Upload folder using huggingface_hub

debug_klora/1/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

debug_klora/1/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": [
+    "down_proj",
+    "v_proj",
+    "o_proj",
+    "q_proj",
+    "up_proj",
+    "k_proj",
+    "gate_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

debug_klora/1/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7690be60adb3846946da8024f2bf779a92a3d4eb98224e0db843c39353385617
+size 791751704

debug_klora/1/latest

@@ -0,0 +1 @@
+global_step934

debug_klora/1/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91e91bf29172fe4593c8d7fe98cad01dff01a8db1d2b259b5b797f8c3e7e5718
+size 15920

debug_klora/1/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:550c18859c913ab88c73fc34389bc6d6a7db54b9afd8c33ac3874748777f9651
+size 15920

debug_klora/1/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64d5a5e91fe43a5d58bcbe83085363bcc7054113badce3eee5b63a61508d0d22
+size 15920

debug_klora/1/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e99aee52786fa23633133ba245cbcdd11b14466ff50a7e764539634b0ff33a4f
+size 15920

debug_klora/1/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:75e997a2a441f93ea8067f7478659feb861d18026ea9304edf285f5872a477c3
+size 15920

debug_klora/1/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aacd5049305c1dce80aa6e21ab6d9b04bd17b8d680cf2f4faeb18a28016ee1d8
+size 15920

debug_klora/1/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2d88acd0f23f6129ffdcad6d316929965e85cd5398d8dd9eeb8e4839232d30b0
+size 15920

debug_klora/1/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:db7df0b69964fd9d124a42d3679f64d3e1eb367740aeaeb9721f028cef73c1d9
+size 15920

debug_klora/1/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:043a315afddd75b425457d271358acac4017d20dff204ed4be3927d173faf3b9
+size 1064

