Upload folder using huggingface_hub

41_128_e5_3e-5/checkpoint-1308/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

41_128_e5_3e-5/checkpoint-1308/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": [
+    "gate_proj",
+    "o_proj",
+    "up_proj",
+    "v_proj",
+    "k_proj",
+    "q_proj",
+    "down_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

41_128_e5_3e-5/checkpoint-1308/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f17caa56e5643d410b16ef7879ffa126151b162c081a8c618f0edea46ed3982b
+size 791751704

41_128_e5_3e-5/checkpoint-1308/latest

@@ -0,0 +1 @@
+global_step1308

41_128_e5_3e-5/checkpoint-1308/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:88671e5a035b58c662645fa0f1d2a8bf1243e06f61840862f5941f4913e99fd0
+size 15920

41_128_e5_3e-5/checkpoint-1308/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b0084a59d2e36dc9e964469fea5e1492ee2b4fc98c5f85f4b312e9fe8e7e5c1
+size 15920

41_128_e5_3e-5/checkpoint-1308/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7fd05527fc85b5b3926b9d54ec5ea2e62debc5fd064a0eb2b871d6d670cf7f39
+size 15920

41_128_e5_3e-5/checkpoint-1308/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c81cc0763c14ca0bde4a7fadc5806fcb1c1ac2fe861d7739dc4d7bf2b1bf8490
+size 15920

41_128_e5_3e-5/checkpoint-1308/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d32744f2a10d8edf3618be506028e52e52caff9648190c006d19ba2cf572a7fb
+size 15920

41_128_e5_3e-5/checkpoint-1308/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5ef0fed8e7684f91fbea4556b4bd8b7d2c43083e0a3a0ae5028b01f2de44a76c
+size 15920

41_128_e5_3e-5/checkpoint-1308/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a16f9c0d043689cff5151507e55e4af9133e21958ed6bd31d12241d12ede58d0
+size 15920

41_128_e5_3e-5/checkpoint-1308/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:336e8f107a8940bf610bfd620641f189e106b76b5889a73ff73390bac559e862
+size 15920

41_128_e5_3e-5/checkpoint-1308/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b6105a1bc67e112c65d157ba6418872c605da66d28cebd483ef2356ba5d950d
+size 1064

41_128_e5_3e-5/checkpoint-1308/special_tokens_map.json

@@ -0,0 +1,45 @@
+{
+  "additional_special_tokens": [
+    "<|endoftext|>",
+    "<fim_prefix>",
+    "<fim_middle>",
+    "<fim_suffix>",
+    "<fim_pad>",
+    "<filename>",
+    "<gh_stars>",
+    "<issue_start>",
+    "<issue_comment>",
+    "<issue_closed>",
+    "<jupyter_start>",
+    "<jupyter_text>",
+    "<jupyter_code>",
+    "<jupyter_output>",
+    "<empty_output>",
+    "<commit_before>",
+    "<commit_msg>",
+    "<commit_after>",
+    "<reponame>"
+  ],
+  "bos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "<reponame>",
+  "unk_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

41_128_e5_3e-5/checkpoint-1308/tokenizer_config.json

@@ -0,0 +1,188 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<fim_prefix>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<fim_middle>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<fim_suffix>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<fim_pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "5": {
+      "content": "<filename>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "6": {
+      "content": "<gh_stars>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "7": {
+      "content": "<issue_start>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "8": {
+      "content": "<issue_comment>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "9": {
+      "content": "<issue_closed>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "10": {
+      "content": "<jupyter_start>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "11": {
+      "content": "<jupyter_text>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "12": {
+      "content": "<jupyter_code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "13": {
+      "content": "<jupyter_output>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "14": {
+      "content": "<empty_output>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "15": {
+      "content": "<commit_before>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "16": {
+      "content": "<commit_msg>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "17": {
+      "content": "<commit_after>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "18": {
+      "content": "<reponame>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<|endoftext|>",
+    "<fim_prefix>",
+    "<fim_middle>",
+    "<fim_suffix>",
+    "<fim_pad>",
+    "<filename>",
+    "<gh_stars>",
+    "<issue_start>",
+    "<issue_comment>",
+    "<issue_closed>",
+    "<jupyter_start>",
+    "<jupyter_text>",
+    "<jupyter_code>",
+    "<jupyter_output>",
+    "<empty_output>",
+    "<commit_before>",
+    "<commit_msg>",
+    "<commit_after>",
+    "<reponame>"
+  ],
+  "bos_token": "<|endoftext|>",
+  "clean_up_tokenization_spaces": true,
+  "eos_token": "<|endoftext|>",
+  "extra_special_tokens": {},
+  "model_max_length": 8192,
+  "pad_token": "<reponame>",
+  "padding_side": "left",
+  "tokenizer_class": "GPT2Tokenizer",
+  "unk_token": "<|endoftext|>",
+  "vocab_size": 49152
+}

