Upload folder using huggingface_hub

29/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

29/adapter_config.json

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

29/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:21f7e9c0f2b59071c13195d2740e312441b84a2fe3ae5d11ec567bb2061eee36
+size 791751704

29/latest

@@ -0,0 +1 @@
+global_step936

29/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:965a55e50454ee4183ad2be9415177e8a3e4d8bb13959340f8fe14f160dee0cc
+size 15920

29/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d8814d0433f0ed420268c375a6ef7d9129dace1ea59e4bc4a9333d314758c3df
+size 15920

29/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0d5098f95c7c8ae4fb4502dc8c3f9d96c7808463e834f3ac59ce5ca18145064a
+size 15920

29/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:29fcf1328ad0c3c88818dd32924c0aceb331e7fa58e211d41e09ada7d604aa8d
+size 15920

29/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:289c50ddba8cc6276de048151f6c4c70928ec0709222a5c40f4511a874575bf5
+size 15920

29/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9d3e187e09ff981860a7d3bcbbc6dae73c0befe9b74573bd51a72804f862c25b
+size 15920

29/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:58148a2a70e5d4b6d4218117c65c565a38f43aa4fc9095f532fd0dce458061a9
+size 15920

29/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bb62f3a45d59dccebf2d6efab2f0bf5ce4a0874220212c4b4af50b9c03b2351a
+size 15920

29/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:563192fc33b9b362099011b7223d39ac950a6514131108814e53dfafdc01ed2e
+size 1064