debug_klora/1/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
+  }
+}

debug_klora/1/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
+}

debug_klora/1/trainer_state.json

@@ -0,0 +1,1336 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 934,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.010718113612004287,
+      "grad_norm": 1.3203599452972412,
+      "learning_rate": 1.0256410256410257e-06,
+      "loss": 1.3095,
+      "step": 5
+    },
+    {
+      "epoch": 0.021436227224008574,
+      "grad_norm": 1.005972981452942,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 1.3756,
+      "step": 10
+    },
+    {
+      "epoch": 0.03215434083601286,
+      "grad_norm": 0.8182684183120728,
+      "learning_rate": 3.58974358974359e-06,
+      "loss": 1.3238,
+      "step": 15
+    },
+    {
+      "epoch": 0.04287245444801715,
+      "grad_norm": 0.791577160358429,
+      "learning_rate": 4.871794871794872e-06,
+      "loss": 1.2938,
+      "step": 20
+    },
+    {
+      "epoch": 0.05359056806002144,
+      "grad_norm": 0.6379374861717224,
+      "learning_rate": 6.153846153846154e-06,
+      "loss": 1.2856,
+      "step": 25
+    },
+    {
+      "epoch": 0.06430868167202572,
+      "grad_norm": 0.6668587327003479,
+      "learning_rate": 7.435897435897436e-06,
+      "loss": 1.2841,
+      "step": 30
+    },
+    {
+      "epoch": 0.07502679528403002,
+      "grad_norm": 0.716699481010437,
+      "learning_rate": 8.717948717948719e-06,
+      "loss": 1.2718,
+      "step": 35
+    },
+    {
+      "epoch": 0.0857449088960343,
+      "grad_norm": 0.5486516952514648,
+      "learning_rate": 9.999999999999999e-06,
+      "loss": 1.273,
+      "step": 40
+    },
+    {
+      "epoch": 0.09646302250803858,
+      "grad_norm": 0.46251392364501953,
+      "learning_rate": 1.1282051282051283e-05,
+      "loss": 1.312,
+      "step": 45
+    },
+    {
+      "epoch": 0.10718113612004287,
+      "grad_norm": 0.4610588848590851,
+      "learning_rate": 1.2564102564102565e-05,
+      "loss": 1.2015,
+      "step": 50
+    },
+    {
+      "epoch": 0.11789924973204716,
+      "grad_norm": 0.4434472918510437,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 1.2457,
+      "step": 55
+    },
+    {
+      "epoch": 0.12861736334405144,
+      "grad_norm": 0.43525955080986023,
+      "learning_rate": 1.5128205128205129e-05,
+      "loss": 1.2111,
+      "step": 60
+    },
+    {
+      "epoch": 0.13933547695605572,
+      "grad_norm": 0.45412635803222656,
+      "learning_rate": 1.641025641025641e-05,
+      "loss": 1.2055,
+      "step": 65
+    },
+    {
+      "epoch": 0.15005359056806003,
+      "grad_norm": 0.47974953055381775,
+      "learning_rate": 1.7692307692307694e-05,
+      "loss": 1.1836,
+      "step": 70
+    },
+    {
+      "epoch": 0.1607717041800643,
+      "grad_norm": 0.5585247278213501,
+      "learning_rate": 1.8974358974358975e-05,
+      "loss": 1.1456,
+      "step": 75
+    },
+    {
+      "epoch": 0.1714898177920686,
+      "grad_norm": 0.4740090072154999,
+      "learning_rate": 2.025641025641026e-05,
+      "loss": 1.1697,
+      "step": 80
+    },
+    {
+      "epoch": 0.18220793140407288,
+      "grad_norm": 0.49104949831962585,
+      "learning_rate": 2.153846153846154e-05,
+      "loss": 1.1723,
+      "step": 85
+    },
+    {
+      "epoch": 0.19292604501607716,
+      "grad_norm": 0.47726204991340637,
+      "learning_rate": 2.282051282051282e-05,
+      "loss": 1.1729,
+      "step": 90
+    },
+    {
+      "epoch": 0.20364415862808147,
+      "grad_norm": 0.5089394450187683,
+      "learning_rate": 2.4102564102564103e-05,
+      "loss": 1.1313,
+      "step": 95
+    },
+    {
+      "epoch": 0.21436227224008575,
+      "grad_norm": 0.4856749475002289,
+      "learning_rate": 2.5384615384615386e-05,
+      "loss": 1.1221,
+      "step": 100
+    },
+    {
+      "epoch": 0.22508038585209003,
+      "grad_norm": 0.4452688992023468,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 1.1334,
+      "step": 105
+    },
+    {
+      "epoch": 0.2357984994640943,
+      "grad_norm": 0.5895005464553833,
+      "learning_rate": 2.794871794871795e-05,
+      "loss": 1.1453,
+      "step": 110
+    },
+    {
+      "epoch": 0.2465166130760986,
+      "grad_norm": 0.5326249003410339,
+      "learning_rate": 2.923076923076923e-05,
+      "loss": 1.0856,
+      "step": 115
+    },
+    {
+      "epoch": 0.2572347266881029,
+      "grad_norm": 0.588718056678772,
+      "learning_rate": 2.9999939813702703e-05,
+      "loss": 1.0957,
+      "step": 120
+    },
+    {
+      "epoch": 0.2679528403001072,
+      "grad_norm": 0.4917924106121063,
+      "learning_rate": 2.9999262723404875e-05,
+      "loss": 1.0265,
+      "step": 125
+    },
+    {
+      "epoch": 0.27867095391211144,
+      "grad_norm": 0.624920666217804,
+      "learning_rate": 2.9997833344010312e-05,