41_128_e5_3e-5/checkpoint-1308/trainer_state.json

@@ -0,0 +1,1861 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 1308,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.015313935681470138,
+      "grad_norm": 1.1070387363433838,
+      "learning_rate": 1.4634146341463416e-06,
+      "loss": 1.2917,
+      "step": 5
+    },
+    {
+      "epoch": 0.030627871362940276,
+      "grad_norm": 1.122970700263977,
+      "learning_rate": 3.2926829268292685e-06,
+      "loss": 1.2965,
+      "step": 10
+    },
+    {
+      "epoch": 0.045941807044410414,
+      "grad_norm": 0.8600009083747864,
+      "learning_rate": 5.121951219512195e-06,
+      "loss": 1.3084,
+      "step": 15
+    },
+    {
+      "epoch": 0.06125574272588055,
+      "grad_norm": 0.6501789093017578,
+      "learning_rate": 6.951219512195123e-06,
+      "loss": 1.3178,
+      "step": 20
+    },
+    {
+      "epoch": 0.07656967840735068,
+      "grad_norm": 0.6199907660484314,
+      "learning_rate": 8.780487804878048e-06,
+      "loss": 1.258,
+      "step": 25
+    },
+    {
+      "epoch": 0.09188361408882083,
+      "grad_norm": 0.525578498840332,
+      "learning_rate": 1.0609756097560975e-05,
+      "loss": 1.2632,
+      "step": 30
+    },
+    {
+      "epoch": 0.10719754977029096,
+      "grad_norm": 0.5301306843757629,
+      "learning_rate": 1.2439024390243903e-05,
+      "loss": 1.1863,
+      "step": 35
+    },
+    {
+      "epoch": 0.1225114854517611,
+      "grad_norm": 0.45140478014945984,
+      "learning_rate": 1.4268292682926829e-05,
+      "loss": 1.1721,
+      "step": 40
+    },
+    {
+      "epoch": 0.13782542113323124,
+      "grad_norm": 0.5104884505271912,
+      "learning_rate": 1.6097560975609757e-05,
+      "loss": 1.2507,
+      "step": 45
+    },
+    {
+      "epoch": 0.15313935681470137,
+      "grad_norm": 0.4204311668872833,
+      "learning_rate": 1.7926829268292684e-05,
+      "loss": 1.1973,
+      "step": 50
+    },
+    {
+      "epoch": 0.16845329249617153,
+      "grad_norm": 0.43958738446235657,
+      "learning_rate": 1.975609756097561e-05,
+      "loss": 1.1404,
+      "step": 55
+    },
+    {
+      "epoch": 0.18376722817764166,
+      "grad_norm": 0.45204630494117737,
+      "learning_rate": 2.1585365853658537e-05,
+      "loss": 1.1811,
+      "step": 60
+    },
+    {
+      "epoch": 0.1990811638591118,
+      "grad_norm": 0.4730817973613739,
+      "learning_rate": 2.3414634146341466e-05,
+      "loss": 1.1368,
+      "step": 65
+    },
+    {
+      "epoch": 0.21439509954058192,
+      "grad_norm": 0.5665387511253357,
+      "learning_rate": 2.524390243902439e-05,
+      "loss": 1.1341,
+      "step": 70
+    },
+    {
+      "epoch": 0.22970903522205208,
+      "grad_norm": 0.49841874837875366,
+      "learning_rate": 2.707317073170732e-05,
+      "loss": 1.1277,
+      "step": 75
+    },
+    {
+      "epoch": 0.2450229709035222,
+      "grad_norm": 0.4274734556674957,
+      "learning_rate": 2.8902439024390242e-05,
+      "loss": 1.0674,
+      "step": 80
+    },
+    {
+      "epoch": 0.26033690658499237,
+      "grad_norm": 0.6278162598609924,
+      "learning_rate": 2.9999877234172298e-05,
+      "loss": 1.0953,
+      "step": 85
+    },
+    {
+      "epoch": 0.27565084226646247,
+      "grad_norm": 0.5454415082931519,
+      "learning_rate": 2.999849614168868e-05,
+      "loss": 1.0974,
+      "step": 90
+    },
+    {
+      "epoch": 0.29096477794793263,
+      "grad_norm": 0.5087949633598328,
+      "learning_rate": 2.999558064119925e-05,
+      "loss": 1.1079,
+      "step": 95
+    },
+    {
+      "epoch": 0.30627871362940273,
+      "grad_norm": 0.5586227774620056,
+      "learning_rate": 2.999113103097174e-05,
+      "loss": 1.0655,
+      "step": 100
+    },
+    {
+      "epoch": 0.3215926493108729,
+      "grad_norm": 0.5441133975982666,
+      "learning_rate": 2.9985147766219615e-05,
+      "loss": 1.0631,
+      "step": 105
+    },
+    {
+      "epoch": 0.33690658499234305,
+      "grad_norm": 0.6769384741783142,
+      "learning_rate": 2.9977631459055537e-05,
+      "loss": 1.0438,
+      "step": 110
+    },
+    {
+      "epoch": 0.35222052067381315,
+      "grad_norm": 0.523868978023529,
+      "learning_rate": 2.996858287842873e-05,
+      "loss": 1.0131,
+      "step": 115
+    },
+    {
+      "epoch": 0.3675344563552833,
+      "grad_norm": 0.5572935342788696,
+      "learning_rate": 2.995800295004629e-05,
+      "loss": 1.0501,
+      "step": 120
+    },
+    {
+      "epoch": 0.38284839203675347,
+      "grad_norm": 0.6209707856178284,
+      "learning_rate": 2.9945892756278543e-05,
+      "loss": 1.0125,
+      "step": 125
+    },
+    {
+      "epoch": 0.3981623277182236,
+      "grad_norm": 0.6483351588249207,
+      "learning_rate": 2.993225353604824e-05,
+      "loss": 0.998,
+      "step": 130
+    },
+    {
+      "epoch": 0.41347626339969373,
+      "grad_norm": 0.6352364420890808,
+      "learning_rate": 2.9917086684703844e-05,
+      "loss": 1.0163,
+      "step": 135
+    },
+    {
+      "epoch": 0.42879019908116384,
+      "grad_norm": 0.6867502331733704,
+      "learning_rate": 2.9900393753876816e-05,
+      "loss": 0.9824,
+      "step": 140
+    },
+    {
+      "epoch": 0.444104134762634,
+      "grad_norm": 0.6153353452682495,
+      "learning_rate": 2.9882176451322798e-05,
+      "loss": 0.9395,
+      "step": 145
+    },
+    {
+      "epoch": 0.45941807044410415,
+      "grad_norm": 0.645903468132019,
+      "learning_rate": 2.9862436640746973e-05,
+      "loss": 0.914,
+      "step": 150
+    },
+    {
+      "epoch": 0.47473200612557426,
+      "grad_norm": 0.671872615814209,
+      "learning_rate": 2.9841176341613364e-05,
+      "loss": 0.9177,
+      "step": 155
+    },
+    {
+      "epoch": 0.4900459418070444,
+      "grad_norm": 0.6303296089172363,
+      "learning_rate": 2.981839772893825e-05,
+      "loss": 0.9576,
+      "step": 160
+    },
+    {
+      "epoch": 0.5053598774885145,
+      "grad_norm": 0.7237289547920227,
+      "learning_rate": 2.9794103133067637e-05,
+      "loss": 0.9649,
+      "step": 165
+    },
+    {
+      "epoch": 0.5206738131699847,
+      "grad_norm": 0.6527206301689148,
+      "learning_rate": 2.9768295039438868e-05,
+      "loss": 0.9158,
+      "step": 170
+    },
+    {
+      "epoch": 0.5359877488514548,
+      "grad_norm": 0.7285829782485962,
+      "learning_rate": 2.974097608832635e-05,
+      "loss": 0.9173,
+      "step": 175
+    },
+    {
+      "epoch": 0.5513016845329249,
+      "grad_norm": 0.8059846758842468,
+      "learning_rate": 2.9712149074571433e-05,
+      "loss": 0.8667,
+      "step": 180
+    },
+    {
+      "epoch": 0.5666156202143952,
+      "grad_norm": 0.7382646799087524,
+      "learning_rate": 2.968181694729651e-05,
+      "loss": 0.877,
+      "step": 185
+    },
+    {
+      "epoch": 0.5819295558958653,
+      "grad_norm": 0.7463013529777527,
+      "learning_rate": 2.964998280960328e-05,
+      "loss": 0.9148,
+      "step": 190
+    },
+    {
+      "epoch": 0.5972434915773354,
+      "grad_norm": 0.7870151400566101,
+      "learning_rate": 2.961664991825531e-05,
+      "loss": 0.8881,
+      "step": 195
+    },
+    {
+      "epoch": 0.6125574272588055,
+      "grad_norm": 0.8590981364250183,
+      "learning_rate": 2.9581821683344832e-05,
+      "loss": 0.8352,
+      "step": 200
+    },
+    {
+      "epoch": 0.6278713629402757,
+      "grad_norm": 0.7620300650596619,
+      "learning_rate": 2.954550166794391e-05,
+      "loss": 0.7867,
+      "step": 205
+    },
+    {
+      "epoch": 0.6431852986217458,
+      "grad_norm": 0.8273492455482483,
+      "learning_rate": 2.9507693587739895e-05,
+      "loss": 0.8456,
+      "step": 210
+    },
+    {
+      "epoch": 0.6584992343032159,
+      "grad_norm": 0.7399709224700928,
+      "learning_rate": 2.9468401310655303e-05,
+      "loss": 0.8156,
+      "step": 215
+    },
+    {
+      "epoch": 0.6738131699846861,
+      "grad_norm": 0.851788341999054,
+      "learning_rate": 2.942762885645211e-05,
+      "loss": 0.7952,
+      "step": 220
+    },
+    {
+      "epoch": 0.6891271056661562,
+      "grad_norm": 0.8381015658378601,
+      "learning_rate": 2.9385380396320523e-05,
+      "loss": 0.8027,
+      "step": 225
+    },
+    {
+      "epoch": 0.7044410413476263,
+      "grad_norm": 0.8373646140098572,
+      "learning_rate": 2.934166025245223e-05,
+      "loss": 0.7943,
+      "step": 230
+    },
+    {
+      "epoch": 0.7197549770290965,
+      "grad_norm": 0.8320114612579346,
+      "learning_rate": 2.9296472897598246e-05,
+      "loss": 0.8064,
+      "step": 235
+    },
+    {
+      "epoch": 0.7350689127105666,
+      "grad_norm": 1.0138018131256104,
+      "learning_rate": 2.924982295461131e-05,
+      "loss": 0.761,
+      "step": 240
+    },
+    {
+      "epoch": 0.7503828483920367,
+      "grad_norm": 0.9759603142738342,
+      "learning_rate": 2.920171519597297e-05,
+      "loss": 0.7114,
+      "step": 245
+    },
+    {
+      "epoch": 0.7656967840735069,
+      "grad_norm": 0.844030499458313,
+      "learning_rate": 2.9152154543305316e-05,
+      "loss": 0.7461,
+      "step": 250
+    },
+    {
+      "epoch": 0.781010719754977,
+      "grad_norm": 0.8687477111816406,
+      "learning_rate": 2.9101146066867502e-05,
+      "loss": 0.728,
+      "step": 255
+    },
+    {
+      "epoch": 0.7963246554364471,
+      "grad_norm": 0.9759243726730347,
+      "learning_rate": 2.9048694985037e-05,
+      "loss": 0.7475,
+      "step": 260
+    },
+    {
+      "epoch": 0.8116385911179173,
+      "grad_norm": 0.9230566620826721,
+      "learning_rate": 2.8994806663775792e-05,
+      "loss": 0.8098,
+      "step": 265
+    },
+    {
+      "epoch": 0.8269525267993875,
+      "grad_norm": 0.8410807251930237,
+      "learning_rate": 2.893948661608136e-05,
+      "loss": 0.7465,
+      "step": 270
+    },
+    {
+      "epoch": 0.8422664624808576,
+      "grad_norm": 1.0104124546051025,
+      "learning_rate": 2.888274050142271e-05,
+      "loss": 0.7799,
+      "step": 275
+    },
+    {
+      "epoch": 0.8575803981623277,
+      "grad_norm": 0.9490285515785217,
+      "learning_rate": 2.8824574125161384e-05,
+      "loss": 0.7541,
+      "step": 280
+    },
+    {
+      "epoch": 0.8728943338437979,
+      "grad_norm": 0.9112844467163086,
+      "learning_rate": 2.876499343795754e-05,
+      "loss": 0.7424,
+      "step": 285
+    },
+    {
+      "epoch": 0.888208269525268,
+      "grad_norm": 0.8824840784072876,
+      "learning_rate": 2.8704004535161172e-05,
+      "loss": 0.7448,
+      "step": 290
+    },
+    {
+      "epoch": 0.9035222052067381,
+      "grad_norm": 1.3259707689285278,
+      "learning_rate": 2.864161365618854e-05,
+      "loss": 0.7165,
+      "step": 295
+    },
+    {
+      "epoch": 0.9188361408882083,
+      "grad_norm": 0.938458263874054,
+      "learning_rate": 2.8577827183883853e-05,
+      "loss": 0.6724,
+      "step": 300
+    },
+    {
+      "epoch": 0.9341500765696784,
+      "grad_norm": 0.940185010433197,
+      "learning_rate": 2.851265164386627e-05,
+      "loss": 0.6722,
+      "step": 305
+    },
+    {
+      "epoch": 0.9494640122511485,
+      "grad_norm": 1.0995936393737793,
+      "learning_rate": 2.84460937038623e-05,
+      "loss": 0.6628,
+      "step": 310
+    },
+    {
+      "epoch": 0.9647779479326187,
+      "grad_norm": 0.9380448460578918,
+      "learning_rate": 2.837816017302368e-05,
+      "loss": 0.6519,
+      "step": 315
+    },
+    {
+      "epoch": 0.9800918836140888,
+      "grad_norm": 1.0398368835449219,
+      "learning_rate": 2.8308858001230757e-05,
+      "loss": 0.6781,
+      "step": 320
+    },
+    {
+      "epoch": 0.9954058192955589,
+      "grad_norm": 1.1420120000839233,
+      "learning_rate": 2.823819427838149e-05,
+      "loss": 0.6554,
+      "step": 325
+    },
+    {
+      "epoch": 1.0091883614088821,
+      "grad_norm": 0.9629755020141602,
+      "learning_rate": 2.8166176233666137e-05,
+      "loss": 0.6067,
+      "step": 330
+    },
+    {
+      "epoch": 1.0245022970903521,
+      "grad_norm": 0.9918197393417358,
+      "learning_rate": 2.8092811234827643e-05,
+      "loss": 0.5579,
+      "step": 335
+    },
+    {
+      "epoch": 1.0398162327718223,
+      "grad_norm": 1.004963994026184,
+      "learning_rate": 2.8018106787407942e-05,
+      "loss": 0.5962,
+      "step": 340
+    },
+    {
+      "epoch": 1.0551301684532925,
+      "grad_norm": 1.010399580001831,
+      "learning_rate": 2.7942070533980072e-05,
+      "loss": 0.5431,
+      "step": 345
+    },
+    {
+      "epoch": 1.0704441041347625,
+      "grad_norm": 1.1038013696670532,
+      "learning_rate": 2.7864710253366303e-05,
+      "loss": 0.5716,
+      "step": 350
+    },
+    {
+      "epoch": 1.0857580398162328,
+      "grad_norm": 1.2030080556869507,
+      "learning_rate": 2.778603385984237e-05,
+      "loss": 0.5613,
+      "step": 355
+    },
+    {
+      "epoch": 1.101071975497703,
+      "grad_norm": 1.0458992719650269,
+      "learning_rate": 2.7706049402327768e-05,
+      "loss": 0.5448,
+      "step": 360
+    },
+    {
+      "epoch": 1.116385911179173,
+      "grad_norm": 1.082198143005371,
+      "learning_rate": 2.762476506356235e-05,
+      "loss": 0.5488,
+      "step": 365
+    },
+    {
+      "epoch": 1.1316998468606432,
+      "grad_norm": 1.0959954261779785,
+      "learning_rate": 2.7542189159269155e-05,
+      "loss": 0.5366,
+      "step": 370
+    },
+    {
+      "epoch": 1.1470137825421134,
+      "grad_norm": 1.0527937412261963,
+      "learning_rate": 2.745833013730373e-05,
+      "loss": 0.53,
+      "step": 375
+    },
+    {
+      "epoch": 1.1623277182235834,
+      "grad_norm": 1.1240211725234985,
+      "learning_rate": 2.737319657678983e-05,
+      "loss": 0.5106,
+      "step": 380
+    },
+    {
+      "epoch": 1.1776416539050536,
+      "grad_norm": 1.073337197303772,
+      "learning_rate": 2.7286797187241772e-05,
+      "loss": 0.5241,
+      "step": 385
+    },
+    {
+      "epoch": 1.1929555895865238,
+      "grad_norm": 1.1111088991165161,
+      "learning_rate": 2.7199140807673398e-05,
+      "loss": 0.5168,
+      "step": 390
+    },
+    {
+      "epoch": 1.2082695252679938,
+      "grad_norm": 1.1780794858932495,
+      "learning_rate": 2.711023640569381e-05,
+      "loss": 0.5503,
+      "step": 395
+    },
+    {
+      "epoch": 1.223583460949464,
+      "grad_norm": 1.138046145439148,
+      "learning_rate": 2.702009307658996e-05,
+      "loss": 0.5197,
+      "step": 400
+    },
+    {
+      "epoch": 1.2388973966309342,
+      "grad_norm": 1.1711301803588867,
+      "learning_rate": 2.692872004239614e-05,
+      "loss": 0.5512,
+      "step": 405
+    },
+    {
+      "epoch": 1.2542113323124042,
+      "grad_norm": 1.0399030447006226,
+      "learning_rate": 2.6836126650950556e-05,
+      "loss": 0.5246,
+      "step": 410
+    },
+    {
+      "epoch": 1.2695252679938744,
+      "grad_norm": 1.0978562831878662,
+      "learning_rate": 2.6742322374938985e-05,
+      "loss": 0.5186,
+      "step": 415
+    },
+    {
+      "epoch": 1.2848392036753444,
+      "grad_norm": 1.1471067667007446,
+      "learning_rate": 2.664731681092569e-05,
+      "loss": 0.5185,
+      "step": 420
+    },
+    {
+      "epoch": 1.3001531393568146,
+      "grad_norm": 1.3175466060638428,
+      "learning_rate": 2.6551119678371655e-05,
+      "loss": 0.4764,
+      "step": 425
+    },
+    {
+      "epoch": 1.3154670750382849,
+      "grad_norm": 0.9900484681129456,
+      "learning_rate": 2.645374081864023e-05,
+      "loss": 0.5133,
+      "step": 430
+    },
+    {
+      "epoch": 1.3307810107197549,
+      "grad_norm": 1.1252256631851196,
+      "learning_rate": 2.635519019399034e-05,