29/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
+  }
+}

29/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
+}

29/trainer_state.json

@@ -0,0 +1,1343 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 936,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.010683760683760684,
+      "grad_norm": 1.1713567972183228,
+      "learning_rate": 1.0256410256410257e-06,
+      "loss": 1.3234,
+      "step": 5
+    },
+    {
+      "epoch": 0.021367521367521368,
+      "grad_norm": 1.1887260675430298,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 1.2851,
+      "step": 10
+    },
+    {
+      "epoch": 0.03205128205128205,
+      "grad_norm": 0.7510101199150085,
+      "learning_rate": 3.58974358974359e-06,
+      "loss": 1.2512,
+      "step": 15
+    },
+    {
+      "epoch": 0.042735042735042736,
+      "grad_norm": 0.6793277263641357,
+      "learning_rate": 4.871794871794872e-06,
+      "loss": 1.2919,
+      "step": 20
+    },
+    {
+      "epoch": 0.053418803418803416,
+      "grad_norm": 0.5821245908737183,
+      "learning_rate": 6.153846153846154e-06,
+      "loss": 1.247,
+      "step": 25
+    },
+    {
+      "epoch": 0.0641025641025641,
+      "grad_norm": 0.5563486814498901,
+      "learning_rate": 7.435897435897436e-06,
+      "loss": 1.2902,
+      "step": 30
+    },
+    {
+      "epoch": 0.07478632478632478,
+      "grad_norm": 0.4339327812194824,
+      "learning_rate": 8.717948717948719e-06,
+      "loss": 1.2194,
+      "step": 35
+    },
+    {
+      "epoch": 0.08547008547008547,
+      "grad_norm": 0.5005843639373779,
+      "learning_rate": 9.999999999999999e-06,
+      "loss": 1.2463,
+      "step": 40
+    },
+    {
+      "epoch": 0.09615384615384616,
+      "grad_norm": 0.4932032823562622,
+      "learning_rate": 1.1282051282051283e-05,
+      "loss": 1.2011,
+      "step": 45
+    },
+    {
+      "epoch": 0.10683760683760683,
+      "grad_norm": 0.4715913236141205,
+      "learning_rate": 1.2564102564102565e-05,
+      "loss": 1.1482,
+      "step": 50
+    },
+    {
+      "epoch": 0.11752136752136752,
+      "grad_norm": 0.7068427205085754,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 1.1867,
+      "step": 55
+    },
+    {
+      "epoch": 0.1282051282051282,
+      "grad_norm": 0.5336880683898926,
+      "learning_rate": 1.5128205128205129e-05,
+      "loss": 1.1946,
+      "step": 60
+    },
+    {
+      "epoch": 0.1388888888888889,
+      "grad_norm": 0.42017748951911926,
+      "learning_rate": 1.641025641025641e-05,
+      "loss": 1.1679,
+      "step": 65
+    },
+    {
+      "epoch": 0.14957264957264957,
+      "grad_norm": 0.479497492313385,
+      "learning_rate": 1.7692307692307694e-05,
+      "loss": 1.2129,
+      "step": 70
+    },
+    {
+      "epoch": 0.16025641025641027,
+      "grad_norm": 0.49006474018096924,
+      "learning_rate": 1.8974358974358975e-05,
+      "loss": 1.1219,
+      "step": 75
+    },
+    {
+      "epoch": 0.17094017094017094,
+      "grad_norm": 0.41606658697128296,
+      "learning_rate": 2.025641025641026e-05,
+      "loss": 1.167,
+      "step": 80
+    },
+    {
+      "epoch": 0.18162393162393162,
+      "grad_norm": 0.4794115424156189,
+      "learning_rate": 2.153846153846154e-05,
+      "loss": 1.1411,
+      "step": 85
+    },
+    {
+      "epoch": 0.19230769230769232,
+      "grad_norm": 0.4326317608356476,
+      "learning_rate": 2.282051282051282e-05,
+      "loss": 1.1313,
+      "step": 90
+    },
+    {
+      "epoch": 0.202991452991453,
+      "grad_norm": 0.491228848695755,
+      "learning_rate": 2.4102564102564103e-05,
+      "loss": 1.099,
+      "step": 95
+    },
+    {
+      "epoch": 0.21367521367521367,
+      "grad_norm": 0.4923720061779022,
+      "learning_rate": 2.5384615384615386e-05,
+      "loss": 1.1109,
+      "step": 100
+    },
+    {
+      "epoch": 0.22435897435897437,
+      "grad_norm": 0.4896738827228546,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 1.1149,
+      "step": 105
+    },
+    {
+      "epoch": 0.23504273504273504,
+      "grad_norm": 0.48596683144569397,
+      "learning_rate": 2.794871794871795e-05,
+      "loss": 1.1151,
+      "step": 110
+    },
+    {
+      "epoch": 0.24572649572649571,
+      "grad_norm": 0.6104574799537659,
+      "learning_rate": 2.923076923076923e-05,
+      "loss": 1.1068,
+      "step": 115
+    },
+    {
+      "epoch": 0.2564102564102564,
+      "grad_norm": 0.5183061361312866,
+      "learning_rate": 2.999994008414162e-05,
+      "loss": 1.0852,
+      "step": 120
+    },
+    {
+      "epoch": 0.2670940170940171,
+      "grad_norm": 0.5682564377784729,
+      "learning_rate": 2.999926603623188e-05,
+      "loss": 1.0771,
+      "step": 125
+    },
+    {
+      "epoch": 0.2777777777777778,
+      "grad_norm": 0.5799983739852905,
+      "learning_rate": 2.999784307935663e-05,
+      "loss": 1.0574,