+      "loss": 1.0845,
+      "step": 130
+    },
+    {
+      "epoch": 0.28938906752411575,
+      "grad_norm": 0.627967119216919,
+      "learning_rate": 2.9995651747209637e-05,
+      "loss": 1.1037,
+      "step": 135
+    },
+    {
+      "epoch": 0.30010718113612006,
+      "grad_norm": 0.5074423551559448,
+      "learning_rate": 2.9992718042420993e-05,
+      "loss": 1.0788,
+      "step": 140
+    },
+    {
+      "epoch": 0.3108252947481243,
+      "grad_norm": 0.5567290186882019,
+      "learning_rate": 2.9989032376784556e-05,
+      "loss": 1.1003,
+      "step": 145
+    },
+    {
+      "epoch": 0.3215434083601286,
+      "grad_norm": 0.5891143083572388,
+      "learning_rate": 2.998459493515515e-05,
+      "loss": 1.0162,
+      "step": 150
+    },
+    {
+      "epoch": 0.3322615219721329,
+      "grad_norm": 0.6841644048690796,
+      "learning_rate": 2.9979405940092978e-05,
+      "loss": 1.0204,
+      "step": 155
+    },
+    {
+      "epoch": 0.3429796355841372,
+      "grad_norm": 0.6039224863052368,
+      "learning_rate": 2.997346565185246e-05,
+      "loss": 1.0219,
+      "step": 160
+    },
+    {
+      "epoch": 0.3536977491961415,
+      "grad_norm": 0.5960545539855957,
+      "learning_rate": 2.9966774368369175e-05,
+      "loss": 1.0065,
+      "step": 165
+    },
+    {
+      "epoch": 0.36441586280814575,
+      "grad_norm": 0.6452131867408752,
+      "learning_rate": 2.995933242524493e-05,
+      "loss": 0.9917,
+      "step": 170
+    },
+    {
+      "epoch": 0.37513397642015006,
+      "grad_norm": 0.5966811776161194,
+      "learning_rate": 2.995114019573091e-05,
+      "loss": 0.9868,
+      "step": 175
+    },
+    {
+      "epoch": 0.3858520900321543,
+      "grad_norm": 0.6129615306854248,
+      "learning_rate": 2.9942198090708976e-05,
+      "loss": 1.034,
+      "step": 180
+    },
+    {
+      "epoch": 0.3965702036441586,
+      "grad_norm": 0.6630488038063049,
+      "learning_rate": 2.993250655867105e-05,
+      "loss": 1.0303,
+      "step": 185
+    },
+    {
+      "epoch": 0.40728831725616294,
+      "grad_norm": 0.6165331602096558,
+      "learning_rate": 2.9922066085696613e-05,
+      "loss": 0.987,
+      "step": 190
+    },
+    {
+      "epoch": 0.4180064308681672,
+      "grad_norm": 0.7296874523162842,
+      "learning_rate": 2.9910877195428335e-05,
+      "loss": 0.9805,
+      "step": 195
+    },
+    {
+      "epoch": 0.4287245444801715,
+      "grad_norm": 0.6991822123527527,
+      "learning_rate": 2.989894044904581e-05,
+      "loss": 0.9377,
+      "step": 200
+    },
+    {
+      "epoch": 0.43944265809217575,
+      "grad_norm": 0.7727455496788025,
+      "learning_rate": 2.9886256445237405e-05,
+      "loss": 0.9726,
+      "step": 205
+    },
+    {
+      "epoch": 0.45016077170418006,
+      "grad_norm": 0.7376512289047241,
+      "learning_rate": 2.987282582017024e-05,
+      "loss": 0.946,
+      "step": 210
+    },
+    {
+      "epoch": 0.4608788853161844,
+      "grad_norm": 0.819678544998169,
+      "learning_rate": 2.9858649247458287e-05,
+      "loss": 0.9861,
+      "step": 215
+    },
+    {
+      "epoch": 0.4715969989281886,
+      "grad_norm": 0.7300285696983337,
+      "learning_rate": 2.984372743812855e-05,
+      "loss": 1.0026,
+      "step": 220
+    },
+    {
+      "epoch": 0.48231511254019294,
+      "grad_norm": 0.7369091510772705,
+      "learning_rate": 2.9828061140585463e-05,
+      "loss": 0.9064,
+      "step": 225
+    },
+    {
+      "epoch": 0.4930332261521972,
+      "grad_norm": 0.6813873052597046,
+      "learning_rate": 2.9811651140573287e-05,
+      "loss": 0.8859,
+      "step": 230
+    },
+    {
+      "epoch": 0.5037513397642015,
+      "grad_norm": 0.7347345352172852,
+      "learning_rate": 2.979449826113676e-05,
+      "loss": 0.8983,
+      "step": 235
+    },
+    {
+      "epoch": 0.5144694533762058,
+      "grad_norm": 0.8283259272575378,
+      "learning_rate": 2.9776603362579775e-05,
+      "loss": 0.8876,
+      "step": 240
+    },
+    {
+      "epoch": 0.5251875669882101,
+      "grad_norm": 0.7388147711753845,
+      "learning_rate": 2.9757967342422246e-05,
+      "loss": 0.899,
+      "step": 245
+    },
+    {
+      "epoch": 0.5359056806002144,
+      "grad_norm": 0.7657457590103149,
+      "learning_rate": 2.9738591135355108e-05,
+      "loss": 0.9412,
+      "step": 250
+    },
+    {
+      "epoch": 0.5466237942122186,
+      "grad_norm": 0.8804264664649963,
+      "learning_rate": 2.971847571319341e-05,
+      "loss": 0.8838,
+      "step": 255
+    },
+    {
+      "epoch": 0.5573419078242229,
+      "grad_norm": 0.771865963935852,
+      "learning_rate": 2.969762208482759e-05,
+      "loss": 0.8697,
+      "step": 260
+    },
+    {
+      "epoch": 0.5680600214362272,
+      "grad_norm": 0.7006550431251526,
+      "learning_rate": 2.967603129617288e-05,