+      "loss": 0.5122,
+      "step": 435
+    },
+    {
+      "epoch": 1.346094946401225,
+      "grad_norm": 1.205263376235962,
+      "learning_rate": 2.6255477886557287e-05,
+      "loss": 0.5093,
+      "step": 440
+    },
+    {
+      "epoch": 1.3614088820826953,
+      "grad_norm": 1.0410200357437134,
+      "learning_rate": 2.615461409732131e-05,
+      "loss": 0.4944,
+      "step": 445
+    },
+    {
+      "epoch": 1.3767228177641653,
+      "grad_norm": 1.0574573278427124,
+      "learning_rate": 2.6052609145064e-05,
+      "loss": 0.4646,
+      "step": 450
+    },
+    {
+      "epoch": 1.3920367534456355,
+      "grad_norm": 1.2100945711135864,
+      "learning_rate": 2.5949473465312602e-05,
+      "loss": 0.5132,
+      "step": 455
+    },
+    {
+      "epoch": 1.4073506891271057,
+      "grad_norm": 1.1366932392120361,
+      "learning_rate": 2.5845217609272474e-05,
+      "loss": 0.5026,
+      "step": 460
+    },
+    {
+      "epoch": 1.4226646248085757,
+      "grad_norm": 1.147862195968628,
+      "learning_rate": 2.5739852242747614e-05,
+      "loss": 0.4918,
+      "step": 465
+    },
+    {
+      "epoch": 1.437978560490046,
+      "grad_norm": 1.06094229221344,
+      "learning_rate": 2.5633388145049525e-05,
+      "loss": 0.4868,
+      "step": 470
+    },
+    {
+      "epoch": 1.4532924961715161,
+      "grad_norm": 1.0694596767425537,
+      "learning_rate": 2.552583620789443e-05,
+      "loss": 0.4744,
+      "step": 475
+    },
+    {
+      "epoch": 1.4686064318529861,
+      "grad_norm": 0.9866558909416199,
+      "learning_rate": 2.5417207434289022e-05,
+      "loss": 0.4992,
+      "step": 480
+    },
+    {
+      "epoch": 1.4839203675344563,
+      "grad_norm": 1.0969899892807007,
+      "learning_rate": 2.5307512937404804e-05,
+      "loss": 0.4718,
+      "step": 485
+    },
+    {
+      "epoch": 1.4992343032159265,
+      "grad_norm": 1.1575231552124023,
+      "learning_rate": 2.5196763939441163e-05,
+      "loss": 0.4851,
+      "step": 490
+    },
+    {
+      "epoch": 1.5145482388973965,
+      "grad_norm": 1.0528773069381714,
+      "learning_rate": 2.5084971770477295e-05,
+      "loss": 0.446,
+      "step": 495
+    },
+    {
+      "epoch": 1.5298621745788668,
+      "grad_norm": 1.1558538675308228,
+      "learning_rate": 2.4972147867313094e-05,
+      "loss": 0.4442,
+      "step": 500
+    },
+    {
+      "epoch": 1.545176110260337,
+      "grad_norm": 1.1066346168518066,
+      "learning_rate": 2.4858303772299115e-05,
+      "loss": 0.4707,
+      "step": 505
+    },
+    {
+      "epoch": 1.560490045941807,
+      "grad_norm": 1.1572457551956177,
+      "learning_rate": 2.4743451132155738e-05,
+      "loss": 0.4463,
+      "step": 510
+    },
+    {
+      "epoch": 1.5758039816232772,
+      "grad_norm": 1.182366132736206,
+      "learning_rate": 2.4627601696781687e-05,
+      "loss": 0.3947,
+      "step": 515
+    },
+    {
+      "epoch": 1.5911179173047474,
+      "grad_norm": 1.0030242204666138,
+      "learning_rate": 2.451076731805193e-05,
+      "loss": 0.4203,
+      "step": 520
+    },
+    {
+      "epoch": 1.6064318529862174,
+      "grad_norm": 1.2075709104537964,
+      "learning_rate": 2.4392959948605214e-05,
+      "loss": 0.4045,
+      "step": 525
+    },
+    {
+      "epoch": 1.6217457886676876,
+      "grad_norm": 1.3273980617523193,
+      "learning_rate": 2.427419164062125e-05,
+      "loss": 0.4074,
+      "step": 530
+    },
+    {
+      "epoch": 1.6370597243491578,
+      "grad_norm": 1.0939041376113892,
+      "learning_rate": 2.415447454458771e-05,
+      "loss": 0.4169,
+      "step": 535
+    },
+    {
+      "epoch": 1.6523736600306278,
+      "grad_norm": 1.3279342651367188,
+      "learning_rate": 2.403382090805721e-05,
+      "loss": 0.4651,
+      "step": 540
+    },
+    {
+      "epoch": 1.667687595712098,
+      "grad_norm": 1.0995126962661743,
+      "learning_rate": 2.391224307439432e-05,
+      "loss": 0.3933,
+      "step": 545
+    },
+    {
+      "epoch": 1.6830015313935682,
+      "grad_norm": 1.221017599105835,
+      "learning_rate": 2.3789753481512777e-05,
+      "loss": 0.4159,
+      "step": 550
+    },
+    {
+      "epoch": 1.6983154670750382,
+      "grad_norm": 1.2090656757354736,
+      "learning_rate": 2.3666364660603058e-05,
+      "loss": 0.4062,
+      "step": 555
+    },
+    {
+      "epoch": 1.7136294027565084,
+      "grad_norm": 1.1304739713668823,
+      "learning_rate": 2.354208923485037e-05,
+      "loss": 0.4137,
+      "step": 560
+    },
+    {
+      "epoch": 1.7289433384379786,
+      "grad_norm": 1.0823583602905273,
+      "learning_rate": 2.341693991814325e-05,
+      "loss": 0.3721,
+      "step": 565
+    },
+    {
+      "epoch": 1.7442572741194486,
+      "grad_norm": 1.3484351634979248,
+      "learning_rate": 2.329092951377289e-05,
+      "loss": 0.3905,
+      "step": 570
+    },
+    {
+      "epoch": 1.7595712098009189,
+      "grad_norm": 1.2623238563537598,
+      "learning_rate": 2.3164070913123288e-05,
+      "loss": 0.3941,
+      "step": 575
+    },
+    {
+      "epoch": 1.774885145482389,
+      "grad_norm": 1.096492052078247,
+      "learning_rate": 2.3036377094352432e-05,
+      "loss": 0.3596,
+      "step": 580
+    },
+    {
+      "epoch": 1.790199081163859,
+      "grad_norm": 1.1170555353164673,
+      "learning_rate": 2.2907861121064567e-05,
+      "loss": 0.3752,
+      "step": 585
+    },
+    {
+      "epoch": 1.8055130168453293,
+      "grad_norm": 1.3183743953704834,
+      "learning_rate": 2.277853614097373e-05,
+      "loss": 0.366,
+      "step": 590
+    },
+    {
+      "epoch": 1.8208269525267995,
+      "grad_norm": 1.2602739334106445,
+      "learning_rate": 2.264841538455869e-05,
+      "loss": 0.386,
+      "step": 595
+    },
+    {
+      "epoch": 1.8361408882082695,
+      "grad_norm": 1.532475471496582,
+      "learning_rate": 2.2517512163709417e-05,
+      "loss": 0.3645,
+      "step": 600
+    },
+    {
+      "epoch": 1.8514548238897397,
+      "grad_norm": 1.1312321424484253,
+      "learning_rate": 2.2385839870365227e-05,
+      "loss": 0.3842,
+      "step": 605
+    },
+    {
+      "epoch": 1.86676875957121,
+      "grad_norm": 1.1037025451660156,
+      "learning_rate": 2.225341197514471e-05,
+      "loss": 0.3528,
+      "step": 610
+    },
+    {
+      "epoch": 1.88208269525268,
+      "grad_norm": 1.1164988279342651,
+      "learning_rate": 2.2120242025967652e-05,
+      "loss": 0.3808,
+      "step": 615
+    },
+    {
+      "epoch": 1.8973966309341501,
+      "grad_norm": 1.316138744354248,
+      "learning_rate": 2.198634364666902e-05,
+      "loss": 0.3882,
+      "step": 620
+    },
+    {
+      "epoch": 1.9127105666156203,
+      "grad_norm": 1.229442834854126,
+      "learning_rate": 2.1851730535605182e-05,
+      "loss": 0.3373,
+      "step": 625
+    },
+    {
+      "epoch": 1.9280245022970903,
+      "grad_norm": 1.0578089952468872,
+      "learning_rate": 2.1716416464252515e-05,
+      "loss": 0.3288,
+      "step": 630
+    },
+    {
+      "epoch": 1.9433384379785605,
+      "grad_norm": 1.1873011589050293,
+      "learning_rate": 2.1580415275798524e-05,
+      "loss": 0.3632,
+      "step": 635
+    },
+    {
+      "epoch": 1.9586523736600308,
+      "grad_norm": 1.2310603857040405,
+      "learning_rate": 2.1443740883725627e-05,
+      "loss": 0.3262,
+      "step": 640
+    },
+    {
+      "epoch": 1.9739663093415007,
+      "grad_norm": 1.2483208179473877,
+      "learning_rate": 2.1306407270387746e-05,
+      "loss": 0.3208,
+      "step": 645
+    },
+    {
+      "epoch": 1.9892802450229707,
+      "grad_norm": 1.269856333732605,
+      "learning_rate": 2.1168428485579864e-05,
+      "loss": 0.3478,
+      "step": 650
+    },
+    {
+      "epoch": 2.003062787136294,
+      "grad_norm": 1.0677129030227661,
+      "learning_rate": 2.1029818645100658e-05,
+      "loss": 0.3044,
+      "step": 655
+    },
+    {
+      "epoch": 2.0183767228177643,
+      "grad_norm": 1.2321646213531494,
+      "learning_rate": 2.0890591929308407e-05,
+      "loss": 0.2795,
+      "step": 660
+    },
+    {
+      "epoch": 2.0336906584992342,
+      "grad_norm": 1.1085941791534424,
+      "learning_rate": 2.075076258167028e-05,
+      "loss": 0.2761,
+      "step": 665
+    },
+    {
+      "epoch": 2.0490045941807042,
+      "grad_norm": 1.2009588479995728,
+      "learning_rate": 2.0610344907305154e-05,
+      "loss": 0.2827,
+      "step": 670
+    },
+    {
+      "epoch": 2.0643185298621747,
+      "grad_norm": 1.0571602582931519,
+      "learning_rate": 2.046935327152018e-05,
+      "loss": 0.2704,
+      "step": 675
+    },
+    {
+      "epoch": 2.0796324655436447,
+      "grad_norm": 1.3781317472457886,
+      "learning_rate": 2.0327802098341116e-05,
+      "loss": 0.2589,
+      "step": 680
+    },
+    {
+      "epoch": 2.0949464012251147,
+      "grad_norm": 1.2654929161071777,
+      "learning_rate": 2.0185705869036704e-05,
+      "loss": 0.258,
+      "step": 685
+    },
+    {
+      "epoch": 2.110260336906585,
+      "grad_norm": 0.9942528605461121,
+      "learning_rate": 2.004307912063717e-05,
+      "loss": 0.2757,
+      "step": 690
+    },
+    {
+      "epoch": 2.125574272588055,
+      "grad_norm": 1.2382903099060059,
+      "learning_rate": 1.989993644444704e-05,
+      "loss": 0.3019,
+      "step": 695
+    },
+    {
+      "epoch": 2.140888208269525,
+      "grad_norm": 1.0553951263427734,
+      "learning_rate": 1.975629248455238e-05,
+      "loss": 0.2578,
+      "step": 700
+    },
+    {
+      "epoch": 2.1562021439509955,
+      "grad_norm": 1.1529054641723633,
+      "learning_rate": 1.9612161936322642e-05,
+      "loss": 0.2792,
+      "step": 705
+    },
+    {
+      "epoch": 2.1715160796324655,
+      "grad_norm": 1.3361412286758423,
+      "learning_rate": 1.9467559544907273e-05,
+      "loss": 0.2516,
+      "step": 710
+    },
+    {
+      "epoch": 2.1868300153139355,
+      "grad_norm": 1.1726348400115967,
+      "learning_rate": 1.932250010372722e-05,
+      "loss": 0.2461,
+      "step": 715
+    },
+    {
+      "epoch": 2.202143950995406,
+      "grad_norm": 1.272294521331787,
+      "learning_rate": 1.9176998452961515e-05,
+      "loss": 0.2612,
+      "step": 720
+    },
+    {
+      "epoch": 2.217457886676876,
+      "grad_norm": 1.0995804071426392,
+      "learning_rate": 1.9031069478029044e-05,
+      "loss": 0.2729,
+      "step": 725
+    },
+    {
+      "epoch": 2.232771822358346,
+      "grad_norm": 1.2195067405700684,
+      "learning_rate": 1.8884728108065723e-05,
+      "loss": 0.2659,
+      "step": 730
+    },
+    {
+      "epoch": 2.2480857580398164,
+      "grad_norm": 1.1243830919265747,
+      "learning_rate": 1.873798931439718e-05,
+      "loss": 0.2736,
+      "step": 735
+    },
+    {
+      "epoch": 2.2633996937212864,
+      "grad_norm": 1.2348970174789429,
+      "learning_rate": 1.8590868109007117e-05,
+      "loss": 0.251,
+      "step": 740
+    },
+    {
+      "epoch": 2.2787136294027563,
+      "grad_norm": 1.087607502937317,
+      "learning_rate": 1.8443379543001556e-05,
+      "loss": 0.2575,
+      "step": 745
+    },
+    {
+      "epoch": 2.294027565084227,
+      "grad_norm": 1.3117108345031738,
+      "learning_rate": 1.8295538705069e-05,
+      "loss": 0.2613,
+      "step": 750
+    },
+    {
+      "epoch": 2.3093415007656968,
+      "grad_norm": 1.1849708557128906,
+      "learning_rate": 1.8147360719936835e-05,
+      "loss": 0.223,
+      "step": 755
+    },
+    {
+      "epoch": 2.3246554364471668,
+      "grad_norm": 1.2858647108078003,
+      "learning_rate": 1.7998860746823995e-05,
+      "loss": 0.2601,
+      "step": 760
+    },
+    {
+      "epoch": 2.339969372128637,
+      "grad_norm": 1.0667446851730347,
+      "learning_rate": 1.7850053977890114e-05,
+      "loss": 0.2543,
+      "step": 765
+    },
+    {
+      "epoch": 2.355283307810107,
+      "grad_norm": 1.1253725290298462,
+      "learning_rate": 1.770095563668129e-05,
+      "loss": 0.219,
+      "step": 770
+    },
+    {
+      "epoch": 2.370597243491577,
+      "grad_norm": 1.078885555267334,
+      "learning_rate": 1.7551580976572686e-05,
+      "loss": 0.2468,
+      "step": 775
+    },
+    {
+      "epoch": 2.3859111791730476,
+      "grad_norm": 1.2456717491149902,
+      "learning_rate": 1.740194527920802e-05,
+      "loss": 0.2257,
+      "step": 780
+    },
+    {
+      "epoch": 2.4012251148545176,
+      "grad_norm": 1.4209325313568115,
+      "learning_rate": 1.7252063852936187e-05,
+      "loss": 0.2296,
+      "step": 785
+    },
+    {
+      "epoch": 2.4165390505359876,
+      "grad_norm": 0.9507177472114563,
+      "learning_rate": 1.7101952031245187e-05,
+      "loss": 0.2168,
+      "step": 790
+    },
+    {
+      "epoch": 2.431852986217458,
+      "grad_norm": 1.307239294052124,
+      "learning_rate": 1.69516251711934e-05,
+      "loss": 0.2191,
+      "step": 795
+    },
+    {
+      "epoch": 2.447166921898928,
+      "grad_norm": 1.2544313669204712,
+      "learning_rate": 1.6801098651838523e-05,
+      "loss": 0.2272,
+      "step": 800
+    },
+    {
+      "epoch": 2.462480857580398,
+      "grad_norm": 1.2705137729644775,
+      "learning_rate": 1.665038787266423e-05,
+      "loss": 0.2415,
+      "step": 805
+    },
+    {
+      "epoch": 2.4777947932618685,
+      "grad_norm": 1.0867475271224976,
+      "learning_rate": 1.649950825200472e-05,
+      "loss": 0.2292,
+      "step": 810
+    },
+    {
+      "epoch": 2.4931087289433385,
+      "grad_norm": 1.1694650650024414,
+      "learning_rate": 1.634847522546739e-05,
+      "loss": 0.2242,
+      "step": 815
+    },
+    {
+      "epoch": 2.5084226646248085,
+      "grad_norm": 1.2106242179870605,
+      "learning_rate": 1.6197304244353677e-05,
+      "loss": 0.222,
+      "step": 820
+    },
+    {
+      "epoch": 2.523736600306279,
+      "grad_norm": 1.1909250020980835,
+      "learning_rate": 1.6046010774078348e-05,
+      "loss": 0.218,
+      "step": 825
+    },
+    {
+      "epoch": 2.539050535987749,
+      "grad_norm": 1.0819553136825562,
+      "learning_rate": 1.5894610292587304e-05,
+      "loss": 0.235,
+      "step": 830
+    },
+    {
+      "epoch": 2.554364471669219,
+      "grad_norm": 1.1172194480895996,
+      "learning_rate": 1.574311828877416e-05,
+      "loss": 0.2246,
+      "step": 835
+    },
+    {
+      "epoch": 2.569678407350689,
+      "grad_norm": 1.6138761043548584,
+      "learning_rate": 1.5591550260895612e-05,
+      "loss": 0.2149,
+      "step": 840
+    },
+    {
+      "epoch": 2.5849923430321593,
+      "grad_norm": 1.1283706426620483,
+      "learning_rate": 1.5439921714985954e-05,
+      "loss": 0.1873,
+      "step": 845
+    },
+    {
+      "epoch": 2.6003062787136293,
+      "grad_norm": 1.1196907758712769,
+      "learning_rate": 1.528824816327071e-05,
+      "loss": 0.2358,
+      "step": 850
+    },
+    {
+      "epoch": 2.6156202143950997,
+      "grad_norm": 1.233160138130188,
+      "learning_rate": 1.5136545122579694e-05,
+      "loss": 0.1997,
+      "step": 855
+    },
+    {
+      "epoch": 2.6309341500765697,
+      "grad_norm": 1.1278220415115356,
+      "learning_rate": 1.4984828112759551e-05,
+      "loss": 0.2217,
+      "step": 860
+    },
+    {
+      "epoch": 2.6462480857580397,
+      "grad_norm": 1.2449628114700317,
+      "learning_rate": 1.4833112655086045e-05,
+      "loss": 0.226,
+      "step": 865
+    },
+    {
+      "epoch": 2.6615620214395097,
+      "grad_norm": 1.0520964860916138,
+      "learning_rate": 1.4681414270676125e-05,
+      "loss": 0.1961,
+      "step": 870
+    },
+    {
+      "epoch": 2.67687595712098,
+      "grad_norm": 1.0986305475234985,
