.6563146997929605,
+      "grad_norm": 1.0793243646621704,
+      "learning_rate": 2.3985427083402645e-05,
+      "loss": 0.43,
+      "step": 800
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 1.0770187377929688,
+      "learning_rate": 2.39029733866067e-05,
+      "loss": 0.4015,
+      "step": 805
+    },
+    {
+      "epoch": 1.6770186335403725,
+      "grad_norm": 1.053779125213623,
+      "learning_rate": 2.3820102257266235e-05,
+      "loss": 0.4033,
+      "step": 810
+    },
+    {
+      "epoch": 1.6873706004140787,
+      "grad_norm": 1.1139789819717407,
+      "learning_rate": 2.373681758094901e-05,
+      "loss": 0.4362,
+      "step": 815
+    },
+    {
+      "epoch": 1.6977225672877847,
+      "grad_norm": 1.059994101524353,
+      "learning_rate": 2.3653123262612728e-05,
+      "loss": 0.4499,
+      "step": 820
+    },
+    {
+      "epoch": 1.7080745341614907,
+      "grad_norm": 0.9898128509521484,
+      "learning_rate": 2.3569023226421885e-05,
+      "loss": 0.3766,
+      "step": 825
+    },
+    {
+      "epoch": 1.7184265010351967,
+      "grad_norm": 1.0510621070861816,
+      "learning_rate": 2.348452141556385e-05,
+      "loss": 0.4066,
+      "step": 830
+    },
+    {
+      "epoch": 1.7287784679089027,
+      "grad_norm": 0.9731054902076721,
+      "learning_rate": 2.339962179206393e-05,
+      "loss": 0.4226,
+      "step": 835
+    },
+    {
+      "epoch": 1.7391304347826086,
+      "grad_norm": 0.9354691505432129,
+      "learning_rate": 2.3314328336599636e-05,
+      "loss": 0.3864,
+      "step": 840
+    },
+    {
+      "epoch": 1.7494824016563149,
+      "grad_norm": 1.0401995182037354,
+      "learning_rate": 2.3228645048314014e-05,
+      "loss": 0.3743,
+      "step": 845
+    },
+    {
+      "epoch": 1.7598343685300208,
+      "grad_norm": 0.9203187227249146,
+      "learning_rate": 2.314257594462815e-05,
+      "loss": 0.4019,
+      "step": 850
+    },
+    {
+      "epoch": 1.7701863354037268,
+      "grad_norm": 1.0538512468338013,
+      "learning_rate": 2.3056125061052816e-05,
+      "loss": 0.397,
+      "step": 855
+    },
+    {
+      "epoch": 1.7805383022774328,
+      "grad_norm": 0.9693679213523865,
+      "learning_rate": 2.296929645099924e-05,
+      "loss": 0.3636,
+      "step": 860
+    },
+    {
+      "epoch": 1.7908902691511388,
+      "grad_norm": 0.9815075397491455,
+      "learning_rate": 2.2882094185589065e-05,
+      "loss": 0.3977,
+      "step": 865
+    },
+    {
+      "epoch": 1.8012422360248448,
+      "grad_norm": 0.983045756816864,
+      "learning_rate": 2.2794522353463464e-05,
+      "loss": 0.3456,
+      "step": 870
+    },
+    {
+      "epoch": 1.8115942028985508,
+      "grad_norm": 0.946188747882843,
+      "learning_rate": 2.270658506059143e-05,
+      "loss": 0.3734,
+      "step": 875
+    },
+    {
+      "epoch": 1.8219461697722568,
+      "grad_norm": 1.0121080875396729,
+      "learning_rate": 2.2618286430077277e-05,
+      "loss": 0.3935,
+      "step": 880
+    },
+    {
+      "epoch": 1.8322981366459627,
+      "grad_norm": 1.0990688800811768,
+      "learning_rate": 2.2529630601967302e-05,
+      "loss": 0.3632,
+      "step": 885
+    },
+    {
+      "epoch": 1.8426501035196687,
+      "grad_norm": 1.0193054676055908,
+      "learning_rate": 2.2440621733055673e-05,
+      "loss": 0.4098,
+      "step": 890
+    },
+    {
+      "epoch": 1.8530020703933747,
+      "grad_norm": 1.0463533401489258,
+      "learning_rate": 2.235126399668955e-05,
+      "loss": 0.3409,
+      "step": 895
+    },
+    {
+      "epoch": 1.8633540372670807,
+      "grad_norm": 1.0082764625549316,
+      "learning_rate": 2.2261561582573388e-05,
+      "loss": 0.378,
+      "step": 900
+    },
+    {
+      "epoch": 1.8737060041407867,
+      "grad_norm": 0.9819716215133667,
+      "learning_rate": 2.2171518696572498e-05,
+      "loss": 0.3971,
+      "step": 905
+    },
+    {
+      "epoch": 1.8840579710144927,
+      "grad_norm": 1.0094093084335327,
+      "learning_rate": 2.208113956051586e-05,
+      "loss": 0.3475,
+      "step": 910
+    },
+    {
+      "epoch": 1.8944099378881987,
+      "grad_norm": 1.1090017557144165,
+      "learning_rate": 2.1990428411998162e-05,
+      "loss": 0.3882,
+      "step": 915
+    },
+    {
+      "epoch": 1.9047619047619047,
+      "grad_norm": 1.0030994415283203,
+      "learning_rate": 2.1899389504181114e-05,
+      "loss": 0.3759,
+      "step": 920
+    },
+    {
+      "epoch": 1.9151138716356106,
+      "grad_norm": 1.071929931640625,
+      "learning_rate": 2.1808027105594047e-05,
+      "loss": 0.3477,
+      "step": 925
+    },
+    {
+      "epoch": 1.9254658385093166,
+      "grad_norm": 1.0278531312942505,
+      "learning_rate": 2.171634549993374e-05,
+      "loss": 0.332,
+      "step": 930
+    },
+    {
+      "epoch": 1.9358178053830226,
+      "grad_norm": 1.0566397905349731,
+      "learning_rate": 2.1624348985863606e-05,
+      "loss": 0.3904,
+      "step": 935
+    },
+    {
+      "epoch": 1.9461697722567288,
+      "grad_norm": 1.1759693622589111,
+      "learning_rate": 2.153204187681212e-05,
+      "loss": 0.3598,
+      "step": 940
+    },
+    {
+      "epoch": 1.9565217391304348,
+      "grad_norm": 1.1160260438919067,
+      "learning_rate": 2.1439428500770598e-05,
+      "loss": 0.3677,
+      "step": 945
+    },
+    {
+      "epoch": 1.9668737060041408,
+      "grad_norm": 1.0783803462982178,
+      "learning_rate": 2.1346513200090238e-05,
+      "loss": 0.3504,
+      "step": 950
+    },
+    {
+      "epoch": 1.9772256728778468,
+      "grad_norm": 1.0974122285842896,
+      "learning_rate": 2.1253300331278545e-05,
+      "loss": 0.3351,
+      "step": 955
+    },
+    {
+      "epoch": 1.9875776397515528,
+      "grad_norm": 1.0041438341140747,
+      "learning_rate": 2.115979426479507e-05,
+      "loss": 0.3765,
+      "step": 960
+    },
+    {
+      "epoch": 1.9979296066252588,
+      "grad_norm": 0.962396502494812,
+      "learning_rate": 2.106599938484647e-05,
+      "loss": 0.3745,
+      "step": 965
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 2415,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 2000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.4734182517980529e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

38/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6fcb4517126fe5c447e98514339e7616f73a6c81c96eafaa468ba0f5a5bbc9a5
+size 7736

38/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)