+      "loss": 0.8801,
+      "step": 265
+    },
+    {
+      "epoch": 0.5787781350482315,
+      "grad_norm": 0.7909011244773865,
+      "learning_rate": 2.9653704430116827e-05,
+      "loss": 0.9199,
+      "step": 270
+    },
+    {
+      "epoch": 0.5894962486602358,
+      "grad_norm": 0.9315905570983887,
+      "learning_rate": 2.9630642606464998e-05,
+      "loss": 0.8249,
+      "step": 275
+    },
+    {
+      "epoch": 0.6002143622722401,
+      "grad_norm": 0.7528437376022339,
+      "learning_rate": 2.9606846981884812e-05,
+      "loss": 0.8656,
+      "step": 280
+    },
+    {
+      "epoch": 0.6109324758842444,
+      "grad_norm": 0.7914732098579407,
+      "learning_rate": 2.958231874984753e-05,
+      "loss": 0.8618,
+      "step": 285
+    },
+    {
+      "epoch": 0.6216505894962486,
+      "grad_norm": 0.7694675922393799,
+      "learning_rate": 2.955705914056838e-05,
+      "loss": 0.8546,
+      "step": 290
+    },
+    {
+      "epoch": 0.632368703108253,
+      "grad_norm": 0.8651600480079651,
+      "learning_rate": 2.9531069420944876e-05,
+      "loss": 0.8654,
+      "step": 295
+    },
+    {
+      "epoch": 0.6430868167202572,
+      "grad_norm": 0.7878230214118958,
+      "learning_rate": 2.9504350894493272e-05,
+      "loss": 0.8624,
+      "step": 300
+    },
+    {
+      "epoch": 0.6538049303322615,
+      "grad_norm": 0.9521771669387817,
+      "learning_rate": 2.9476904901283175e-05,
+      "loss": 0.8175,
+      "step": 305
+    },
+    {
+      "epoch": 0.6645230439442658,
+      "grad_norm": 0.8132469058036804,
+      "learning_rate": 2.944873281787034e-05,
+      "loss": 0.8087,
+      "step": 310
+    },
+    {
+      "epoch": 0.6752411575562701,
+      "grad_norm": 0.8626110553741455,
+      "learning_rate": 2.9419836057227634e-05,
+      "loss": 0.8033,
+      "step": 315
+    },
+    {
+      "epoch": 0.6859592711682744,
+      "grad_norm": 0.858932614326477,
+      "learning_rate": 2.9390216068674154e-05,
+      "loss": 0.7535,
+      "step": 320
+    },
+    {
+      "epoch": 0.6966773847802786,
+      "grad_norm": 1.0335079431533813,
+      "learning_rate": 2.9359874337802542e-05,
+      "loss": 0.8272,
+      "step": 325
+    },
+    {
+      "epoch": 0.707395498392283,
+      "grad_norm": 0.819476842880249,
+      "learning_rate": 2.9328812386404503e-05,
+      "loss": 0.8184,
+      "step": 330
+    },
+    {
+      "epoch": 0.7181136120042872,
+      "grad_norm": 0.9225292205810547,
+      "learning_rate": 2.9297031772394412e-05,
+      "loss": 0.8009,
+      "step": 335
+    },
+    {
+      "epoch": 0.7288317256162915,
+      "grad_norm": 0.9197941422462463,
+      "learning_rate": 2.9264534089731262e-05,
+      "loss": 0.7841,
+      "step": 340
+    },
+    {
+      "epoch": 0.7395498392282959,
+      "grad_norm": 0.8611735701560974,
+      "learning_rate": 2.9231320968338644e-05,
+      "loss": 0.7975,
+      "step": 345
+    },
+    {
+      "epoch": 0.7502679528403001,
+      "grad_norm": 0.8244771361351013,
+      "learning_rate": 2.9197394074023042e-05,
+      "loss": 0.7895,
+      "step": 350
+    },
+    {
+      "epoch": 0.7609860664523044,
+      "grad_norm": 0.8496798872947693,
+      "learning_rate": 2.9162755108390278e-05,
+      "loss": 0.7931,
+      "step": 355
+    },
+    {
+      "epoch": 0.7717041800643086,
+      "grad_norm": 0.9810552000999451,
+      "learning_rate": 2.9127405808760152e-05,
+      "loss": 0.7507,
+      "step": 360
+    },
+    {
+      "epoch": 0.782422293676313,
+      "grad_norm": 0.9915773272514343,
+      "learning_rate": 2.9091347948079318e-05,
+      "loss": 0.8251,
+      "step": 365
+    },
+    {
+      "epoch": 0.7931404072883173,
+      "grad_norm": 1.0386714935302734,
+      "learning_rate": 2.9054583334832366e-05,
+      "loss": 0.8063,
+      "step": 370
+    },
+    {
+      "epoch": 0.8038585209003215,
+      "grad_norm": 0.9649940729141235,
+      "learning_rate": 2.9017113812951105e-05,
+      "loss": 0.7797,
+      "step": 375
+    },
+    {
+      "epoch": 0.8145766345123259,
+      "grad_norm": 1.0805562734603882,
+      "learning_rate": 2.897894126172209e-05,
+      "loss": 0.7873,
+      "step": 380
+    },
+    {
+      "epoch": 0.8252947481243301,
+      "grad_norm": 0.9175759553909302,
+      "learning_rate": 2.8940067595692367e-05,
+      "loss": 0.7306,
+      "step": 385
+    },
+    {
+      "epoch": 0.8360128617363344,
+      "grad_norm": 1.0049532651901245,
+      "learning_rate": 2.890049476457344e-05,
+      "loss": 0.7357,
+      "step": 390
+    },
+    {
+      "epoch": 0.8467309753483387,
+      "grad_norm": 0.8982154726982117,
+      "learning_rate": 2.8860224753143486e-05,
+      "loss": 0.7478,
+      "step": 395
+    },
+    {
+      "epoch": 0.857449088960343,
+      "grad_norm": 0.898063600063324,
+      "learning_rate": 2.881925958114782e-05,