+      "step": 130
+    },
+    {
+      "epoch": 0.28846153846153844,
+      "grad_norm": 0.5900704264640808,
+      "learning_rate": 2.999567128456368e-05,
+      "loss": 1.0898,
+      "step": 135
+    },
+    {
+      "epoch": 0.29914529914529914,
+      "grad_norm": 0.6018744111061096,
+      "learning_rate": 2.999275076029011e-05,
+      "loss": 1.0212,
+      "step": 140
+    },
+    {
+      "epoch": 0.30982905982905984,
+      "grad_norm": 0.5195294618606567,
+      "learning_rate": 2.9989081652356853e-05,
+      "loss": 1.0384,
+      "step": 145
+    },
+    {
+      "epoch": 0.32051282051282054,
+      "grad_norm": 0.560064971446991,
+      "learning_rate": 2.998466414396139e-05,
+      "loss": 1.0092,
+      "step": 150
+    },
+    {
+      "epoch": 0.3311965811965812,
+      "grad_norm": 0.6255720257759094,
+      "learning_rate": 2.9979498455668615e-05,
+      "loss": 0.9929,
+      "step": 155
+    },
+    {
+      "epoch": 0.3418803418803419,
+      "grad_norm": 0.6955039501190186,
+      "learning_rate": 2.9973584845399847e-05,
+      "loss": 1.006,
+      "step": 160
+    },
+    {
+      "epoch": 0.3525641025641026,
+      "grad_norm": 0.6638875603675842,
+      "learning_rate": 2.9966923608419906e-05,
+      "loss": 1.0245,
+      "step": 165
+    },
+    {
+      "epoch": 0.36324786324786323,
+      "grad_norm": 0.7732335925102234,
+      "learning_rate": 2.9959515077322413e-05,
+      "loss": 1.0425,
+      "step": 170
+    },
+    {
+      "epoch": 0.37393162393162394,
+      "grad_norm": 0.6519403457641602,
+      "learning_rate": 2.995135962201315e-05,
+      "loss": 1.0148,
+      "step": 175
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 0.6945505738258362,
+      "learning_rate": 2.9942457649691624e-05,
+      "loss": 1.0235,
+      "step": 180
+    },
+    {
+      "epoch": 0.3952991452991453,
+      "grad_norm": 0.6265760660171509,
+      "learning_rate": 2.9932809604830703e-05,
+      "loss": 0.9227,
+      "step": 185
+    },
+    {
+      "epoch": 0.405982905982906,
+      "grad_norm": 0.6135931611061096,
+      "learning_rate": 2.9922415969154445e-05,
+      "loss": 1.0083,
+      "step": 190
+    },
+    {
+      "epoch": 0.4166666666666667,
+      "grad_norm": 0.5732436180114746,
+      "learning_rate": 2.9911277261614044e-05,
+      "loss": 1.0089,
+      "step": 195
+    },
+    {
+      "epoch": 0.42735042735042733,
+      "grad_norm": 0.730195939540863,
+      "learning_rate": 2.9899394038361906e-05,
+      "loss": 0.9937,
+      "step": 200
+    },
+    {
+      "epoch": 0.43803418803418803,
+      "grad_norm": 0.7022405862808228,
+      "learning_rate": 2.9886766892723887e-05,
+      "loss": 0.986,
+      "step": 205
+    },
+    {
+      "epoch": 0.44871794871794873,
+      "grad_norm": 0.5751799941062927,
+      "learning_rate": 2.987339645516968e-05,
+      "loss": 0.9643,
+      "step": 210
+    },
+    {
+      "epoch": 0.4594017094017094,
+      "grad_norm": 0.7066762447357178,
+      "learning_rate": 2.9859283393281325e-05,
+      "loss": 0.9148,
+      "step": 215
+    },
+    {
+      "epoch": 0.4700854700854701,
+      "grad_norm": 0.6306741833686829,
+      "learning_rate": 2.9844428411719864e-05,
+      "loss": 0.9486,
+      "step": 220
+    },
+    {
+      "epoch": 0.4807692307692308,
+      "grad_norm": 0.7719146609306335,
+      "learning_rate": 2.982883225219019e-05,
+      "loss": 0.8962,
+      "step": 225
+    },
+    {
+      "epoch": 0.49145299145299143,
+      "grad_norm": 0.6854239106178284,
+      "learning_rate": 2.981249569340398e-05,
+      "loss": 0.9415,
+      "step": 230
+    },
+    {
+      "epoch": 0.5021367521367521,
+      "grad_norm": 0.6685622334480286,
+      "learning_rate": 2.9795419551040836e-05,
+      "loss": 0.894,
+      "step": 235
+    },
+    {
+      "epoch": 0.5128205128205128,
+      "grad_norm": 0.6961314082145691,
+      "learning_rate": 2.9777604677707553e-05,
+      "loss": 0.9099,
+      "step": 240
+    },
+    {
+      "epoch": 0.5235042735042735,
+      "grad_norm": 0.705095112323761,
+      "learning_rate": 2.975905196289555e-05,
+      "loss": 0.9117,
+      "step": 245
+    },
+    {
+      "epoch": 0.5341880341880342,
+      "grad_norm": 0.7737231850624084,
+      "learning_rate": 2.9739762332936433e-05,
+      "loss": 0.8638,
+      "step": 250
+    },
+    {
+      "epoch": 0.5448717948717948,
+      "grad_norm": 0.9252617359161377,
+      "learning_rate": 2.97197367509558e-05,
+      "loss": 0.9247,
+      "step": 255
+    },
+    {
+      "epoch": 0.5555555555555556,
+      "grad_norm": 0.7449509501457214,
+      "learning_rate": 2.96989762168251e-05,
+      "loss": 0.8522,
+      "step": 260
+    },
+    {
+      "epoch": 0.5662393162393162,
+      "grad_norm": 0.7612582445144653,
+      "learning_rate": 2.967748176711172e-05,
+      "loss": 0.912,
+      "step": 265