+      "learning_rate": 1.4529748478900096e-05,
+      "loss": 0.2014,
+      "step": 875
+    },
+    {
+      "epoch": 2.69218989280245,
+      "grad_norm": 1.0960958003997803,
+      "learning_rate": 1.4378130795793884e-05,
+      "loss": 0.1939,
+      "step": 880
+    },
+    {
+      "epoch": 2.7075038284839206,
+      "grad_norm": 1.0452426671981812,
+      "learning_rate": 1.4226576732471724e-05,
+      "loss": 0.2051,
+      "step": 885
+    },
+    {
+      "epoch": 2.7228177641653906,
+      "grad_norm": 1.2079676389694214,
+      "learning_rate": 1.4075101793539254e-05,
+      "loss": 0.18,
+      "step": 890
+    },
+    {
+      "epoch": 2.7381316998468606,
+      "grad_norm": 1.0217554569244385,
+      "learning_rate": 1.3923721475507384e-05,
+      "loss": 0.2024,
+      "step": 895
+    },
+    {
+      "epoch": 2.7534456355283305,
+      "grad_norm": 1.0963138341903687,
+      "learning_rate": 1.3772451265206905e-05,
+      "loss": 0.2098,
+      "step": 900
+    },
+    {
+      "epoch": 2.768759571209801,
+      "grad_norm": 1.184330940246582,
+      "learning_rate": 1.362130663820414e-05,
+      "loss": 0.1946,
+      "step": 905
+    },
+    {
+      "epoch": 2.784073506891271,
+      "grad_norm": 1.3663582801818848,
+      "learning_rate": 1.3470303057217711e-05,
+      "loss": 0.2119,
+      "step": 910
+    },
+    {
+      "epoch": 2.7993874425727414,
+      "grad_norm": 1.4678473472595215,
+      "learning_rate": 1.3319455970536674e-05,
+      "loss": 0.2213,
+      "step": 915
+    },
+    {
+      "epoch": 2.8147013782542114,
+      "grad_norm": 1.1842464208602905,
+      "learning_rate": 1.3168780810440038e-05,
+      "loss": 0.1932,
+      "step": 920
+    },
+    {
+      "epoch": 2.8300153139356814,
+      "grad_norm": 1.1453865766525269,
+      "learning_rate": 1.3018292991618046e-05,
+      "loss": 0.1916,
+      "step": 925
+    },
+    {
+      "epoch": 2.8453292496171514,
+      "grad_norm": 1.286776065826416,
+      "learning_rate": 1.2868007909595131e-05,
+      "loss": 0.1943,
+      "step": 930
+    },
+    {
+      "epoch": 2.860643185298622,
+      "grad_norm": 0.9827431440353394,
+      "learning_rate": 1.271794093915494e-05,
+      "loss": 0.196,
+      "step": 935
+    },
+    {
+      "epoch": 2.875957120980092,
+      "grad_norm": 1.1773149967193604,
+      "learning_rate": 1.256810743276738e-05,
+      "loss": 0.1945,
+      "step": 940
+    },
+    {
+      "epoch": 2.8912710566615623,
+      "grad_norm": 1.141097903251648,
+      "learning_rate": 1.2418522719018047e-05,
+      "loss": 0.181,
+      "step": 945
+    },
+    {
+      "epoch": 2.9065849923430322,
+      "grad_norm": 1.1019505262374878,
+      "learning_rate": 1.2269202101039999e-05,
+      "loss": 0.1769,
+      "step": 950
+    },
+    {
+      "epoch": 2.9218989280245022,
+      "grad_norm": 1.0214898586273193,
+      "learning_rate": 1.212016085494823e-05,
+      "loss": 0.2148,
+      "step": 955
+    },
+    {
+      "epoch": 2.9372128637059722,
+      "grad_norm": 1.0284066200256348,
+      "learning_rate": 1.1971414228276823e-05,
+      "loss": 0.1777,
+      "step": 960
+    },
+    {
+      "epoch": 2.9525267993874427,
+      "grad_norm": 1.144411563873291,
+      "learning_rate": 1.1822977438419092e-05,
+      "loss": 0.1839,
+      "step": 965
+    },
+    {
+      "epoch": 2.9678407350689127,
+      "grad_norm": 1.1772691011428833,
+      "learning_rate": 1.1674865671070751e-05,
+      "loss": 0.2072,
+      "step": 970
+    },
+    {
+      "epoch": 2.983154670750383,
+      "grad_norm": 1.053539752960205,
+      "learning_rate": 1.1527094078676395e-05,
+      "loss": 0.1685,
+      "step": 975
+    },
+    {
+      "epoch": 2.998468606431853,
+      "grad_norm": 1.1275657415390015,
+      "learning_rate": 1.1379677778879283e-05,
+      "loss": 0.1848,
+      "step": 980
+    },
+    {
+      "epoch": 3.012251148545176,
+      "grad_norm": 0.9341105818748474,
+      "learning_rate": 1.1232631852974814e-05,
+      "loss": 0.1463,
+      "step": 985
+    },
+    {
+      "epoch": 3.027565084226646,
+      "grad_norm": 1.0302433967590332,
+      "learning_rate": 1.1085971344367585e-05,
+      "loss": 0.124,
+      "step": 990
+    },
+    {
+      "epoch": 3.0428790199081166,
+      "grad_norm": 1.0990079641342163,
+      "learning_rate": 1.0939711257032429e-05,
+      "loss": 0.138,
+      "step": 995
+    },
+    {
+      "epoch": 3.0581929555895866,
+      "grad_norm": 1.0678660869598389,
+      "learning_rate": 1.0793866553979411e-05,
+      "loss": 0.1297,
+      "step": 1000
+    },
+    {
+      "epoch": 3.0735068912710566,
+      "grad_norm": 1.1288793087005615,
+      "learning_rate": 1.0648452155723103e-05,
+      "loss": 0.1418,
+      "step": 1005
+    },
+    {
+      "epoch": 3.088820826952527,
+      "grad_norm": 1.0335837602615356,
+      "learning_rate": 1.0503482938756091e-05,
+      "loss": 0.1367,
+      "step": 1010
+    },
+    {
+      "epoch": 3.104134762633997,
+      "grad_norm": 0.9697948694229126,
+      "learning_rate": 1.0358973734027098e-05,
+      "loss": 0.1264,
+      "step": 1015
+    },
+    {
+      "epoch": 3.119448698315467,
+      "grad_norm": 1.1903094053268433,
+      "learning_rate": 1.0214939325423686e-05,
+      "loss": 0.1372,
+      "step": 1020
+    },
+    {
+      "epoch": 3.134762633996937,
+      "grad_norm": 1.1920007467269897,
+      "learning_rate": 1.0071394448259834e-05,
+      "loss": 0.1284,
+      "step": 1025
+    },
+    {
+      "epoch": 3.1500765696784074,
+      "grad_norm": 0.9497950673103333,
+      "learning_rate": 9.928353787768404e-06,
+      "loss": 0.1293,
+      "step": 1030
+    },
+    {
+      "epoch": 3.1653905053598774,
+      "grad_norm": 1.056449294090271,
+      "learning_rate": 9.785831977598855e-06,
+      "loss": 0.1191,
+      "step": 1035
+    },
+    {
+      "epoch": 3.1807044410413474,
+      "grad_norm": 1.0650913715362549,
+      "learning_rate": 9.643843598320078e-06,
+      "loss": 0.1298,
+      "step": 1040
+    },
+    {
+      "epoch": 3.196018376722818,
+      "grad_norm": 0.9316422939300537,
+      "learning_rate": 9.50240317592881e-06,
+      "loss": 0.1318,
+      "step": 1045
+    },
+    {
+      "epoch": 3.211332312404288,
+      "grad_norm": 1.1120059490203857,
+      "learning_rate": 9.36152518036353e-06,
+      "loss": 0.1315,
+      "step": 1050
+    },
+    {
+      "epoch": 3.226646248085758,
+      "grad_norm": 1.050228238105774,
+      "learning_rate": 9.22122402402414e-06,
+      "loss": 0.1145,
+      "step": 1055
+    },
+    {
+      "epoch": 3.2419601837672283,
+      "grad_norm": 1.0715663433074951,
+      "learning_rate": 9.08151406029749e-06,
+      "loss": 0.1299,
+      "step": 1060
+    },
+    {
+      "epoch": 3.2572741194486983,
+      "grad_norm": 1.7908285856246948,
+      "learning_rate": 8.942409582089018e-06,
+      "loss": 0.1442,
+      "step": 1065
+    },
+    {
+      "epoch": 3.2725880551301683,
+      "grad_norm": 1.0539591312408447,
+      "learning_rate": 8.803924820360469e-06,
+      "loss": 0.1291,
+      "step": 1070
+    },
+    {
+      "epoch": 3.2879019908116387,
+      "grad_norm": 0.9626232981681824,
+      "learning_rate": 8.666073942674064e-06,
+      "loss": 0.1401,
+      "step": 1075
+    },
+    {
+      "epoch": 3.3032159264931087,
+      "grad_norm": 1.1888229846954346,
+      "learning_rate": 8.528871051743038e-06,
+      "loss": 0.1368,
+      "step": 1080
+    },
+    {
+      "epoch": 3.3185298621745787,
+      "grad_norm": 1.115984559059143,
+      "learning_rate": 8.392330183988942e-06,
+      "loss": 0.1272,
+      "step": 1085
+    },
+    {
+      "epoch": 3.333843797856049,
+      "grad_norm": 0.9548808932304382,
+      "learning_rate": 8.256465308105608e-06,
+      "loss": 0.1236,
+      "step": 1090
+    },
+    {
+      "epoch": 3.349157733537519,
+      "grad_norm": 1.2673202753067017,
+      "learning_rate": 8.121290323630116e-06,
+      "loss": 0.1233,
+      "step": 1095
+    },
+    {
+      "epoch": 3.364471669218989,
+      "grad_norm": 1.1379413604736328,
+      "learning_rate": 7.986819059520803e-06,
+      "loss": 0.1312,
+      "step": 1100
+    },
+    {
+      "epoch": 3.3797856049004595,
+      "grad_norm": 1.0691787004470825,
+      "learning_rate": 7.853065272742536e-06,
+      "loss": 0.1318,
+      "step": 1105
+    },
+    {
+      "epoch": 3.3950995405819295,
+      "grad_norm": 1.019787073135376,
+      "learning_rate": 7.720042646859247e-06,
+      "loss": 0.1341,
+      "step": 1110
+    },
+    {
+      "epoch": 3.4104134762633995,
+      "grad_norm": 1.0537774562835693,
+      "learning_rate": 7.587764790634143e-06,
+      "loss": 0.1391,
+      "step": 1115
+    },
+    {
+      "epoch": 3.42572741194487,
+      "grad_norm": 1.0213468074798584,
+      "learning_rate": 7.456245236637392e-06,
+      "loss": 0.1144,
+      "step": 1120
+    },
+    {
+      "epoch": 3.44104134762634,
+      "grad_norm": 1.038906455039978,
+      "learning_rate": 7.325497439861725e-06,
+      "loss": 0.1216,
+      "step": 1125
+    },
+    {
+      "epoch": 3.45635528330781,
+      "grad_norm": 1.2020183801651,
+      "learning_rate": 7.195534776345923e-06,
+      "loss": 0.1266,
+      "step": 1130
+    },
+    {
+      "epoch": 3.4716692189892804,
+      "grad_norm": 0.8986299633979797,
+      "learning_rate": 7.066370541806418e-06,
+      "loss": 0.1341,
+      "step": 1135
+    },
+    {
+      "epoch": 3.4869831546707504,
+      "grad_norm": 0.9827359914779663,
+      "learning_rate": 6.938017950277025e-06,
+      "loss": 0.1213,
+      "step": 1140
+    },
+    {
+      "epoch": 3.5022970903522204,
+      "grad_norm": 1.0107229948043823,
+      "learning_rate": 6.810490132757175e-06,
+      "loss": 0.1288,
+      "step": 1145
+    },
+    {
+      "epoch": 3.517611026033691,
+      "grad_norm": 1.0034672021865845,
+      "learning_rate": 6.683800135868487e-06,
+      "loss": 0.1193,
+      "step": 1150
+    },
+    {
+      "epoch": 3.532924961715161,
+      "grad_norm": 1.0432562828063965,
+      "learning_rate": 6.557960920520089e-06,
+      "loss": 0.1337,
+      "step": 1155
+    },
+    {
+      "epoch": 3.548238897396631,
+      "grad_norm": 0.9052329659461975,
+      "learning_rate": 6.432985360582641e-06,
+      "loss": 0.1091,
+      "step": 1160
+    },
+    {
+      "epoch": 3.5635528330781012,
+      "grad_norm": 0.9641390442848206,
+      "learning_rate": 6.308886241571299e-06,
+      "loss": 0.1101,
+      "step": 1165
+    },
+    {
+      "epoch": 3.578866768759571,
+      "grad_norm": 1.187159776687622,
+      "learning_rate": 6.18567625933769e-06,
+      "loss": 0.122,
+      "step": 1170
+    },
+    {
+      "epoch": 3.594180704441041,
+      "grad_norm": 0.9154672622680664,
+      "learning_rate": 6.063368018771096e-06,
+      "loss": 0.1059,
+      "step": 1175
+    },
+    {
+      "epoch": 3.6094946401225116,
+      "grad_norm": 0.9932954907417297,
+      "learning_rate": 5.941974032508896e-06,
+      "loss": 0.1183,
+      "step": 1180
+    },
+    {
+      "epoch": 3.6248085758039816,
+      "grad_norm": 0.9219599366188049,
+      "learning_rate": 5.821506719656485e-06,
+      "loss": 0.1058,
+      "step": 1185
+    },
+    {
+      "epoch": 3.6401225114854516,
+      "grad_norm": 0.8277255296707153,
+      "learning_rate": 5.70197840451675e-06,
+      "loss": 0.1076,
+      "step": 1190
+    },
+    {
+      "epoch": 3.655436447166922,
+      "grad_norm": 0.8804378509521484,
+      "learning_rate": 5.5834013153292396e-06,
+      "loss": 0.113,
+      "step": 1195
+    },
+    {
+      "epoch": 3.670750382848392,
+      "grad_norm": 0.9384981989860535,
+      "learning_rate": 5.465787583019156e-06,
+      "loss": 0.101,
+      "step": 1200
+    },
+    {
+      "epoch": 3.686064318529862,
+      "grad_norm": 1.1971505880355835,
+      "learning_rate": 5.349149239956339e-06,
+      "loss": 0.1061,
+      "step": 1205
+    },
+    {
+      "epoch": 3.701378254211332,
+      "grad_norm": 1.0984221696853638,
+      "learning_rate": 5.23349821872427e-06,
+      "loss": 0.1276,
+      "step": 1210
+    },
+    {
+      "epoch": 3.7166921898928025,
+      "grad_norm": 0.9261698126792908,
+      "learning_rate": 5.1188463508993396e-06,
+      "loss": 0.1094,
+      "step": 1215
+    },
+    {
+      "epoch": 3.7320061255742725,
+      "grad_norm": 0.9417651891708374,
+      "learning_rate": 5.0052053658404285e-06,
+      "loss": 0.1085,
+      "step": 1220
+    },
+    {
+      "epoch": 3.747320061255743,
+      "grad_norm": 0.9424726963043213,
+      "learning_rate": 4.892586889488935e-06,
+      "loss": 0.1183,
+      "step": 1225
+    },
+    {
+      "epoch": 3.762633996937213,
+      "grad_norm": 0.9524731636047363,
+      "learning_rate": 4.7810024431794e-06,
+      "loss": 0.1177,
+      "step": 1230
+    },
+    {
+      "epoch": 3.777947932618683,
+      "grad_norm": 0.8342095017433167,
+      "learning_rate": 4.670463442460842e-06,
+      "loss": 0.1,
+      "step": 1235
+    },
+    {
+      "epoch": 3.793261868300153,
+      "grad_norm": 0.9804592132568359,
+      "learning_rate": 4.560981195928869e-06,
+      "loss": 0.1109,
+      "step": 1240
+    },
+    {
+      "epoch": 3.8085758039816233,
+      "grad_norm": 1.1521655321121216,
+      "learning_rate": 4.452566904068778e-06,
+      "loss": 0.1081,
+      "step": 1245
+    },
+    {
+      "epoch": 3.8238897396630933,
+      "grad_norm": 1.014464259147644,
+      "learning_rate": 4.345231658109707e-06,
+      "loss": 0.1126,
+      "step": 1250
+    },
+    {
+      "epoch": 3.8392036753445637,
+      "grad_norm": 1.0303317308425903,
+      "learning_rate": 4.238986438889942e-06,
+      "loss": 0.1125,
+      "step": 1255
+    },
+    {
+      "epoch": 3.8545176110260337,
+      "grad_norm": 1.0285574197769165,
+      "learning_rate": 4.133842115733537e-06,
+      "loss": 0.1163,
+      "step": 1260
+    },
+    {
+      "epoch": 3.8698315467075037,
+      "grad_norm": 0.921039342880249,
+      "learning_rate": 4.029809445338359e-06,
+      "loss": 0.0999,
+      "step": 1265
+    },
+    {
+      "epoch": 3.8851454823889737,
+      "grad_norm": 0.8448448777198792,
+      "learning_rate": 3.926899070675577e-06,
+      "loss": 0.1109,
+      "step": 1270
+    },
+    {
+      "epoch": 3.900459418070444,
+      "grad_norm": 0.9851784706115723,
+      "learning_rate": 3.825121519900913e-06,
+      "loss": 0.1054,
+      "step": 1275
+    },
+    {
+      "epoch": 3.915773353751914,
+      "grad_norm": 1.327043890953064,
+      "learning_rate": 3.7244872052775165e-06,
+      "loss": 0.1034,
+      "step": 1280
+    },
+    {
+      "epoch": 3.9310872894333846,
+      "grad_norm": 0.8687987923622131,
+      "learning_rate": 3.6250064221107638e-06,
+      "loss": 0.0915,
+      "step": 1285
+    },
+    {
+      "epoch": 3.9464012251148546,
+      "grad_norm": 0.9403600692749023,
+      "learning_rate": 3.5266893476950096e-06,
+      "loss": 0.1012,
+      "step": 1290
+    },
+    {
+      "epoch": 3.9617151607963246,
+      "grad_norm": 0.7973408699035645,
+      "learning_rate": 3.4295460402724148e-06,
+      "loss": 0.0837,
+      "step": 1295
+    },
+    {
+      "epoch": 3.9770290964777946,
+      "grad_norm": 0.8510920405387878,
+      "learning_rate": 3.3335864380039123e-06,
+      "loss": 0.088,
+      "step": 1300
+    },
+    {
+      "epoch": 3.992343032159265,
+      "grad_norm": 0.9783728122711182,
+      "learning_rate": 3.238820357952539e-06,
+      "loss": 0.0981,
+      "step": 1305
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1635,
+  "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": 1.8916381367405117e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