+      "loss": 0.7576,
+      "step": 400
+    },
+    {
+      "epoch": 0.8681672025723473,
+      "grad_norm": 0.9208119511604309,
+      "learning_rate": 2.877760130319757e-05,
+      "loss": 0.7413,
+      "step": 405
+    },
+    {
+      "epoch": 0.8788853161843515,
+      "grad_norm": 0.9576625823974609,
+      "learning_rate": 2.8735252008666663e-05,
+      "loss": 0.6863,
+      "step": 410
+    },
+    {
+      "epoch": 0.8896034297963559,
+      "grad_norm": 0.9474856853485107,
+      "learning_rate": 2.8692213821586988e-05,
+      "loss": 0.7466,
+      "step": 415
+    },
+    {
+      "epoch": 0.9003215434083601,
+      "grad_norm": 1.0231027603149414,
+      "learning_rate": 2.8648488900541915e-05,
+      "loss": 0.7399,
+      "step": 420
+    },
+    {
+      "epoch": 0.9110396570203644,
+      "grad_norm": 0.9832124710083008,
+      "learning_rate": 2.8604079438557985e-05,
+      "loss": 0.7037,
+      "step": 425
+    },
+    {
+      "epoch": 0.9217577706323687,
+      "grad_norm": 1.0980058908462524,
+      "learning_rate": 2.8558987662994948e-05,
+      "loss": 0.7189,
+      "step": 430
+    },
+    {
+      "epoch": 0.932475884244373,
+      "grad_norm": 1.1283589601516724,
+      "learning_rate": 2.851321583543404e-05,
+      "loss": 0.6951,
+      "step": 435
+    },
+    {
+      "epoch": 0.9431939978563773,
+      "grad_norm": 0.9087287187576294,
+      "learning_rate": 2.8466766251564554e-05,
+      "loss": 0.6932,
+      "step": 440
+    },
+    {
+      "epoch": 0.9539121114683816,
+      "grad_norm": 1.0392457246780396,
+      "learning_rate": 2.8419641241068687e-05,
+      "loss": 0.7065,
+      "step": 445
+    },
+    {
+      "epoch": 0.9646302250803859,
+      "grad_norm": 0.9226632118225098,
+      "learning_rate": 2.837184316750473e-05,
+      "loss": 0.7066,
+      "step": 450
+    },
+    {
+      "epoch": 0.9753483386923901,
+      "grad_norm": 0.9685462713241577,
+      "learning_rate": 2.832337442818848e-05,
+      "loss": 0.6805,
+      "step": 455
+    },
+    {
+      "epoch": 0.9860664523043944,
+      "grad_norm": 0.9703472852706909,
+      "learning_rate": 2.8274237454073034e-05,
+      "loss": 0.668,
+      "step": 460
+    },
+    {
+      "epoch": 0.9967845659163987,
+      "grad_norm": 0.902520477771759,
+      "learning_rate": 2.822443470962685e-05,
+      "loss": 0.6379,
+      "step": 465
+    },
+    {
+      "epoch": 1.0064308681672025,
+      "grad_norm": 1.2527391910552979,
+      "learning_rate": 2.8173968692710142e-05,
+      "loss": 0.6063,
+      "step": 470
+    },
+    {
+      "epoch": 1.0171489817792068,
+      "grad_norm": 1.0625535249710083,
+      "learning_rate": 2.8122841934449607e-05,
+      "loss": 0.5916,
+      "step": 475
+    },
+    {
+      "epoch": 1.0278670953912112,
+      "grad_norm": 1.1103384494781494,
+      "learning_rate": 2.807105699911147e-05,
+      "loss": 0.6323,
+      "step": 480
+    },
+    {
+      "epoch": 1.0385852090032155,
+      "grad_norm": 1.0214422941207886,
+      "learning_rate": 2.8018616483972874e-05,
+      "loss": 0.573,
+      "step": 485
+    },
+    {
+      "epoch": 1.0493033226152197,
+      "grad_norm": 0.9643238186836243,
+      "learning_rate": 2.7965523019191606e-05,
+      "loss": 0.5675,
+      "step": 490
+    },
+    {
+      "epoch": 1.060021436227224,
+      "grad_norm": 0.9685465097427368,
+      "learning_rate": 2.7911779267674207e-05,
+      "loss": 0.5708,
+      "step": 495
+    },
+    {
+      "epoch": 1.0707395498392283,
+      "grad_norm": 1.0887079238891602,
+      "learning_rate": 2.7857387924942387e-05,
+      "loss": 0.5954,
+      "step": 500
+    },
+    {
+      "epoch": 1.0814576634512325,
+      "grad_norm": 0.9755250811576843,
+      "learning_rate": 2.7802351718997825e-05,
+      "loss": 0.5384,
+      "step": 505
+    },
+    {
+      "epoch": 1.092175777063237,
+      "grad_norm": 0.9574642181396484,
+      "learning_rate": 2.7746673410185387e-05,
+      "loss": 0.5627,
+      "step": 510
+    },
+    {
+      "epoch": 1.1028938906752412,
+      "grad_norm": 1.0391875505447388,
+      "learning_rate": 2.7690355791054632e-05,
+      "loss": 0.5621,
+      "step": 515
+    },
+    {
+      "epoch": 1.1136120042872455,
+      "grad_norm": 0.9786142706871033,
+      "learning_rate": 2.7633401686219793e-05,
+      "loss": 0.5991,
+      "step": 520
+    },
+    {
+      "epoch": 1.1243301178992497,
+      "grad_norm": 1.0193188190460205,
+      "learning_rate": 2.7575813952218075e-05,
+      "loss": 0.5723,
+      "step": 525
+    },
+    {
+      "epoch": 1.135048231511254,
+      "grad_norm": 1.0080389976501465,
+      "learning_rate": 2.75175954773664e-05,
+      "loss": 0.5206,
+      "step": 530
+    },
+    {
+      "epoch": 1.1457663451232583,
+      "grad_norm": 0.9439852833747864,
+      "learning_rate": 2.7458749181616545e-05,