+    },
+    {
+      "epoch": 0.5769230769230769,
+      "grad_norm": 0.7640935182571411,
+      "learning_rate": 2.9655254475027256e-05,
+      "loss": 0.8693,
+      "step": 270
+    },
+    {
+      "epoch": 0.5876068376068376,
+      "grad_norm": 0.9326266646385193,
+      "learning_rate": 2.9632295450373893e-05,
+      "loss": 0.9266,
+      "step": 275
+    },
+    {
+      "epoch": 0.5982905982905983,
+      "grad_norm": 0.9562191963195801,
+      "learning_rate": 2.960860583948902e-05,
+      "loss": 0.8783,
+      "step": 280
+    },
+    {
+      "epoch": 0.6089743589743589,
+      "grad_norm": 0.8497742414474487,
+      "learning_rate": 2.958418682518798e-05,
+      "loss": 0.8753,
+      "step": 285
+    },
+    {
+      "epoch": 0.6196581196581197,
+      "grad_norm": 0.7647716403007507,
+      "learning_rate": 2.9559039626705006e-05,
+      "loss": 0.8345,
+      "step": 290
+    },
+    {
+      "epoch": 0.6303418803418803,
+      "grad_norm": 0.7795203924179077,
+      "learning_rate": 2.9533165499632367e-05,
+      "loss": 0.9003,
+      "step": 295
+    },
+    {
+      "epoch": 0.6410256410256411,
+      "grad_norm": 0.8634848594665527,
+      "learning_rate": 2.950656573585766e-05,
+      "loss": 0.7992,
+      "step": 300
+    },
+    {
+      "epoch": 0.6517094017094017,
+      "grad_norm": 0.8322213292121887,
+      "learning_rate": 2.947924166349931e-05,
+      "loss": 0.8705,
+      "step": 305
+    },
+    {
+      "epoch": 0.6623931623931624,
+      "grad_norm": 0.794114351272583,
+      "learning_rate": 2.9451194646840255e-05,
+      "loss": 0.8142,
+      "step": 310
+    },
+    {
+      "epoch": 0.6730769230769231,
+      "grad_norm": 0.8197933435440063,
+      "learning_rate": 2.942242608625983e-05,
+      "loss": 0.7872,
+      "step": 315
+    },
+    {
+      "epoch": 0.6837606837606838,
+      "grad_norm": 0.8497008681297302,
+      "learning_rate": 2.939293741816385e-05,
+      "loss": 0.8132,
+      "step": 320
+    },
+    {
+      "epoch": 0.6944444444444444,
+      "grad_norm": 0.9021241068840027,
+      "learning_rate": 2.9362730114912892e-05,
+      "loss": 0.8512,
+      "step": 325
+    },
+    {
+      "epoch": 0.7051282051282052,
+      "grad_norm": 0.8258333206176758,
+      "learning_rate": 2.9331805684748766e-05,
+      "loss": 0.7375,
+      "step": 330
+    },
+    {
+      "epoch": 0.7158119658119658,
+      "grad_norm": 0.8060581684112549,
+      "learning_rate": 2.930016567171922e-05,
+      "loss": 0.7768,
+      "step": 335
+    },
+    {
+      "epoch": 0.7264957264957265,
+      "grad_norm": 0.8077674508094788,
+      "learning_rate": 2.926781165560086e-05,
+      "loss": 0.8057,
+      "step": 340
+    },
+    {
+      "epoch": 0.7371794871794872,
+      "grad_norm": 0.7880494594573975,
+      "learning_rate": 2.9234745251820242e-05,
+      "loss": 0.8406,
+      "step": 345
+    },
+    {
+      "epoch": 0.7478632478632479,
+      "grad_norm": 0.8711897134780884,
+      "learning_rate": 2.920096811137323e-05,
+      "loss": 0.8124,
+      "step": 350
+    },
+    {
+      "epoch": 0.7585470085470085,
+      "grad_norm": 0.8994575142860413,
+      "learning_rate": 2.9166481920742565e-05,
+      "loss": 0.76,
+      "step": 355
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.869128942489624,
+      "learning_rate": 2.9131288401813664e-05,
+      "loss": 0.7901,
+      "step": 360
+    },
+    {
+      "epoch": 0.7799145299145299,
+      "grad_norm": 0.9975675940513611,
+      "learning_rate": 2.9095389311788626e-05,
+      "loss": 0.7521,
+      "step": 365
+    },
+    {
+      "epoch": 0.7905982905982906,
+      "grad_norm": 0.9022954106330872,
+      "learning_rate": 2.905878644309852e-05,
+      "loss": 0.7632,
+      "step": 370
+    },
+    {
+      "epoch": 0.8012820512820513,
+      "grad_norm": 0.7905294299125671,
+      "learning_rate": 2.9021481623313864e-05,
+      "loss": 0.7631,
+      "step": 375
+    },
+    {
+      "epoch": 0.811965811965812,
+      "grad_norm": 0.7954912781715393,
+      "learning_rate": 2.8983476715053412e-05,
+      "loss": 0.7703,
+      "step": 380
+    },
+    {
+      "epoch": 0.8226495726495726,
+      "grad_norm": 0.9469508528709412,
+      "learning_rate": 2.8944773615891097e-05,
+      "loss": 0.7086,
+      "step": 385
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 0.8869741559028625,
+      "learning_rate": 2.8905374258261357e-05,
+      "loss": 0.7342,
+      "step": 390
+    },
+    {
+      "epoch": 0.844017094017094,
+      "grad_norm": 0.9316343665122986,
+      "learning_rate": 2.8865280609362578e-05,
+      "loss": 0.7139,
+      "step": 395
+    },
+    {
+      "epoch": 0.8547008547008547,
+      "grad_norm": 0.9334567189216614,
+      "learning_rate": 2.8824494671058927e-05,
+      "loss": 0.7348,
+      "step": 400
+    },
+    {
+      "epoch": 0.8653846153846154,