41_128_e5_3e-5/checkpoint-1308/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ccd25f5c530b5e7648b66503aed21679838fee027eb7c7df40e963df785d384d
+size 7736

41_128_e5_3e-5/checkpoint-1308/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)

41_128_e5_3e-5/checkpoint-1635/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

41_128_e5_3e-5/checkpoint-1635/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": [
+    "gate_proj",
+    "o_proj",
+    "up_proj",
+    "v_proj",
+    "k_proj",
+    "q_proj",
+    "down_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

41_128_e5_3e-5/checkpoint-1635/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b451b0d2156d710dab849e7423b0405677ffc908cc6f5f662195c674a4d0197
+size 791751704

41_128_e5_3e-5/checkpoint-1635/latest

@@ -0,0 +1 @@
+global_step1635

41_128_e5_3e-5/checkpoint-1635/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:781d88121a7a906517f0d36478fa9a5bd33ed8adfd3becfe0f21c9f0cf90301d
+size 15920

41_128_e5_3e-5/checkpoint-1635/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7fe8b5cce98a56397157b6f7b1b0771910497410848d86234cb2bd8da7fab67f
+size 15920

41_128_e5_3e-5/checkpoint-1635/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c7fb847840acc3fc3902e43fa24553ce726c7bc5d102ed477be12baf1795482f
+size 15920

41_128_e5_3e-5/checkpoint-1635/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b4b5d140bebca020cf38b4387fe3a9cff5c8875bcb2e31a894154e243e5ae92e
+size 15920

41_128_e5_3e-5/checkpoint-1635/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7eac80f47def9c267278ca45d8f9d26ff2038268fb91d99f31a7c4e83a5d7488
+size 15920

41_128_e5_3e-5/checkpoint-1635/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f14f9747d8765a431bee0f3c3a99dfa4ad0f6ca99d75ad71c2224b92c65dfabe
+size 15920

41_128_e5_3e-5/checkpoint-1635/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0443da98d87082f6671b2c46819651453561c6fd5202d725182328ef1cdc53ae
+size 15920

41_128_e5_3e-5/checkpoint-1635/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b51ae8caabe70425dd6649d9a9ff55b102e123317364e8941fe7ad5c1c087efa
+size 15920

41_128_e5_3e-5/checkpoint-1635/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:23b385d103e4c2cad45f7aedc2c427529293b6446b015f3ae47689b608eb04d8
+size 1064

41_128_e5_3e-5/checkpoint-1635/special_tokens_map.json

@@ -0,0 +1,45 @@
+{
+  "additional_special_tokens": [
+    "<|endoftext|>",
+    "<fim_prefix>",
+    "<fim_middle>",
+    "<fim_suffix>",
+    "<fim_pad>",
+    "<filename>",
+    "<gh_stars>",
+    "<issue_start>",
+    "<issue_comment>",
+    "<issue_closed>",
+    "<jupyter_start>",
+    "<jupyter_text>",
+    "<jupyter_code>",
+    "<jupyter_output>",
+    "<empty_output>",
+    "<commit_before>",
+    "<commit_msg>",
+    "<commit_after>",
+    "<reponame>"
+  ],
+  "bos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "<reponame>",
+  "unk_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

41_128_e5_3e-5/checkpoint-1635/tokenizer_config.json

@@ -0,0 +1,188 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<fim_prefix>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<fim_middle>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<fim_suffix>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<fim_pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "5": {
+      "content": "<filename>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "6": {
+      "content": "<gh_stars>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "7": {
+      "content": "<issue_start>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "8": {
+      "content": "<issue_comment>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "9": {
+      "content": "<issue_closed>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "10": {
+      "content": "<jupyter_start>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "11": {
+      "content": "<jupyter_text>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "12": {
+      "content": "<jupyter_code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "13": {
+      "content": "<jupyter_output>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "14": {
+      "content": "<empty_output>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "15": {
+      "content": "<commit_before>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "16": {
+      "content": "<commit_msg>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "17": {
+      "content": "<commit_after>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "18": {
+      "content": "<reponame>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<|endoftext|>",
+    "<fim_prefix>",
+    "<fim_middle>",
+    "<fim_suffix>",
+    "<fim_pad>",
+    "<filename>",
+    "<gh_stars>",
+    "<issue_start>",
+    "<issue_comment>",
+    "<issue_closed>",
+    "<jupyter_start>",
+    "<jupyter_text>",
+    "<jupyter_code>",
+    "<jupyter_output>",
+    "<empty_output>",
+    "<commit_before>",
+    "<commit_msg>",
+    "<commit_after>",
+    "<reponame>"
+  ],
+  "bos_token": "<|endoftext|>",
+  "clean_up_tokenization_spaces": true,
+  "eos_token": "<|endoftext|>",
+  "extra_special_tokens": {},
+  "model_max_length": 8192,
+  "pad_token": "<reponame>",
+  "padding_side": "left",
+  "tokenizer_class": "GPT2Tokenizer",
+  "unk_token": "<|endoftext|>",
+  "vocab_size": 49152
+}