+      "loss": 0.571,
+      "step": 535
+    },
+    {
+      "epoch": 1.1564844587352625,
+      "grad_norm": 0.9455674290657043,
+      "learning_rate": 2.7399278016408695e-05,
+      "loss": 0.547,
+      "step": 540
+    },
+    {
+      "epoch": 1.167202572347267,
+      "grad_norm": 1.138099193572998,
+      "learning_rate": 2.7339184964523404e-05,
+      "loss": 0.5942,
+      "step": 545
+    },
+    {
+      "epoch": 1.1779206859592712,
+      "grad_norm": 1.1031590700149536,
+      "learning_rate": 2.7278473039931992e-05,
+      "loss": 0.5378,
+      "step": 550
+    },
+    {
+      "epoch": 1.1886387995712755,
+      "grad_norm": 1.1105057001113892,
+      "learning_rate": 2.7217145287645393e-05,
+      "loss": 0.5186,
+      "step": 555
+    },
+    {
+      "epoch": 1.1993569131832797,
+      "grad_norm": 1.2202413082122803,
+      "learning_rate": 2.715520478356141e-05,
+      "loss": 0.5721,
+      "step": 560
+    },
+    {
+      "epoch": 1.210075026795284,
+      "grad_norm": 0.9721954464912415,
+      "learning_rate": 2.7092654634310468e-05,
+      "loss": 0.5433,
+      "step": 565
+    },
+    {
+      "epoch": 1.2207931404072883,
+      "grad_norm": 1.2260032892227173,
+      "learning_rate": 2.7029497977099784e-05,
+      "loss": 0.543,
+      "step": 570
+    },
+    {
+      "epoch": 1.2315112540192925,
+      "grad_norm": 0.9644677639007568,
+      "learning_rate": 2.6965737979556025e-05,
+      "loss": 0.552,
+      "step": 575
+    },
+    {
+      "epoch": 1.242229367631297,
+      "grad_norm": 0.9937833547592163,
+      "learning_rate": 2.6901377839566443e-05,
+      "loss": 0.535,
+      "step": 580
+    },
+    {
+      "epoch": 1.2529474812433012,
+      "grad_norm": 1.031482219696045,
+      "learning_rate": 2.683642078511846e-05,
+      "loss": 0.5322,
+      "step": 585
+    },
+    {
+      "epoch": 1.2636655948553055,
+      "grad_norm": 0.9676262140274048,
+      "learning_rate": 2.6770870074137808e-05,
+      "loss": 0.5006,
+      "step": 590
+    },
+    {
+      "epoch": 1.2743837084673098,
+      "grad_norm": 1.3593294620513916,
+      "learning_rate": 2.670472899432509e-05,
+      "loss": 0.5064,
+      "step": 595
+    },
+    {
+      "epoch": 1.285101822079314,
+      "grad_norm": 1.0694024562835693,
+      "learning_rate": 2.66380008629909e-05,
+      "loss": 0.5848,
+      "step": 600
+    },
+    {
+      "epoch": 1.2958199356913183,
+      "grad_norm": 1.0443016290664673,
+      "learning_rate": 2.657068902688945e-05,
+      "loss": 0.538,
+      "step": 605
+    },
+    {
+      "epoch": 1.3065380493033225,
+      "grad_norm": 1.023464560508728,
+      "learning_rate": 2.65027968620507e-05,
+      "loss": 0.5587,
+      "step": 610
+    },
+    {
+      "epoch": 1.317256162915327,
+      "grad_norm": 0.9948901534080505,
+      "learning_rate": 2.643432777361105e-05,
+      "loss": 0.534,
+      "step": 615
+    },
+    {
+      "epoch": 1.3279742765273312,
+      "grad_norm": 1.0392403602600098,
+      "learning_rate": 2.636528519564253e-05,
+      "loss": 0.5365,
+      "step": 620
+    },
+    {
+      "epoch": 1.3386923901393355,
+      "grad_norm": 1.0413520336151123,
+      "learning_rate": 2.6295672590980586e-05,
+      "loss": 0.4708,
+      "step": 625
+    },
+    {
+      "epoch": 1.3494105037513398,
+      "grad_norm": 1.1629996299743652,
+      "learning_rate": 2.6225493451050397e-05,
+      "loss": 0.5554,
+      "step": 630
+    },
+    {
+      "epoch": 1.360128617363344,
+      "grad_norm": 1.023750901222229,
+      "learning_rate": 2.6154751295691756e-05,
+      "loss": 0.4827,
+      "step": 635
+    },
+    {
+      "epoch": 1.3708467309753483,
+      "grad_norm": 1.13310706615448,
+      "learning_rate": 2.608344967298253e-05,
+      "loss": 0.4657,
+      "step": 640
+    },
+    {
+      "epoch": 1.3815648445873525,
+      "grad_norm": 1.083983063697815,
+      "learning_rate": 2.6011592159060717e-05,
+      "loss": 0.5213,
+      "step": 645
+    },
+    {
+      "epoch": 1.392282958199357,
+      "grad_norm": 1.048169493675232,
+      "learning_rate": 2.593918235794508e-05,
+      "loss": 0.4819,
+      "step": 650
+    },
+    {
+      "epoch": 1.4030010718113612,
+      "grad_norm": 0.9783421158790588,
+      "learning_rate": 2.5866223901354372e-05,
+      "loss": 0.5378,
+      "step": 655
+    },
+    {
+      "epoch": 1.4137191854233655,
+      "grad_norm": 1.2764405012130737,
+      "learning_rate": 2.579272044852523e-05,
+      "loss": 0.4782,
+      "step": 660
+    },
+    {
+      "epoch": 1.4244372990353698,
+      "grad_norm": 1.0567833185195923,
+      "learning_rate": 2.5718675686028582e-05,
+      "loss": 0.5,
+      "step": 665
+    },
+    {
+      "epoch": 1.435155412647374,
+      "grad_norm": 1.150490164756775,
+      "learning_rate": 2.56440933275848e-05,
+      "loss": 0.4415,
+      "step": 670