+      "grad_norm": 0.8888137936592102,
+      "learning_rate": 2.8783018479780378e-05,
+      "loss": 0.7234,
+      "step": 405
+    },
+    {
+      "epoch": 0.8760683760683761,
+      "grad_norm": 0.8082792162895203,
+      "learning_rate": 2.874085410642102e-05,
+      "loss": 0.7877,
+      "step": 410
+    },
+    {
+      "epoch": 0.8867521367521367,
+      "grad_norm": 0.9728290438652039,
+      "learning_rate": 2.8698003656235686e-05,
+      "loss": 0.7513,
+      "step": 415
+    },
+    {
+      "epoch": 0.8974358974358975,
+      "grad_norm": 0.915022611618042,
+      "learning_rate": 2.8654469268734813e-05,
+      "loss": 0.7276,
+      "step": 420
+    },
+    {
+      "epoch": 0.9081196581196581,
+      "grad_norm": 0.876768171787262,
+      "learning_rate": 2.8610253117577644e-05,
+      "loss": 0.6979,
+      "step": 425
+    },
+    {
+      "epoch": 0.9188034188034188,
+      "grad_norm": 1.037745475769043,
+      "learning_rate": 2.8565357410463664e-05,
+      "loss": 0.7515,
+      "step": 430
+    },
+    {
+      "epoch": 0.9294871794871795,
+      "grad_norm": 0.8949560523033142,
+      "learning_rate": 2.85197843890224e-05,
+      "loss": 0.6866,
+      "step": 435
+    },
+    {
+      "epoch": 0.9401709401709402,
+      "grad_norm": 0.9212034940719604,
+      "learning_rate": 2.8473536328701472e-05,
+      "loss": 0.6836,
+      "step": 440
+    },
+    {
+      "epoch": 0.9508547008547008,
+      "grad_norm": 0.9272440671920776,
+      "learning_rate": 2.8426615538653016e-05,
+      "loss": 0.6802,
+      "step": 445
+    },
+    {
+      "epoch": 0.9615384615384616,
+      "grad_norm": 0.906293511390686,
+      "learning_rate": 2.8379024361618348e-05,
+      "loss": 0.6654,
+      "step": 450
+    },
+    {
+      "epoch": 0.9722222222222222,
+      "grad_norm": 0.8837043642997742,
+      "learning_rate": 2.8330765173811016e-05,
+      "loss": 0.6963,
+      "step": 455
+    },
+    {
+      "epoch": 0.9829059829059829,
+      "grad_norm": 0.8492494821548462,
+      "learning_rate": 2.8281840384798147e-05,
+      "loss": 0.6697,
+      "step": 460
+    },
+    {
+      "epoch": 0.9935897435897436,
+      "grad_norm": 0.8252466917037964,
+      "learning_rate": 2.8232252437380153e-05,
+      "loss": 0.6546,
+      "step": 465
+    },
+    {
+      "epoch": 1.0042735042735043,
+      "grad_norm": 0.8824915289878845,
+      "learning_rate": 2.8182003807468733e-05,
+      "loss": 0.6714,
+      "step": 470
+    },
+    {
+      "epoch": 1.014957264957265,
+      "grad_norm": 0.9439343214035034,
+      "learning_rate": 2.8131097003963285e-05,
+      "loss": 0.6008,
+      "step": 475
+    },
+    {
+      "epoch": 1.0256410256410255,
+      "grad_norm": 1.0198942422866821,
+      "learning_rate": 2.8079534568625615e-05,
+      "loss": 0.6261,
+      "step": 480
+    },
+    {
+      "epoch": 1.0363247863247864,
+      "grad_norm": 1.01052987575531,
+      "learning_rate": 2.8027319075953038e-05,
+      "loss": 0.5412,
+      "step": 485
+    },
+    {
+      "epoch": 1.047008547008547,
+      "grad_norm": 0.9888526201248169,
+      "learning_rate": 2.7974453133049824e-05,
+      "loss": 0.6405,
+      "step": 490
+    },
+    {
+      "epoch": 1.0576923076923077,
+      "grad_norm": 0.982110857963562,
+      "learning_rate": 2.792093937949704e-05,
+      "loss": 0.5479,
+      "step": 495
+    },
+    {
+      "epoch": 1.0683760683760684,
+      "grad_norm": 0.9982971549034119,
+      "learning_rate": 2.786678048722075e-05,
+      "loss": 0.5939,
+      "step": 500
+    },
+    {
+      "epoch": 1.079059829059829,
+      "grad_norm": 1.0138237476348877,
+      "learning_rate": 2.78119791603586e-05,
+      "loss": 0.5611,
+      "step": 505
+    },
+    {
+      "epoch": 1.0897435897435896,
+      "grad_norm": 0.969343364238739,
+      "learning_rate": 2.775653813512481e-05,
+      "loss": 0.5585,
+      "step": 510
+    },
+    {
+      "epoch": 1.1004273504273505,
+      "grad_norm": 1.051679253578186,
+      "learning_rate": 2.7700460179673548e-05,
+      "loss": 0.5654,
+      "step": 515
+    },
+    {
+      "epoch": 1.1111111111111112,
+      "grad_norm": 1.1189768314361572,
+      "learning_rate": 2.764374809396073e-05,
+      "loss": 0.5373,
+      "step": 520
+    },
+    {
+      "epoch": 1.1217948717948718,
+      "grad_norm": 1.017276644706726,
+      "learning_rate": 2.75864047096042e-05,
+      "loss": 0.5593,
+      "step": 525
+    },
+    {
+      "epoch": 1.1324786324786325,
+      "grad_norm": 0.9930076003074646,
+      "learning_rate": 2.7528432889742378e-05,
+      "loss": 0.5386,
+      "step": 530
+    },
+    {
+      "epoch": 1.143162393162393,
+      "grad_norm": 0.9654243588447571,
+      "learning_rate": 2.7469835528891257e-05,
+      "loss": 0.5643,
+      "step": 535
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 1.0912607908248901,