41_128_e5_3e-5/checkpoint-1635/trainer_state.json

@@ -0,0 +1,2323 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 1635,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.015313935681470138,
+      "grad_norm": 1.1070387363433838,
+      "learning_rate": 1.4634146341463416e-06,
+      "loss": 1.2917,
+      "step": 5
+    },
+    {
+      "epoch": 0.030627871362940276,
+      "grad_norm": 1.122970700263977,
+      "learning_rate": 3.2926829268292685e-06,
+      "loss": 1.2965,
+      "step": 10
+    },
+    {
+      "epoch": 0.045941807044410414,
+      "grad_norm": 0.8600009083747864,
+      "learning_rate": 5.121951219512195e-06,
+      "loss": 1.3084,
+      "step": 15
+    },
+    {
+      "epoch": 0.06125574272588055,
+      "grad_norm": 0.6501789093017578,
+      "learning_rate": 6.951219512195123e-06,
+      "loss": 1.3178,
+      "step": 20
+    },
+    {
+      "epoch": 0.07656967840735068,
+      "grad_norm": 0.6199907660484314,
+      "learning_rate": 8.780487804878048e-06,
+      "loss": 1.258,
+      "step": 25
+    },
+    {
+      "epoch": 0.09188361408882083,
+      "grad_norm": 0.525578498840332,
+      "learning_rate": 1.0609756097560975e-05,
+      "loss": 1.2632,
+      "step": 30
+    },
+    {
+      "epoch": 0.10719754977029096,
+      "grad_norm": 0.5301306843757629,
+      "learning_rate": 1.2439024390243903e-05,
+      "loss": 1.1863,
+      "step": 35
+    },
+    {
+      "epoch": 0.1225114854517611,
+      "grad_norm": 0.45140478014945984,
+      "learning_rate": 1.4268292682926829e-05,
+      "loss": 1.1721,
+      "step": 40
+    },
+    {
+      "epoch": 0.13782542113323124,
+      "grad_norm": 0.5104884505271912,
+      "learning_rate": 1.6097560975609757e-05,
+      "loss": 1.2507,
+      "step": 45
+    },
+    {
+      "epoch": 0.15313935681470137,
+      "grad_norm": 0.4204311668872833,
+      "learning_rate": 1.7926829268292684e-05,
+      "loss": 1.1973,
+      "step": 50
+    },
+    {
+      "epoch": 0.16845329249617153,
+      "grad_norm": 0.43958738446235657,
+      "learning_rate": 1.975609756097561e-05,
+      "loss": 1.1404,
+      "step": 55
+    },
+    {
+      "epoch": 0.18376722817764166,
+      "grad_norm": 0.45204630494117737,
+      "learning_rate": 2.1585365853658537e-05,
+      "loss": 1.1811,
+      "step": 60
+    },
+    {
+      "epoch": 0.1990811638591118,
+      "grad_norm": 0.4730817973613739,
+      "learning_rate": 2.3414634146341466e-05,
+      "loss": 1.1368,
+      "step": 65
+    },
+    {
+      "epoch": 0.21439509954058192,
+      "grad_norm": 0.5665387511253357,
+      "learning_rate": 2.524390243902439e-05,
+      "loss": 1.1341,
+      "step": 70
+    },
+    {
+      "epoch": 0.22970903522205208,
+      "grad_norm": 0.49841874837875366,
+      "learning_rate": 2.707317073170732e-05,
+      "loss": 1.1277,
+      "step": 75
+    },
+    {
+      "epoch": 0.2450229709035222,
+      "grad_norm": 0.4274734556674957,
+      "learning_rate": 2.8902439024390242e-05,
+      "loss": 1.0674,
+      "step": 80
+    },
+    {
+      "epoch": 0.26033690658499237,
+      "grad_norm": 0.6278162598609924,
+      "learning_rate": 2.9999877234172298e-05,
+      "loss": 1.0953,
+      "step": 85
+    },
+    {
+      "epoch": 0.27565084226646247,
+      "grad_norm": 0.5454415082931519,
+      "learning_rate": 2.999849614168868e-05,
+      "loss": 1.0974,
+      "step": 90
+    },
+    {
+      "epoch": 0.29096477794793263,
+      "grad_norm": 0.5087949633598328,
+      "learning_rate": 2.999558064119925e-05,
+      "loss": 1.1079,
+      "step": 95
+    },
+    {
+      "epoch": 0.30627871362940273,
+      "grad_norm": 0.5586227774620056,
+      "learning_rate": 2.999113103097174e-05,
+      "loss": 1.0655,
+      "step": 100
+    },
+    {
+      "epoch": 0.3215926493108729,
+      "grad_norm": 0.5441133975982666,
+      "learning_rate": 2.9985147766219615e-05,
+      "loss": 1.0631,
+      "step": 105
+    },
+    {
+      "epoch": 0.33690658499234305,
+      "grad_norm": 0.6769384741783142,
+      "learning_rate": 2.9977631459055537e-05,
+      "loss": 1.0438,
+      "step": 110
+    },
+    {
+      "epoch": 0.35222052067381315,
+      "grad_norm": 0.523868978023529,
+      "learning_rate": 2.996858287842873e-05,
+      "loss": 1.0131,
+      "step": 115
+    },
+    {
+      "epoch": 0.3675344563552833,
+      "grad_norm": 0.5572935342788696,
+      "learning_rate": 2.995800295004629e-05,
+      "loss": 1.0501,
+      "step": 120
+    },
+    {
+      "epoch": 0.38284839203675347,
+      "grad_norm": 0.6209707856178284,
+      "learning_rate": 2.9945892756278543e-05,
+      "loss": 1.0125,
+      "step": 125
+    },
+    {
+      "epoch": 0.3981623277182236,
+      "grad_norm": 0.6483351588249207,
+      "learning_rate": 2.993225353604824e-05,
+      "loss": 0.998,
+      "step": 130
+    },
+    {
+      "epoch": 0.41347626339969373,
+      "grad_norm": 0.6352364420890808,
+      "learning_rate": 2.9917086684703844e-05,
+      "loss": 1.0163,
+      "step": 135
+    },
+    {
+      "epoch": 0.42879019908116384,
+      "grad_norm": 0.6867502331733704,
+      "learning_rate": 2.9900393753876816e-05,
+      "loss": 0.9824,
+      "step": 140
+    },
+    {
+      "epoch": 0.444104134762634,
+      "grad_norm": 0.6153353452682495,
+      "learning_rate": 2.9882176451322798e-05,
+      "loss": 0.9395,
+      "step": 145
+    },
+    {
+      "epoch": 0.45941807044410415,
+      "grad_norm": 0.645903468132019,
+      "learning_rate": 2.9862436640746973e-05,
+      "loss": 0.914,
+      "step": 150
+    },
+    {
+      "epoch": 0.47473200612557426,
+      "grad_norm": 0.671872615814209,
+      "learning_rate": 2.9841176341613364e-05,
+      "loss": 0.9177,
+      "step": 155
+    },
+    {
+      "epoch": 0.4900459418070444,
+      "grad_norm": 0.6303296089172363,
+      "learning_rate": 2.981839772893825e-05,
+      "loss": 0.9576,
+      "step": 160
+    },
+    {
+      "epoch": 0.5053598774885145,
+      "grad_norm": 0.7237289547920227,
+      "learning_rate": 2.9794103133067637e-05,
+      "loss": 0.9649,
+      "step": 165
+    },
+    {
+      "epoch": 0.5206738131699847,
+      "grad_norm": 0.6527206301689148,
+      "learning_rate": 2.9768295039438868e-05,
+      "loss": 0.9158,
+      "step": 170
+    },
+    {
+      "epoch": 0.5359877488514548,
+      "grad_norm": 0.7285829782485962,
+      "learning_rate": 2.974097608832635e-05,
+      "loss": 0.9173,
+      "step": 175
+    },
+    {
+      "epoch": 0.5513016845329249,
+      "grad_norm": 0.8059846758842468,
+      "learning_rate": 2.9712149074571433e-05,
+      "loss": 0.8667,
+      "step": 180
+    },
+    {
+      "epoch": 0.5666156202143952,
+      "grad_norm": 0.7382646799087524,
+      "learning_rate": 2.968181694729651e-05,
+      "loss": 0.877,
+      "step": 185
+    },
+    {
+      "epoch": 0.5819295558958653,
+      "grad_norm": 0.7463013529777527,
+      "learning_rate": 2.964998280960328e-05,
+      "loss": 0.9148,
+      "step": 190
+    },
+    {
+      "epoch": 0.5972434915773354,
+      "grad_norm": 0.7870151400566101,
+      "learning_rate": 2.961664991825531e-05,
+      "loss": 0.8881,
+      "step": 195
+    },
+    {
+      "epoch": 0.6125574272588055,
+      "grad_norm": 0.8590981364250183,
+      "learning_rate": 2.9581821683344832e-05,
+      "loss": 0.8352,
+      "step": 200
+    },
+    {
+      "epoch": 0.6278713629402757,
+      "grad_norm": 0.7620300650596619,
+      "learning_rate": 2.954550166794391e-05,
+      "loss": 0.7867,
+      "step": 205
+    },
+    {
+      "epoch": 0.6431852986217458,
+      "grad_norm": 0.8273492455482483,
+      "learning_rate": 2.9507693587739895e-05,
+      "loss": 0.8456,
+      "step": 210
+    },
+    {
+      "epoch": 0.6584992343032159,
+      "grad_norm": 0.7399709224700928,
+      "learning_rate": 2.9468401310655303e-05,
+      "loss": 0.8156,
+      "step": 215
+    },
+    {
+      "epoch": 0.6738131699846861,
+      "grad_norm": 0.851788341999054,
+      "learning_rate": 2.942762885645211e-05,
+      "loss": 0.7952,
+      "step": 220
+    },
+    {
+      "epoch": 0.6891271056661562,
+      "grad_norm": 0.8381015658378601,
+      "learning_rate": 2.9385380396320523e-05,
+      "loss": 0.8027,
+      "step": 225
+    },
+    {
+      "epoch": 0.7044410413476263,
+      "grad_norm": 0.8373646140098572,
+      "learning_rate": 2.934166025245223e-05,
+      "loss": 0.7943,
+      "step": 230
+    },
+    {
+      "epoch": 0.7197549770290965,
+      "grad_norm": 0.8320114612579346,
+      "learning_rate": 2.9296472897598246e-05,
+      "loss": 0.8064,
+      "step": 235
+    },
+    {
+      "epoch": 0.7350689127105666,
+      "grad_norm": 1.0138018131256104,
+      "learning_rate": 2.924982295461131e-05,
+      "loss": 0.761,
+      "step": 240
+    },
+    {
+      "epoch": 0.7503828483920367,
+      "grad_norm": 0.9759603142738342,
+      "learning_rate": 2.920171519597297e-05,
+      "loss": 0.7114,
+      "step": 245
+    },
+    {
+      "epoch": 0.7656967840735069,
+      "grad_norm": 0.844030499458313,
+      "learning_rate": 2.9152154543305316e-05,
+      "loss": 0.7461,
+      "step": 250
+    },
+    {
+      "epoch": 0.781010719754977,
+      "grad_norm": 0.8687477111816406,
+      "learning_rate": 2.9101146066867502e-05,
+      "loss": 0.728,
+      "step": 255
+    },
+    {
+      "epoch": 0.7963246554364471,
+      "grad_norm": 0.9759243726730347,
+      "learning_rate": 2.9048694985037e-05,
+      "loss": 0.7475,
+      "step": 260
+    },
+    {
+      "epoch": 0.8116385911179173,
+      "grad_norm": 0.9230566620826721,
+      "learning_rate": 2.8994806663775792e-05,
+      "loss": 0.8098,
+      "step": 265
+    },
+    {
+      "epoch": 0.8269525267993875,
+      "grad_norm": 0.8410807251930237,
+      "learning_rate": 2.893948661608136e-05,
+      "loss": 0.7465,
+      "step": 270
+    },
+    {
+      "epoch": 0.8422664624808576,
+      "grad_norm": 1.0104124546051025,
+      "learning_rate": 2.888274050142271e-05,
+      "loss": 0.7799,
+      "step": 275
+    },
+    {
+      "epoch": 0.8575803981623277,
+      "grad_norm": 0.9490285515785217,
+      "learning_rate": 2.8824574125161384e-05,
+      "loss": 0.7541,
+      "step": 280
+    },
+    {
+      "epoch": 0.8728943338437979,
+      "grad_norm": 0.9112844467163086,
+      "learning_rate": 2.876499343795754e-05,
+      "loss": 0.7424,
+      "step": 285
+    },
+    {
+      "epoch": 0.888208269525268,
+      "grad_norm": 0.8824840784072876,
+      "learning_rate": 2.8704004535161172e-05,
+      "loss": 0.7448,
+      "step": 290
+    },
+    {
+      "epoch": 0.9035222052067381,
+      "grad_norm": 1.3259707689285278,
+      "learning_rate": 2.864161365618854e-05,
+      "loss": 0.7165,
+      "step": 295
+    },
+    {
+      "epoch": 0.9188361408882083,
+      "grad_norm": 0.938458263874054,
+      "learning_rate": 2.8577827183883853e-05,
+      "loss": 0.6724,
+      "step": 300
+    },
+    {
+      "epoch": 0.9341500765696784,
+      "grad_norm": 0.940185010433197,
+      "learning_rate": 2.851265164386627e-05,
+      "loss": 0.6722,
+      "step": 305
+    },
+    {
+      "epoch": 0.9494640122511485,
+      "grad_norm": 1.0995936393737793,
+      "learning_rate": 2.84460937038623e-05,
+      "loss": 0.6628,
+      "step": 310
+    },
+    {
+      "epoch": 0.9647779479326187,
+      "grad_norm": 0.9380448460578918,
+      "learning_rate": 2.837816017302368e-05,
+      "loss": 0.6519,
+      "step": 315
+    },
+    {
+      "epoch": 0.9800918836140888,
+      "grad_norm": 1.0398368835449219,
+      "learning_rate": 2.8308858001230757e-05,
+      "loss": 0.6781,
+      "step": 320
+    },
+    {
+      "epoch": 0.9954058192955589,
+      "grad_norm": 1.1420120000839233,
+      "learning_rate": 2.823819427838149e-05,
+      "loss": 0.6554,
+      "step": 325
+    },
+    {
+      "epoch": 1.0091883614088821,
+      "grad_norm": 0.9629755020141602,
+      "learning_rate": 2.8166176233666137e-05,
+      "loss": 0.6067,
+      "step": 330
+    },
+    {
+      "epoch": 1.0245022970903521,
+      "grad_norm": 0.9918197393417358,
+      "learning_rate": 2.8092811234827643e-05,
+      "loss": 0.5579,
+      "step": 335
+    },
+    {
+      "epoch": 1.0398162327718223,
+      "grad_norm": 1.004963994026184,
+      "learning_rate": 2.8018106787407942e-05,
+      "loss": 0.5962,
+      "step": 340
+    },
+    {
+      "epoch": 1.0551301684532925,
+      "grad_norm": 1.010399580001831,
+      "learning_rate": 2.7942070533980072e-05,
+      "loss": 0.5431,
+      "step": 345
+    },
+    {
+      "epoch": 1.0704441041347625,
+      "grad_norm": 1.1038013696670532,
+      "learning_rate": 2.7864710253366303e-05,
+      "loss": 0.5716,
+      "step": 350
+    },
+    {
+      "epoch": 1.0857580398162328,
+      "grad_norm": 1.2030080556869507,
+      "learning_rate": 2.778603385984237e-05,
+      "loss": 0.5613,
+      "step": 355
+    },
+    {
+      "epoch": 1.101071975497703,
+      "grad_norm": 1.0458992719650269,
+      "learning_rate": 2.7706049402327768e-05,
+      "loss": 0.5448,
+      "step": 360
+    },
+    {
+      "epoch": 1.116385911179173,
+      "grad_norm": 1.082198143005371,
+      "learning_rate": 2.762476506356235e-05,
+      "loss": 0.5488,
+      "step": 365
+    },
+    {
+      "epoch": 1.1316998468606432,
+      "grad_norm": 1.0959954261779785,
+      "learning_rate": 2.7542189159269155e-05,
+      "loss": 0.5366,
+      "step": 370
+    },
+    {
+      "epoch": 1.1470137825421134,
+      "grad_norm": 1.0527937412261963,
+      "learning_rate": 2.745833013730373e-05,
+      "loss": 0.53,
+      "step": 375
+    },
+    {
+      "epoch": 1.1623277182235834,
+      "grad_norm": 1.1240211725234985,
+      "learning_rate": 2.737319657678983e-05,
+      "loss": 0.5106,