+    },
+    {
+      "epoch": 1.4458735262593785,
+      "grad_norm": 1.0718622207641602,
+      "learning_rate": 2.5568977113877426e-05,
+      "loss": 0.5022,
+      "step": 675
+    },
+    {
+      "epoch": 1.4565916398713825,
+      "grad_norm": 1.0396467447280884,
+      "learning_rate": 2.549333081236553e-05,
+      "loss": 0.4582,
+      "step": 680
+    },
+    {
+      "epoch": 1.467309753483387,
+      "grad_norm": 1.0076903104782104,
+      "learning_rate": 2.5417158217094788e-05,
+      "loss": 0.4865,
+      "step": 685
+    },
+    {
+      "epoch": 1.4780278670953912,
+      "grad_norm": 1.1110858917236328,
+      "learning_rate": 2.5340463148507178e-05,
+      "loss": 0.4707,
+      "step": 690
+    },
+    {
+      "epoch": 1.4887459807073955,
+      "grad_norm": 1.0835309028625488,
+      "learning_rate": 2.5263249453249356e-05,
+      "loss": 0.4569,
+      "step": 695
+    },
+    {
+      "epoch": 1.4994640943193998,
+      "grad_norm": 1.1380281448364258,
+      "learning_rate": 2.5185521003979746e-05,
+      "loss": 0.4699,
+      "step": 700
+    },
+    {
+      "epoch": 1.510182207931404,
+      "grad_norm": 1.2212399244308472,
+      "learning_rate": 2.5107281699174287e-05,
+      "loss": 0.4579,
+      "step": 705
+    },
+    {
+      "epoch": 1.5209003215434085,
+      "grad_norm": 1.080427646636963,
+      "learning_rate": 2.5028535462930924e-05,
+      "loss": 0.4509,
+      "step": 710
+    },
+    {
+      "epoch": 1.5316184351554125,
+      "grad_norm": 1.0146664381027222,
+      "learning_rate": 2.494928624477277e-05,
+      "loss": 0.5209,
+      "step": 715
+    },
+    {
+      "epoch": 1.542336548767417,
+      "grad_norm": 1.023618221282959,
+      "learning_rate": 2.4869538019450045e-05,
+      "loss": 0.459,
+      "step": 720
+    },
+    {
+      "epoch": 1.5530546623794212,
+      "grad_norm": 1.0760775804519653,
+      "learning_rate": 2.4789294786740705e-05,
+      "loss": 0.4522,
+      "step": 725
+    },
+    {
+      "epoch": 1.5637727759914255,
+      "grad_norm": 1.0976240634918213,
+      "learning_rate": 2.4708560571249845e-05,
+      "loss": 0.4853,
+      "step": 730
+    },
+    {
+      "epoch": 1.5744908896034298,
+      "grad_norm": 1.2561285495758057,
+      "learning_rate": 2.4627339422207805e-05,
+      "loss": 0.4584,
+      "step": 735
+    },
+    {
+      "epoch": 1.585209003215434,
+      "grad_norm": 1.1148028373718262,
+      "learning_rate": 2.454563541326716e-05,
+      "loss": 0.4492,
+      "step": 740
+    },
+    {
+      "epoch": 1.5959271168274385,
+      "grad_norm": 1.085665225982666,
+      "learning_rate": 2.4463452642298324e-05,
+      "loss": 0.4545,
+      "step": 745
+    },
+    {
+      "epoch": 1.6066452304394425,
+      "grad_norm": 0.9975435137748718,
+      "learning_rate": 2.438079523118406e-05,
+      "loss": 0.4392,
+      "step": 750
+    },
+    {
+      "epoch": 1.617363344051447,
+      "grad_norm": 1.1030360460281372,
+      "learning_rate": 2.4297667325612756e-05,
+      "loss": 0.4255,
+      "step": 755
+    },
+    {
+      "epoch": 1.6280814576634512,
+      "grad_norm": 1.02976655960083,
+      "learning_rate": 2.421407309487046e-05,
+      "loss": 0.4452,
+      "step": 760
+    },
+    {
+      "epoch": 1.6387995712754555,
+      "grad_norm": 1.0580244064331055,
+      "learning_rate": 2.4130016731631815e-05,
+      "loss": 0.4275,
+      "step": 765
+    },
+    {
+      "epoch": 1.6495176848874598,
+      "grad_norm": 1.1036502122879028,
+      "learning_rate": 2.4045502451749736e-05,
+      "loss": 0.4672,
+      "step": 770
+    },
+    {
+      "epoch": 1.660235798499464,
+      "grad_norm": 1.1784788370132446,
+      "learning_rate": 2.3960534494043993e-05,
+      "loss": 0.4036,
+      "step": 775
+    },
+    {
+      "epoch": 1.6709539121114685,
+      "grad_norm": 1.1079624891281128,
+      "learning_rate": 2.387511712008859e-05,
+      "loss": 0.4294,
+      "step": 780
+    },
+    {
+      "epoch": 1.6816720257234725,
+      "grad_norm": 1.0606523752212524,
+      "learning_rate": 2.3789254613998043e-05,
+      "loss": 0.449,
+      "step": 785
+    },
+    {
+      "epoch": 1.692390139335477,
+      "grad_norm": 0.9858099818229675,
+      "learning_rate": 2.3702951282212492e-05,
+      "loss": 0.3832,
+      "step": 790
+    },
+    {
+      "epoch": 1.7031082529474812,
+      "grad_norm": 1.1735355854034424,
+      "learning_rate": 2.3616211453281726e-05,
+      "loss": 0.445,
+      "step": 795
+    },
+    {
+      "epoch": 1.7138263665594855,
+      "grad_norm": 1.083713412284851,
+      "learning_rate": 2.3529039477648087e-05,
+      "loss": 0.4306,
+      "step": 800
+    },
+    {
+      "epoch": 1.72454448017149,
+      "grad_norm": 1.1281569004058838,
+      "learning_rate": 2.344143972742826e-05,
+      "loss": 0.3807,
+      "step": 805
+    },
+    {