+      "learning_rate": 2.741061555279993e-05,
+      "loss": 0.5672,
+      "step": 540
+    },
+    {
+      "epoch": 1.1645299145299146,
+      "grad_norm": 0.9735100269317627,
+      "learning_rate": 2.735077591830448e-05,
+      "loss": 0.5565,
+      "step": 545
+    },
+    {
+      "epoch": 1.1752136752136753,
+      "grad_norm": 1.0461339950561523,
+      "learning_rate": 2.7290319613180348e-05,
+      "loss": 0.5727,
+      "step": 550
+    },
+    {
+      "epoch": 1.185897435897436,
+      "grad_norm": 1.0950112342834473,
+      "learning_rate": 2.7229249655993164e-05,
+      "loss": 0.5453,
+      "step": 555
+    },
+    {
+      "epoch": 1.1965811965811965,
+      "grad_norm": 1.028745174407959,
+      "learning_rate": 2.7167569095948032e-05,
+      "loss": 0.5749,
+      "step": 560
+    },
+    {
+      "epoch": 1.2072649572649572,
+      "grad_norm": 1.0277397632598877,
+      "learning_rate": 2.7105281012737272e-05,
+      "loss": 0.5933,
+      "step": 565
+    },
+    {
+      "epoch": 1.217948717948718,
+      "grad_norm": 0.9753595590591431,
+      "learning_rate": 2.7042388516386648e-05,
+      "loss": 0.5223,
+      "step": 570
+    },
+    {
+      "epoch": 1.2286324786324787,
+      "grad_norm": 0.9498695135116577,
+      "learning_rate": 2.697889474710012e-05,
+      "loss": 0.5452,
+      "step": 575
+    },
+    {
+      "epoch": 1.2393162393162394,
+      "grad_norm": 1.110336422920227,
+      "learning_rate": 2.6914802875103003e-05,
+      "loss": 0.5681,
+      "step": 580
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.9814836382865906,
+      "learning_rate": 2.6850116100483714e-05,
+      "loss": 0.5497,
+      "step": 585
+    },
+    {
+      "epoch": 1.2606837606837606,
+      "grad_norm": 1.027953028678894,
+      "learning_rate": 2.6784837653034e-05,
+      "loss": 0.5137,
+      "step": 590
+    },
+    {
+      "epoch": 1.2713675213675213,
+      "grad_norm": 0.9982088804244995,
+      "learning_rate": 2.6718970792087642e-05,
+      "loss": 0.5624,
+      "step": 595
+    },
+    {
+      "epoch": 1.282051282051282,
+      "grad_norm": 1.4840160608291626,
+      "learning_rate": 2.6652518806357733e-05,
+      "loss": 0.4867,
+      "step": 600
+    },
+    {
+      "epoch": 1.2927350427350428,
+      "grad_norm": 1.0056880712509155,
+      "learning_rate": 2.658548501377249e-05,
+      "loss": 0.5635,
+      "step": 605
+    },
+    {
+      "epoch": 1.3034188034188035,
+      "grad_norm": 0.9386727809906006,
+      "learning_rate": 2.651787276130956e-05,
+      "loss": 0.5026,
+      "step": 610
+    },
+    {
+      "epoch": 1.314102564102564,
+      "grad_norm": 1.0231010913848877,
+      "learning_rate": 2.6449685424828935e-05,
+      "loss": 0.5213,
+      "step": 615
+    },
+    {
+      "epoch": 1.3247863247863247,
+      "grad_norm": 1.1094346046447754,
+      "learning_rate": 2.638092640890438e-05,
+      "loss": 0.4772,
+      "step": 620
+    },
+    {
+      "epoch": 1.3354700854700854,
+      "grad_norm": 1.0024440288543701,
+      "learning_rate": 2.6311599146653446e-05,
+      "loss": 0.4642,
+      "step": 625
+    },
+    {
+      "epoch": 1.3461538461538463,
+      "grad_norm": 0.9997662901878357,
+      "learning_rate": 2.6241707099566052e-05,
+      "loss": 0.4885,
+      "step": 630
+    },
+    {
+      "epoch": 1.356837606837607,
+      "grad_norm": 1.0848814249038696,
+      "learning_rate": 2.617125375733167e-05,
+      "loss": 0.4751,
+      "step": 635
+    },
+    {
+      "epoch": 1.3675213675213675,
+      "grad_norm": 0.9730331301689148,
+      "learning_rate": 2.6100242637665064e-05,
+      "loss": 0.4822,
+      "step": 640
+    },
+    {
+      "epoch": 1.3782051282051282,
+      "grad_norm": 1.269994854927063,
+      "learning_rate": 2.6028677286130664e-05,
+      "loss": 0.5376,
+      "step": 645
+    },
+    {
+      "epoch": 1.3888888888888888,
+      "grad_norm": 1.0568530559539795,
+      "learning_rate": 2.5956561275965548e-05,
+      "loss": 0.5015,
+      "step": 650
+    },
+    {
+      "epoch": 1.3995726495726495,
+      "grad_norm": 1.109546422958374,
+      "learning_rate": 2.5883898207900997e-05,
+      "loss": 0.4934,
+      "step": 655
+    },
+    {
+      "epoch": 1.4102564102564101,
+      "grad_norm": 1.1545758247375488,
+      "learning_rate": 2.5810691709982747e-05,
+      "loss": 0.468,
+      "step": 660
+    },
+    {
+      "epoch": 1.420940170940171,
+      "grad_norm": 1.0068634748458862,
+      "learning_rate": 2.573694543738983e-05,
+      "loss": 0.4859,
+      "step": 665
+    },
+    {
+      "epoch": 1.4316239316239316,
+      "grad_norm": 1.120348572731018,
+      "learning_rate": 2.5662663072252056e-05,
+      "loss": 0.4657,
+      "step": 670
+    },
+    {
+      "epoch": 1.4423076923076923,
+      "grad_norm": 1.042707085609436,
+      "learning_rate": 2.558784832346619e-05,