+      "step": 380
+    },
+    {
+      "epoch": 1.1776416539050536,
+      "grad_norm": 1.073337197303772,
+      "learning_rate": 2.7286797187241772e-05,
+      "loss": 0.5241,
+      "step": 385
+    },
+    {
+      "epoch": 1.1929555895865238,
+      "grad_norm": 1.1111088991165161,
+      "learning_rate": 2.7199140807673398e-05,
+      "loss": 0.5168,
+      "step": 390
+    },
+    {
+      "epoch": 1.2082695252679938,
+      "grad_norm": 1.1780794858932495,
+      "learning_rate": 2.711023640569381e-05,
+      "loss": 0.5503,
+      "step": 395
+    },
+    {
+      "epoch": 1.223583460949464,
+      "grad_norm": 1.138046145439148,
+      "learning_rate": 2.702009307658996e-05,
+      "loss": 0.5197,
+      "step": 400
+    },
+    {
+      "epoch": 1.2388973966309342,
+      "grad_norm": 1.1711301803588867,
+      "learning_rate": 2.692872004239614e-05,
+      "loss": 0.5512,
+      "step": 405
+    },
+    {
+      "epoch": 1.2542113323124042,
+      "grad_norm": 1.0399030447006226,
+      "learning_rate": 2.6836126650950556e-05,
+      "loss": 0.5246,
+      "step": 410
+    },
+    {
+      "epoch": 1.2695252679938744,
+      "grad_norm": 1.0978562831878662,
+      "learning_rate": 2.6742322374938985e-05,
+      "loss": 0.5186,
+      "step": 415
+    },
+    {
+      "epoch": 1.2848392036753444,
+      "grad_norm": 1.1471067667007446,
+      "learning_rate": 2.664731681092569e-05,
+      "loss": 0.5185,
+      "step": 420
+    },
+    {
+      "epoch": 1.3001531393568146,
+      "grad_norm": 1.3175466060638428,
+      "learning_rate": 2.6551119678371655e-05,
+      "loss": 0.4764,
+      "step": 425
+    },
+    {
+      "epoch": 1.3154670750382849,
+      "grad_norm": 0.9900484681129456,
+      "learning_rate": 2.645374081864023e-05,
+      "loss": 0.5133,
+      "step": 430
+    },
+    {
+      "epoch": 1.3307810107197549,
+      "grad_norm": 1.1252256631851196,
+      "learning_rate": 2.635519019399034e-05,
+      "loss": 0.5122,
+      "step": 435
+    },
+    {
+      "epoch": 1.346094946401225,
+      "grad_norm": 1.205263376235962,
+      "learning_rate": 2.6255477886557287e-05,
+      "loss": 0.5093,
+      "step": 440
+    },
+    {
+      "epoch": 1.3614088820826953,
+      "grad_norm": 1.0410200357437134,
+      "learning_rate": 2.615461409732131e-05,
+      "loss": 0.4944,
+      "step": 445
+    },
+    {
+      "epoch": 1.3767228177641653,
+      "grad_norm": 1.0574573278427124,
+      "learning_rate": 2.6052609145064e-05,
+      "loss": 0.4646,
+      "step": 450
+    },
+    {
+      "epoch": 1.3920367534456355,
+      "grad_norm": 1.2100945711135864,
+      "learning_rate": 2.5949473465312602e-05,
+      "loss": 0.5132,
+      "step": 455
+    },
+    {
+      "epoch": 1.4073506891271057,
+      "grad_norm": 1.1366932392120361,
+      "learning_rate": 2.5845217609272474e-05,
+      "loss": 0.5026,
+      "step": 460
+    },
+    {
+      "epoch": 1.4226646248085757,
+      "grad_norm": 1.147862195968628,
+      "learning_rate": 2.5739852242747614e-05,
+      "loss": 0.4918,
+      "step": 465
+    },
+    {
+      "epoch": 1.437978560490046,
+      "grad_norm": 1.06094229221344,
+      "learning_rate": 2.5633388145049525e-05,
+      "loss": 0.4868,
+      "step": 470
+    },
+    {
+      "epoch": 1.4532924961715161,
+      "grad_norm": 1.0694596767425537,
+      "learning_rate": 2.552583620789443e-05,
+      "loss": 0.4744,
+      "step": 475
+    },
+    {
+      "epoch": 1.4686064318529861,
+      "grad_norm": 0.9866558909416199,
+      "learning_rate": 2.5417207434289022e-05,
+      "loss": 0.4992,
+      "step": 480
+    },
+    {
+      "epoch": 1.4839203675344563,
+      "grad_norm": 1.0969899892807007,
+      "learning_rate": 2.5307512937404804e-05,
+      "loss": 0.4718,
+      "step": 485
+    },
+    {
+      "epoch": 1.4992343032159265,
+      "grad_norm": 1.1575231552124023,
+      "learning_rate": 2.5196763939441163e-05,
+      "loss": 0.4851,
+      "step": 490
+    },
+    {
+      "epoch": 1.5145482388973965,
+      "grad_norm": 1.0528773069381714,
+      "learning_rate": 2.5084971770477295e-05,
+      "loss": 0.446,
+      "step": 495
+    },
+    {
+      "epoch": 1.5298621745788668,
+      "grad_norm": 1.1558538675308228,
+      "learning_rate": 2.4972147867313094e-05,
+      "loss": 0.4442,
+      "step": 500
+    },
+    {
+      "epoch": 1.545176110260337,
+      "grad_norm": 1.1066346168518066,
+      "learning_rate": 2.4858303772299115e-05,
+      "loss": 0.4707,
+      "step": 505
+    },
+    {
+      "epoch": 1.560490045941807,
+      "grad_norm": 1.1572457551956177,
+      "learning_rate": 2.4743451132155738e-05,
+      "loss": 0.4463,
+      "step": 510
+    },
+    {
+      "epoch": 1.5758039816232772,
+      "grad_norm": 1.182366132736206,
+      "learning_rate": 2.4627601696781687e-05,
+      "loss": 0.3947,
+      "step": 515
+    },
+    {
+      "epoch": 1.5911179173047474,
+      "grad_norm": 1.0030242204666138,
+      "learning_rate": 2.451076731805193e-05,
+      "loss": 0.4203,
+      "step": 520
+    },
+    {
+      "epoch": 1.6064318529862174,
+      "grad_norm": 1.2075709104537964,
+      "learning_rate": 2.4392959948605214e-05,
+      "loss": 0.4045,
+      "step": 525
+    },
+    {
+      "epoch": 1.6217457886676876,
+      "grad_norm": 1.3273980617523193,
+      "learning_rate": 2.427419164062125e-05,
+      "loss": 0.4074,
+      "step": 530
+    },
+    {
+      "epoch": 1.6370597243491578,
+      "grad_norm": 1.0939041376113892,
+      "learning_rate": 2.415447454458771e-05,
+      "loss": 0.4169,
+      "step": 535
+    },
+    {
+      "epoch": 1.6523736600306278,
+      "grad_norm": 1.3279342651367188,
+      "learning_rate": 2.403382090805721e-05,
+      "loss": 0.4651,
+      "step": 540
+    },
+    {
+      "epoch": 1.667687595712098,
+      "grad_norm": 1.0995126962661743,
+      "learning_rate": 2.391224307439432e-05,
+      "loss": 0.3933,
+      "step": 545
+    },
+    {
+      "epoch": 1.6830015313935682,
+      "grad_norm": 1.221017599105835,
+      "learning_rate": 2.3789753481512777e-05,
+      "loss": 0.4159,
+      "step": 550
+    },
+    {
+      "epoch": 1.6983154670750382,
+      "grad_norm": 1.2090656757354736,
+      "learning_rate": 2.3666364660603058e-05,
+      "loss": 0.4062,
+      "step": 555
+    },
+    {
+      "epoch": 1.7136294027565084,
+      "grad_norm": 1.1304739713668823,
+      "learning_rate": 2.354208923485037e-05,
+      "loss": 0.4137,
+      "step": 560
+    },
+    {
+      "epoch": 1.7289433384379786,
+      "grad_norm": 1.0823583602905273,
+      "learning_rate": 2.341693991814325e-05,
+      "loss": 0.3721,
+      "step": 565
+    },
+    {
+      "epoch": 1.7442572741194486,
+      "grad_norm": 1.3484351634979248,
+      "learning_rate": 2.329092951377289e-05,
+      "loss": 0.3905,
+      "step": 570
+    },
+    {
+      "epoch": 1.7595712098009189,
+      "grad_norm": 1.2623238563537598,
+      "learning_rate": 2.3164070913123288e-05,
+      "loss": 0.3941,
+      "step": 575
+    },
+    {
+      "epoch": 1.774885145482389,
+      "grad_norm": 1.096492052078247,
+      "learning_rate": 2.3036377094352432e-05,
+      "loss": 0.3596,
+      "step": 580
+    },
+    {
+      "epoch": 1.790199081163859,
+      "grad_norm": 1.1170555353164673,
+      "learning_rate": 2.2907861121064567e-05,
+      "loss": 0.3752,
+      "step": 585
+    },
+    {
+      "epoch": 1.8055130168453293,
+      "grad_norm": 1.3183743953704834,
+      "learning_rate": 2.277853614097373e-05,
+      "loss": 0.366,
+      "step": 590
+    },
+    {
+      "epoch": 1.8208269525267995,
+      "grad_norm": 1.2602739334106445,
+      "learning_rate": 2.264841538455869e-05,
+      "loss": 0.386,
+      "step": 595
+    },
+    {
+      "epoch": 1.8361408882082695,
+      "grad_norm": 1.532475471496582,
+      "learning_rate": 2.2517512163709417e-05,
+      "loss": 0.3645,
+      "step": 600
+    },
+    {
+      "epoch": 1.8514548238897397,
+      "grad_norm": 1.1312321424484253,
+      "learning_rate": 2.2385839870365227e-05,
+      "loss": 0.3842,
+      "step": 605
+    },
+    {
+      "epoch": 1.86676875957121,
+      "grad_norm": 1.1037025451660156,
+      "learning_rate": 2.225341197514471e-05,
+      "loss": 0.3528,
+      "step": 610
+    },
+    {
+      "epoch": 1.88208269525268,
+      "grad_norm": 1.1164988279342651,
+      "learning_rate": 2.2120242025967652e-05,
+      "loss": 0.3808,
+      "step": 615
+    },
+    {
+      "epoch": 1.8973966309341501,
+      "grad_norm": 1.316138744354248,
+      "learning_rate": 2.198634364666902e-05,
+      "loss": 0.3882,
+      "step": 620
+    },
+    {
+      "epoch": 1.9127105666156203,
+      "grad_norm": 1.229442834854126,
+      "learning_rate": 2.1851730535605182e-05,
+      "loss": 0.3373,
+      "step": 625
+    },
+    {
+      "epoch": 1.9280245022970903,
+      "grad_norm": 1.0578089952468872,
+      "learning_rate": 2.1716416464252515e-05,
+      "loss": 0.3288,
+      "step": 630
+    },
+    {
+      "epoch": 1.9433384379785605,
+      "grad_norm": 1.1873011589050293,
+      "learning_rate": 2.1580415275798524e-05,
+      "loss": 0.3632,
+      "step": 635
+    },
+    {
+      "epoch": 1.9586523736600308,
+      "grad_norm": 1.2310603857040405,
+      "learning_rate": 2.1443740883725627e-05,
+      "loss": 0.3262,
+      "step": 640
+    },
+    {
+      "epoch": 1.9739663093415007,
+      "grad_norm": 1.2483208179473877,
+      "learning_rate": 2.1306407270387746e-05,
+      "loss": 0.3208,
+      "step": 645
+    },
+    {
+      "epoch": 1.9892802450229707,
+      "grad_norm": 1.269856333732605,
+      "learning_rate": 2.1168428485579864e-05,
+      "loss": 0.3478,
+      "step": 650
+    },
+    {
+      "epoch": 2.003062787136294,
+      "grad_norm": 1.0677129030227661,
+      "learning_rate": 2.1029818645100658e-05,
+      "loss": 0.3044,
+      "step": 655
+    },
+    {
+      "epoch": 2.0183767228177643,
+      "grad_norm": 1.2321646213531494,
+      "learning_rate": 2.0890591929308407e-05,
+      "loss": 0.2795,
+      "step": 660
+    },
+    {
+      "epoch": 2.0336906584992342,
+      "grad_norm": 1.1085941791534424,
+      "learning_rate": 2.075076258167028e-05,
+      "loss": 0.2761,
+      "step": 665
+    },
+    {
+      "epoch": 2.0490045941807042,
+      "grad_norm": 1.2009588479995728,
+      "learning_rate": 2.0610344907305154e-05,
+      "loss": 0.2827,
+      "step": 670
+    },
+    {
+      "epoch": 2.0643185298621747,
+      "grad_norm": 1.0571602582931519,
+      "learning_rate": 2.046935327152018e-05,
+      "loss": 0.2704,
+      "step": 675
+    },
+    {
+      "epoch": 2.0796324655436447,
+      "grad_norm": 1.3781317472457886,
+      "learning_rate": 2.0327802098341116e-05,
+      "loss": 0.2589,
+      "step": 680
+    },
+    {
+      "epoch": 2.0949464012251147,
+      "grad_norm": 1.2654929161071777,
+      "learning_rate": 2.0185705869036704e-05,
+      "loss": 0.258,
+      "step": 685
+    },
+    {
+      "epoch": 2.110260336906585,
+      "grad_norm": 0.9942528605461121,
+      "learning_rate": 2.004307912063717e-05,
+      "loss": 0.2757,
+      "step": 690
+    },
+    {
+      "epoch": 2.125574272588055,
+      "grad_norm": 1.2382903099060059,
+      "learning_rate": 1.989993644444704e-05,
+      "loss": 0.3019,
+      "step": 695
+    },
+    {
+      "epoch": 2.140888208269525,
+      "grad_norm": 1.0553951263427734,
+      "learning_rate": 1.975629248455238e-05,
+      "loss": 0.2578,
+      "step": 700
+    },
+    {
+      "epoch": 2.1562021439509955,
+      "grad_norm": 1.1529054641723633,
+      "learning_rate": 1.9612161936322642e-05,
+      "loss": 0.2792,
+      "step": 705
+    },
+    {
+      "epoch": 2.1715160796324655,
+      "grad_norm": 1.3361412286758423,
+      "learning_rate": 1.9467559544907273e-05,
+      "loss": 0.2516,
+      "step": 710
+    },
+    {
+      "epoch": 2.1868300153139355,
+      "grad_norm": 1.1726348400115967,
+      "learning_rate": 1.932250010372722e-05,
+      "loss": 0.2461,
+      "step": 715
+    },
+    {
+      "epoch": 2.202143950995406,
+      "grad_norm": 1.272294521331787,
+      "learning_rate": 1.9176998452961515e-05,
+      "loss": 0.2612,
+      "step": 720
+    },
+    {
+      "epoch": 2.217457886676876,
+      "grad_norm": 1.0995804071426392,
+      "learning_rate": 1.9031069478029044e-05,
+      "loss": 0.2729,
+      "step": 725
+    },
+    {
+      "epoch": 2.232771822358346,
+      "grad_norm": 1.2195067405700684,
+      "learning_rate": 1.8884728108065723e-05,
+      "loss": 0.2659,
+      "step": 730
+    },
+    {
+      "epoch": 2.2480857580398164,
+      "grad_norm": 1.1243830919265747,
+      "learning_rate": 1.873798931439718e-05,
+      "loss": 0.2736,
+      "step": 735
+    },
+    {
+      "epoch": 2.2633996937212864,
+      "grad_norm": 1.2348970174789429,
+      "learning_rate": 1.8590868109007117e-05,
+      "loss": 0.251,
+      "step": 740
+    },
+    {
+      "epoch": 2.2787136294027563,
+      "grad_norm": 1.087607502937317,
+      "learning_rate": 1.8443379543001556e-05,
+      "loss": 0.2575,
+      "step": 745
+    },
+    {
+      "epoch": 2.294027565084227,
+      "grad_norm": 1.3117108345031738,
+      "learning_rate": 1.8295538705069e-05,
+      "loss": 0.2613,
+      "step": 750
+    },
+    {
+      "epoch": 2.3093415007656968,
+      "grad_norm": 1.1849708557128906,
+      "learning_rate": 1.8147360719936835e-05,
+      "loss": 0.223,
+      "step": 755
+    },
+    {
+      "epoch": 2.3246554364471668,
+      "grad_norm": 1.2858647108078003,
+      "learning_rate": 1.7998860746823995e-05,
+      "loss": 0.2601,
+      "step": 760
+    },
+    {
+      "epoch": 2.339969372128637,
+      "grad_norm": 1.0667446851730347,