+      "epoch": 1.735262593783494,
+      "grad_norm": 1.0552798509597778,
+      "learning_rate": 2.3353416596193985e-05,
+      "loss": 0.4013,
+      "step": 810
+    },
+    {
+      "epoch": 1.7459807073954985,
+      "grad_norm": 1.0200108289718628,
+      "learning_rate": 2.326497449875173e-05,
+      "loss": 0.369,
+      "step": 815
+    },
+    {
+      "epoch": 1.7566988210075025,
+      "grad_norm": 1.0658568143844604,
+      "learning_rate": 2.3176117870921218e-05,
+      "loss": 0.3814,
+      "step": 820
+    },
+    {
+      "epoch": 1.767416934619507,
+      "grad_norm": 1.164661169052124,
+      "learning_rate": 2.3086851169312992e-05,
+      "loss": 0.4044,
+      "step": 825
+    },
+    {
+      "epoch": 1.7781350482315113,
+      "grad_norm": 1.178032636642456,
+      "learning_rate": 2.299717887110487e-05,
+      "loss": 0.3985,
+      "step": 830
+    },
+    {
+      "epoch": 1.7888531618435155,
+      "grad_norm": 1.1956833600997925,
+      "learning_rate": 2.290710547381739e-05,
+      "loss": 0.3905,
+      "step": 835
+    },
+    {
+      "epoch": 1.79957127545552,
+      "grad_norm": 1.205641269683838,
+      "learning_rate": 2.2816635495088244e-05,
+      "loss": 0.3915,
+      "step": 840
+    },
+    {
+      "epoch": 1.810289389067524,
+      "grad_norm": 1.208314299583435,
+      "learning_rate": 2.272577347244571e-05,
+      "loss": 0.403,
+      "step": 845
+    },
+    {
+      "epoch": 1.8210075026795285,
+      "grad_norm": 1.1529418230056763,
+      "learning_rate": 2.2634523963081033e-05,
+      "loss": 0.3932,
+      "step": 850
+    },
+    {
+      "epoch": 1.8317256162915327,
+      "grad_norm": 1.1101915836334229,
+      "learning_rate": 2.25428915436199e-05,
+      "loss": 0.3763,
+      "step": 855
+    },
+    {
+      "epoch": 1.842443729903537,
+      "grad_norm": 1.071894884109497,
+      "learning_rate": 2.245088080989287e-05,
+      "loss": 0.3824,
+      "step": 860
+    },
+    {
+      "epoch": 1.8531618435155413,
+      "grad_norm": 1.1817514896392822,
+      "learning_rate": 2.2358496376704898e-05,
+      "loss": 0.4261,
+      "step": 865
+    },
+    {
+      "epoch": 1.8638799571275455,
+      "grad_norm": 1.1303304433822632,
+      "learning_rate": 2.2265742877603838e-05,
+      "loss": 0.3604,
+      "step": 870
+    },
+    {
+      "epoch": 1.87459807073955,
+      "grad_norm": 1.0623246431350708,
+      "learning_rate": 2.2172624964648094e-05,
+      "loss": 0.4049,
+      "step": 875
+    },
+    {
+      "epoch": 1.885316184351554,
+      "grad_norm": 1.171410322189331,
+      "learning_rate": 2.2079147308173258e-05,
+      "loss": 0.3633,
+      "step": 880
+    },
+    {
+      "epoch": 1.8960342979635585,
+      "grad_norm": 1.20453941822052,
+      "learning_rate": 2.1985314596557887e-05,
+      "loss": 0.4022,
+      "step": 885
+    },
+    {
+      "epoch": 1.9067524115755627,
+      "grad_norm": 1.171126365661621,
+      "learning_rate": 2.1891131535988364e-05,
+      "loss": 0.3967,
+      "step": 890
+    },
+    {
+      "epoch": 1.917470525187567,
+      "grad_norm": 1.1599109172821045,
+      "learning_rate": 2.1796602850222832e-05,
+      "loss": 0.349,
+      "step": 895
+    },
+    {
+      "epoch": 1.9281886387995713,
+      "grad_norm": 1.056682825088501,
+      "learning_rate": 2.170173328035431e-05,
+      "loss": 0.3705,
+      "step": 900
+    },
+    {
+      "epoch": 1.9389067524115755,
+      "grad_norm": 1.1990083456039429,
+      "learning_rate": 2.1606527584572854e-05,
+      "loss": 0.367,
+      "step": 905
+    },
+    {
+      "epoch": 1.94962486602358,
+      "grad_norm": 1.1245200634002686,
+      "learning_rate": 2.1510990537926983e-05,
+      "loss": 0.3734,
+      "step": 910
+    },
+    {
+      "epoch": 1.960342979635584,
+      "grad_norm": 1.2177960872650146,
+      "learning_rate": 2.1415126932084104e-05,
+      "loss": 0.4062,
+      "step": 915
+    },
+    {
+      "epoch": 1.9710610932475885,
+      "grad_norm": 1.1364030838012695,
+      "learning_rate": 2.1318941575090233e-05,
+      "loss": 0.3492,
+      "step": 920
+    },
+    {
+      "epoch": 1.9817792068595927,
+      "grad_norm": 1.1492586135864258,
+      "learning_rate": 2.1222439291128855e-05,
+      "loss": 0.3597,
+      "step": 925
+    },
+    {
+      "epoch": 1.992497320471597,
+      "grad_norm": 1.0529663562774658,
+      "learning_rate": 2.112562492027893e-05,
+      "loss": 0.3996,
+      "step": 930
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 2335,
+  "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": 1.394156316393472e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

debug_klora/1/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:627ff0db6f2b267b1535e0fea14759ea7e3b0dcb42df707143d60ceba8842b33
+size 7736

debug_klora/1/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)