+      "loss": 0.4592,
+      "step": 675
+    },
+    {
+      "epoch": 1.452991452991453,
+      "grad_norm": 1.0495193004608154,
+      "learning_rate": 2.5512504926510754e-05,
+      "loss": 0.4912,
+      "step": 680
+    },
+    {
+      "epoch": 1.4636752136752138,
+      "grad_norm": 1.0407277345657349,
+      "learning_rate": 2.5436636643259515e-05,
+      "loss": 0.4824,
+      "step": 685
+    },
+    {
+      "epoch": 1.4743589743589745,
+      "grad_norm": 0.9787584543228149,
+      "learning_rate": 2.5360247261793666e-05,
+      "loss": 0.4947,
+      "step": 690
+    },
+    {
+      "epoch": 1.485042735042735,
+      "grad_norm": 0.9804218411445618,
+      "learning_rate": 2.5283340596212677e-05,
+      "loss": 0.4647,
+      "step": 695
+    },
+    {
+      "epoch": 1.4957264957264957,
+      "grad_norm": 1.1483441591262817,
+      "learning_rate": 2.5205920486443863e-05,
+      "loss": 0.4774,
+      "step": 700
+    },
+    {
+      "epoch": 1.5064102564102564,
+      "grad_norm": 1.0858572721481323,
+      "learning_rate": 2.5127990798050658e-05,
+      "loss": 0.4508,
+      "step": 705
+    },
+    {
+      "epoch": 1.517094017094017,
+      "grad_norm": 1.031877040863037,
+      "learning_rate": 2.5049555422039602e-05,
+      "loss": 0.438,
+      "step": 710
+    },
+    {
+      "epoch": 1.5277777777777777,
+      "grad_norm": 1.0131568908691406,
+      "learning_rate": 2.4970618274666087e-05,
+      "loss": 0.4563,
+      "step": 715
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 1.027461290359497,
+      "learning_rate": 2.4891183297238776e-05,
+      "loss": 0.4397,
+      "step": 720
+    },
+    {
+      "epoch": 1.5491452991452992,
+      "grad_norm": 0.9542386531829834,
+      "learning_rate": 2.4811254455922858e-05,
+      "loss": 0.4614,
+      "step": 725
+    },
+    {
+      "epoch": 1.5598290598290598,
+      "grad_norm": 1.0060298442840576,
+      "learning_rate": 2.4730835741542007e-05,
+      "loss": 0.4567,
+      "step": 730
+    },
+    {
+      "epoch": 1.5705128205128205,
+      "grad_norm": 0.9602590203285217,
+      "learning_rate": 2.4649931169379102e-05,
+      "loss": 0.4182,
+      "step": 735
+    },
+    {
+      "epoch": 1.5811965811965814,
+      "grad_norm": 0.9110608696937561,
+      "learning_rate": 2.4568544778975765e-05,
+      "loss": 0.4523,
+      "step": 740
+    },
+    {
+      "epoch": 1.591880341880342,
+      "grad_norm": 1.1888507604599,
+      "learning_rate": 2.4486680633930658e-05,
+      "loss": 0.4128,
+      "step": 745
+    },
+    {
+      "epoch": 1.6025641025641026,
+      "grad_norm": 1.1152507066726685,
+      "learning_rate": 2.44043428216966e-05,
+      "loss": 0.4708,
+      "step": 750
+    },
+    {
+      "epoch": 1.6132478632478633,
+      "grad_norm": 1.0559046268463135,
+      "learning_rate": 2.4321535453376472e-05,
+      "loss": 0.434,
+      "step": 755
+    },
+    {
+      "epoch": 1.623931623931624,
+      "grad_norm": 1.0688294172286987,
+      "learning_rate": 2.4238262663517944e-05,
+      "loss": 0.4318,
+      "step": 760
+    },
+    {
+      "epoch": 1.6346153846153846,
+      "grad_norm": 0.979230523109436,
+      "learning_rate": 2.4154528609907057e-05,
+      "loss": 0.4138,
+      "step": 765
+    },
+    {
+      "epoch": 1.6452991452991452,
+      "grad_norm": 1.081634283065796,
+      "learning_rate": 2.407033747336062e-05,
+      "loss": 0.48,
+      "step": 770
+    },
+    {
+      "epoch": 1.6559829059829059,
+      "grad_norm": 1.0213351249694824,
+      "learning_rate": 2.3985693457517444e-05,
+      "loss": 0.4375,
+      "step": 775
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 1.061716914176941,
+      "learning_rate": 2.390060078862848e-05,
+      "loss": 0.4272,
+      "step": 780
+    },
+    {
+      "epoch": 1.6773504273504274,
+      "grad_norm": 1.0171300172805786,
+      "learning_rate": 2.38150637153458e-05,
+      "loss": 0.4202,
+      "step": 785
+    },
+    {
+      "epoch": 1.688034188034188,
+      "grad_norm": 1.0210041999816895,
+      "learning_rate": 2.372908650851044e-05,
+      "loss": 0.4101,
+      "step": 790
+    },
+    {
+      "epoch": 1.6987179487179487,
+      "grad_norm": 1.0023560523986816,
+      "learning_rate": 2.3642673460939205e-05,
+      "loss": 0.4524,
+      "step": 795
+    },
+    {
+      "epoch": 1.7094017094017095,
+      "grad_norm": 0.9828326106071472,
+      "learning_rate": 2.355582888721026e-05,
+      "loss": 0.3943,
+      "step": 800
+    },
+    {
+      "epoch": 1.7200854700854702,
+      "grad_norm": 2.0454530715942383,
+      "learning_rate": 2.3468557123447784e-05,
+      "loss": 0.4166,
+      "step": 805
+    },
+    {
+      "epoch": 1.7307692307692308,
+      "grad_norm": 0.9595752954483032,
+      "learning_rate": 2.338086252710541e-05,
+      "loss": 0.4313,
+      "step": 810
+    },
+    {
+      "epoch": 1.7414529914529915,
+      "grad_norm": 1.0404924154281616,
+      "learning_rate": 2.3292749476748668e-05,
+      "loss": 0.3915,
+      "step": 815
+    },
+    {
+      "epoch": 1.7521367521367521,
+      "grad_norm": 0.9760511517524719,
+      "learning_rate": 2.320422237183641e-05,
+      "loss": 0.4098,
+      "step": 820
+    },
+    {
+      "epoch": 1.7628205128205128,
+      "grad_norm": 1.131895661354065,
+      "learning_rate": 2.3115285632501058e-05,
+      "loss": 0.4083,
+      "step": 825
+    },
+    {
+      "epoch": 1.7735042735042734,
+      "grad_norm": 1.0025837421417236,
+      "learning_rate": 2.3025943699328016e-05,
+      "loss": 0.4034,
+      "step": 830
+    },
+    {
+      "epoch": 1.784188034188034,
+      "grad_norm": 1.1371080875396729,
+      "learning_rate": 2.293620103313386e-05,
+      "loss": 0.3842,
+      "step": 835
+    },
+    {
+      "epoch": 1.7948717948717947,
+      "grad_norm": 1.0681666135787964,
+      "learning_rate": 2.2846062114743665e-05,
+      "loss": 0.405,
+      "step": 840
+    },
+    {
+      "epoch": 1.8055555555555556,
+      "grad_norm": 1.0509333610534668,
+      "learning_rate": 2.275553144476726e-05,
+      "loss": 0.3983,
+      "step": 845
+    },
+    {
+      "epoch": 1.8162393162393162,
+      "grad_norm": 0.9554222822189331,
+      "learning_rate": 2.2664613543374507e-05,
+      "loss": 0.3617,
+      "step": 850
+    },
+    {
+      "epoch": 1.8269230769230769,
+      "grad_norm": 1.0312929153442383,
+      "learning_rate": 2.2573312950069632e-05,
+      "loss": 0.4084,
+      "step": 855
+    },
+    {
+      "epoch": 1.8376068376068377,
+      "grad_norm": 1.067847490310669,
+      "learning_rate": 2.2481634223464552e-05,
+      "loss": 0.4053,
+      "step": 860
+    },
+    {
+      "epoch": 1.8482905982905984,
+      "grad_norm": 1.2075289487838745,
+      "learning_rate": 2.2389581941051264e-05,
+      "loss": 0.3847,
+      "step": 865
+    },
+    {
+      "epoch": 1.858974358974359,
+      "grad_norm": 1.0252904891967773,
+      "learning_rate": 2.2297160698973307e-05,
+      "loss": 0.3814,
+      "step": 870
+    },
+    {
+      "epoch": 1.8696581196581197,
+      "grad_norm": 1.0117474794387817,
+      "learning_rate": 2.220437511179626e-05,
+      "loss": 0.371,
+      "step": 875
+    },
+    {
+      "epoch": 1.8803418803418803,
+      "grad_norm": 1.084829568862915,
+      "learning_rate": 2.211122981227734e-05,
+      "loss": 0.3972,
+      "step": 880
+    },
+    {
+      "epoch": 1.891025641025641,
+      "grad_norm": 1.0629417896270752,
+      "learning_rate": 2.201772945113412e-05,
+      "loss": 0.3423,
+      "step": 885
+    },
+    {
+      "epoch": 1.9017094017094016,
+      "grad_norm": 0.9619771242141724,
+      "learning_rate": 2.192387869681227e-05,
+      "loss": 0.3673,
+      "step": 890
+    },
+    {
+      "epoch": 1.9123931623931623,
+      "grad_norm": 1.0823090076446533,
+      "learning_rate": 2.18296822352525e-05,
+      "loss": 0.3562,
+      "step": 895
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 1.0876190662384033,
+      "learning_rate": 2.1735144769656584e-05,
+      "loss": 0.3394,
+      "step": 900
+    },
+    {
+      "epoch": 1.9337606837606838,
+      "grad_norm": 0.9800743460655212,
+      "learning_rate": 2.164027102025252e-05,
+      "loss": 0.3849,
+      "step": 905
+    },
+    {
+      "epoch": 1.9444444444444444,
+      "grad_norm": 1.1516016721725464,
+      "learning_rate": 2.1545065724058873e-05,
+      "loss": 0.4077,
+      "step": 910
+    },
+    {
+      "epoch": 1.9551282051282053,
+      "grad_norm": 1.0895222425460815,
+      "learning_rate": 2.1449533634648223e-05,
+      "loss": 0.3865,
+      "step": 915
+    },
+    {
+      "epoch": 1.965811965811966,
+      "grad_norm": 1.1499086618423462,
+      "learning_rate": 2.1353679521909848e-05,
+      "loss": 0.3547,
+      "step": 920
+    },
+    {
+      "epoch": 1.9764957264957266,
+      "grad_norm": 1.1656689643859863,
+      "learning_rate": 2.1257508171811563e-05,
+      "loss": 0.3852,
+      "step": 925
+    },
+    {
+      "epoch": 1.9871794871794872,
+      "grad_norm": 0.9602221250534058,
+      "learning_rate": 2.116102438616074e-05,
+      "loss": 0.3572,
+      "step": 930
+    },
+    {
+      "epoch": 1.9978632478632479,
+      "grad_norm": 1.0611974000930786,
+      "learning_rate": 2.1064232982364572e-05,
+      "loss": 0.3635,
+      "step": 935
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 2340,
+  "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": 1.4249203492118856e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

29/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:296526ae4e1aa846a2c085b140cba5f70a034f754dad00ea7ce96dc3906fe6cf
+size 7736

29/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)