+      "learning_rate": 1.7850053977890114e-05,
+      "loss": 0.2543,
+      "step": 765
+    },
+    {
+      "epoch": 2.355283307810107,
+      "grad_norm": 1.1253725290298462,
+      "learning_rate": 1.770095563668129e-05,
+      "loss": 0.219,
+      "step": 770
+    },
+    {
+      "epoch": 2.370597243491577,
+      "grad_norm": 1.078885555267334,
+      "learning_rate": 1.7551580976572686e-05,
+      "loss": 0.2468,
+      "step": 775
+    },
+    {
+      "epoch": 2.3859111791730476,
+      "grad_norm": 1.2456717491149902,
+      "learning_rate": 1.740194527920802e-05,
+      "loss": 0.2257,
+      "step": 780
+    },
+    {
+      "epoch": 2.4012251148545176,
+      "grad_norm": 1.4209325313568115,
+      "learning_rate": 1.7252063852936187e-05,
+      "loss": 0.2296,
+      "step": 785
+    },
+    {
+      "epoch": 2.4165390505359876,
+      "grad_norm": 0.9507177472114563,
+      "learning_rate": 1.7101952031245187e-05,
+      "loss": 0.2168,
+      "step": 790
+    },
+    {
+      "epoch": 2.431852986217458,
+      "grad_norm": 1.307239294052124,
+      "learning_rate": 1.69516251711934e-05,
+      "loss": 0.2191,
+      "step": 795
+    },
+    {
+      "epoch": 2.447166921898928,
+      "grad_norm": 1.2544313669204712,
+      "learning_rate": 1.6801098651838523e-05,
+      "loss": 0.2272,
+      "step": 800
+    },
+    {
+      "epoch": 2.462480857580398,
+      "grad_norm": 1.2705137729644775,
+      "learning_rate": 1.665038787266423e-05,
+      "loss": 0.2415,
+      "step": 805
+    },
+    {
+      "epoch": 2.4777947932618685,
+      "grad_norm": 1.0867475271224976,
+      "learning_rate": 1.649950825200472e-05,
+      "loss": 0.2292,
+      "step": 810
+    },
+    {
+      "epoch": 2.4931087289433385,
+      "grad_norm": 1.1694650650024414,
+      "learning_rate": 1.634847522546739e-05,
+      "loss": 0.2242,
+      "step": 815
+    },
+    {
+      "epoch": 2.5084226646248085,
+      "grad_norm": 1.2106242179870605,
+      "learning_rate": 1.6197304244353677e-05,
+      "loss": 0.222,
+      "step": 820
+    },
+    {
+      "epoch": 2.523736600306279,
+      "grad_norm": 1.1909250020980835,
+      "learning_rate": 1.6046010774078348e-05,
+      "loss": 0.218,
+      "step": 825
+    },
+    {
+      "epoch": 2.539050535987749,
+      "grad_norm": 1.0819553136825562,
+      "learning_rate": 1.5894610292587304e-05,
+      "loss": 0.235,
+      "step": 830
+    },
+    {
+      "epoch": 2.554364471669219,
+      "grad_norm": 1.1172194480895996,
+      "learning_rate": 1.574311828877416e-05,
+      "loss": 0.2246,
+      "step": 835
+    },
+    {
+      "epoch": 2.569678407350689,
+      "grad_norm": 1.6138761043548584,
+      "learning_rate": 1.5591550260895612e-05,
+      "loss": 0.2149,
+      "step": 840
+    },
+    {
+      "epoch": 2.5849923430321593,
+      "grad_norm": 1.1283706426620483,
+      "learning_rate": 1.5439921714985954e-05,
+      "loss": 0.1873,
+      "step": 845
+    },
+    {
+      "epoch": 2.6003062787136293,
+      "grad_norm": 1.1196907758712769,
+      "learning_rate": 1.528824816327071e-05,
+      "loss": 0.2358,
+      "step": 850
+    },
+    {
+      "epoch": 2.6156202143950997,
+      "grad_norm": 1.233160138130188,
+      "learning_rate": 1.5136545122579694e-05,
+      "loss": 0.1997,
+      "step": 855
+    },
+    {
+      "epoch": 2.6309341500765697,
+      "grad_norm": 1.1278220415115356,
+      "learning_rate": 1.4984828112759551e-05,
+      "loss": 0.2217,
+      "step": 860
+    },
+    {
+      "epoch": 2.6462480857580397,
+      "grad_norm": 1.2449628114700317,
+      "learning_rate": 1.4833112655086045e-05,
+      "loss": 0.226,
+      "step": 865
+    },
+    {
+      "epoch": 2.6615620214395097,
+      "grad_norm": 1.0520964860916138,
+      "learning_rate": 1.4681414270676125e-05,
+      "loss": 0.1961,
+      "step": 870
+    },
+    {
+      "epoch": 2.67687595712098,
+      "grad_norm": 1.0986305475234985,
+      "learning_rate": 1.4529748478900096e-05,
+      "loss": 0.2014,
+      "step": 875
+    },
+    {
+      "epoch": 2.69218989280245,
+      "grad_norm": 1.0960958003997803,
+      "learning_rate": 1.4378130795793884e-05,
+      "loss": 0.1939,
+      "step": 880
+    },
+    {
+      "epoch": 2.7075038284839206,
+      "grad_norm": 1.0452426671981812,
+      "learning_rate": 1.4226576732471724e-05,
+      "loss": 0.2051,
+      "step": 885
+    },
+    {
+      "epoch": 2.7228177641653906,
+      "grad_norm": 1.2079676389694214,
+      "learning_rate": 1.4075101793539254e-05,
+      "loss": 0.18,
+      "step": 890
+    },
+    {
+      "epoch": 2.7381316998468606,
+      "grad_norm": 1.0217554569244385,
+      "learning_rate": 1.3923721475507384e-05,
+      "loss": 0.2024,
+      "step": 895
+    },
+    {
+      "epoch": 2.7534456355283305,
+      "grad_norm": 1.0963138341903687,
+      "learning_rate": 1.3772451265206905e-05,
+      "loss": 0.2098,
+      "step": 900
+    },
+    {
+      "epoch": 2.768759571209801,
+      "grad_norm": 1.184330940246582,
+      "learning_rate": 1.362130663820414e-05,
+      "loss": 0.1946,
+      "step": 905
+    },
+    {
+      "epoch": 2.784073506891271,
+      "grad_norm": 1.3663582801818848,
+      "learning_rate": 1.3470303057217711e-05,
+      "loss": 0.2119,
+      "step": 910
+    },
+    {
+      "epoch": 2.7993874425727414,
+      "grad_norm": 1.4678473472595215,
+      "learning_rate": 1.3319455970536674e-05,
+      "loss": 0.2213,
+      "step": 915
+    },
+    {
+      "epoch": 2.8147013782542114,
+      "grad_norm": 1.1842464208602905,
+      "learning_rate": 1.3168780810440038e-05,
+      "loss": 0.1932,
+      "step": 920
+    },
+    {
+      "epoch": 2.8300153139356814,
+      "grad_norm": 1.1453865766525269,
+      "learning_rate": 1.3018292991618046e-05,
+      "loss": 0.1916,
+      "step": 925
+    },
+    {
+      "epoch": 2.8453292496171514,
+      "grad_norm": 1.286776065826416,
+      "learning_rate": 1.2868007909595131e-05,
+      "loss": 0.1943,
+      "step": 930
+    },
+    {
+      "epoch": 2.860643185298622,
+      "grad_norm": 0.9827431440353394,
+      "learning_rate": 1.271794093915494e-05,
+      "loss": 0.196,
+      "step": 935
+    },
+    {
+      "epoch": 2.875957120980092,
+      "grad_norm": 1.1773149967193604,
+      "learning_rate": 1.256810743276738e-05,
+      "loss": 0.1945,
+      "step": 940
+    },
+    {
+      "epoch": 2.8912710566615623,
+      "grad_norm": 1.141097903251648,
+      "learning_rate": 1.2418522719018047e-05,
+      "loss": 0.181,
+      "step": 945
+    },
+    {
+      "epoch": 2.9065849923430322,
+      "grad_norm": 1.1019505262374878,
+      "learning_rate": 1.2269202101039999e-05,
+      "loss": 0.1769,
+      "step": 950
+    },
+    {
+      "epoch": 2.9218989280245022,
+      "grad_norm": 1.0214898586273193,
+      "learning_rate": 1.212016085494823e-05,
+      "loss": 0.2148,
+      "step": 955
+    },
+    {
+      "epoch": 2.9372128637059722,
+      "grad_norm": 1.0284066200256348,
+      "learning_rate": 1.1971414228276823e-05,
+      "loss": 0.1777,
+      "step": 960
+    },
+    {
+      "epoch": 2.9525267993874427,
+      "grad_norm": 1.144411563873291,
+      "learning_rate": 1.1822977438419092e-05,
+      "loss": 0.1839,
+      "step": 965
+    },
+    {
+      "epoch": 2.9678407350689127,
+      "grad_norm": 1.1772691011428833,
+      "learning_rate": 1.1674865671070751e-05,
+      "loss": 0.2072,
+      "step": 970
+    },
+    {
+      "epoch": 2.983154670750383,
+      "grad_norm": 1.053539752960205,
+      "learning_rate": 1.1527094078676395e-05,
+      "loss": 0.1685,
+      "step": 975
+    },
+    {
+      "epoch": 2.998468606431853,
+      "grad_norm": 1.1275657415390015,
+      "learning_rate": 1.1379677778879283e-05,
+      "loss": 0.1848,
+      "step": 980
+    },
+    {
+      "epoch": 3.012251148545176,
+      "grad_norm": 0.9341105818748474,
+      "learning_rate": 1.1232631852974814e-05,
+      "loss": 0.1463,
+      "step": 985
+    },
+    {
+      "epoch": 3.027565084226646,
+      "grad_norm": 1.0302433967590332,
+      "learning_rate": 1.1085971344367585e-05,
+      "loss": 0.124,
+      "step": 990
+    },
+    {
+      "epoch": 3.0428790199081166,
+      "grad_norm": 1.0990079641342163,
+      "learning_rate": 1.0939711257032429e-05,
+      "loss": 0.138,
+      "step": 995
+    },
+    {
+      "epoch": 3.0581929555895866,
+      "grad_norm": 1.0678660869598389,
+      "learning_rate": 1.0793866553979411e-05,
+      "loss": 0.1297,
+      "step": 1000
+    },
+    {
+      "epoch": 3.0735068912710566,
+      "grad_norm": 1.1288793087005615,
+      "learning_rate": 1.0648452155723103e-05,
+      "loss": 0.1418,
+      "step": 1005
+    },
+    {
+      "epoch": 3.088820826952527,
+      "grad_norm": 1.0335837602615356,
+      "learning_rate": 1.0503482938756091e-05,
+      "loss": 0.1367,
+      "step": 1010
+    },
+    {
+      "epoch": 3.104134762633997,
+      "grad_norm": 0.9697948694229126,
+      "learning_rate": 1.0358973734027098e-05,
+      "loss": 0.1264,
+      "step": 1015
+    },
+    {
+      "epoch": 3.119448698315467,
+      "grad_norm": 1.1903094053268433,
+      "learning_rate": 1.0214939325423686e-05,
+      "loss": 0.1372,
+      "step": 1020
+    },
+    {
+      "epoch": 3.134762633996937,
+      "grad_norm": 1.1920007467269897,
+      "learning_rate": 1.0071394448259834e-05,
+      "loss": 0.1284,
+      "step": 1025
+    },
+    {
+      "epoch": 3.1500765696784074,
+      "grad_norm": 0.9497950673103333,
+      "learning_rate": 9.928353787768404e-06,
+      "loss": 0.1293,
+      "step": 1030
+    },
+    {
+      "epoch": 3.1653905053598774,
+      "grad_norm": 1.056449294090271,
+      "learning_rate": 9.785831977598855e-06,
+      "loss": 0.1191,
+      "step": 1035
+    },
+    {
+      "epoch": 3.1807044410413474,
+      "grad_norm": 1.0650913715362549,
+      "learning_rate": 9.643843598320078e-06,
+      "loss": 0.1298,
+      "step": 1040
+    },
+    {
+      "epoch": 3.196018376722818,
+      "grad_norm": 0.9316422939300537,
+      "learning_rate": 9.50240317592881e-06,
+      "loss": 0.1318,
+      "step": 1045
+    },
+    {
+      "epoch": 3.211332312404288,
+      "grad_norm": 1.1120059490203857,
+      "learning_rate": 9.36152518036353e-06,
+      "loss": 0.1315,
+      "step": 1050
+    },
+    {
+      "epoch": 3.226646248085758,
+      "grad_norm": 1.050228238105774,
+      "learning_rate": 9.22122402402414e-06,
+      "loss": 0.1145,
+      "step": 1055
+    },
+    {
+      "epoch": 3.2419601837672283,
+      "grad_norm": 1.0715663433074951,
+      "learning_rate": 9.08151406029749e-06,
+      "loss": 0.1299,
+      "step": 1060
+    },
+    {
+      "epoch": 3.2572741194486983,
+      "grad_norm": 1.7908285856246948,
+      "learning_rate": 8.942409582089018e-06,
+      "loss": 0.1442,
+      "step": 1065
+    },
+    {
+      "epoch": 3.2725880551301683,
+      "grad_norm": 1.0539591312408447,
+      "learning_rate": 8.803924820360469e-06,
+      "loss": 0.1291,
+      "step": 1070
+    },
+    {
+      "epoch": 3.2879019908116387,
+      "grad_norm": 0.9626232981681824,
+      "learning_rate": 8.666073942674064e-06,
+      "loss": 0.1401,
+      "step": 1075
+    },
+    {
+      "epoch": 3.3032159264931087,
+      "grad_norm": 1.1888229846954346,
+      "learning_rate": 8.528871051743038e-06,
+      "loss": 0.1368,
+      "step": 1080
+    },
+    {
+      "epoch": 3.3185298621745787,
+      "grad_norm": 1.115984559059143,
+      "learning_rate": 8.392330183988942e-06,
+      "loss": 0.1272,
+      "step": 1085
+    },
+    {
+      "epoch": 3.333843797856049,
+      "grad_norm": 0.9548808932304382,
+      "learning_rate": 8.256465308105608e-06,
+      "loss": 0.1236,
+      "step": 1090
+    },
+    {
+      "epoch": 3.349157733537519,
+      "grad_norm": 1.2673202753067017,
+      "learning_rate": 8.121290323630116e-06,
+      "loss": 0.1233,
+      "step": 1095
+    },
+    {
+      "epoch": 3.364471669218989,
+      "grad_norm": 1.1379413604736328,
+      "learning_rate": 7.986819059520803e-06,
+      "loss": 0.1312,
+      "step": 1100
+    },
+    {
+      "epoch": 3.3797856049004595,
+      "grad_norm": 1.0691787004470825,
+      "learning_rate": 7.853065272742536e-06,
+      "loss": 0.1318,
+      "step": 1105
+    },
+    {
+      "epoch": 3.3950995405819295,
+      "grad_norm": 1.019787073135376,
+      "learning_rate": 7.720042646859247e-06,
+      "loss": 0.1341,
+      "step": 1110
+    },
+    {
+      "epoch": 3.4104134762633995,
+      "grad_norm": 1.0537774562835693,
+      "learning_rate": 7.587764790634143e-06,
+      "loss": 0.1391,
+      "step": 1115
+    },
+    {
+      "epoch": 3.42572741194487,
+      "grad_norm": 1.0213468074798584,
+      "learning_rate": 7.456245236637392e-06,
+      "loss": 0.1144,
+      "step": 1120
+    },
+    {
+      "epoch": 3.44104134762634,
+      "grad_norm": 1.038906455039978,
+      "learning_rate": 7.325497439861725e-06,
+      "loss": 0.1216,
+      "step": 1125
+    },
+    {
+      "epoch": 3.45635528330781,
+      "grad_norm": 1.2020183801651,
+      "learning_rate": 7.195534776345923e-06,
+      "loss": 0.1266,
+      "step": 1130
+    },
+    {
+      "epoch": 3.4716692189892804,
+      "grad_norm": 0.8986299633979797,
+      "learning_rate": 7.066370541806418e-06,
+      "loss": 0.1341,
+      "step": 1135
+    },
+    {
+      "epoch": 3.4869831546707504,
+      "grad_norm": 0.9827359914779663,
+      "learning_rate": 6.938017950277025e-06,
+      "loss": 0.1213,
+      "step": 1140
+    },
+    {
+      "epoch": 3.5022970903522204,
+      "grad_norm": 1.0107229948043823,
+      "learning_rate": 6.810490132757175e-06,
+      "loss": 0.1288,
+      "step": 1145
+    },
+    {
+      "epoch": 3.517611026033691,
+      "grad_norm": 1.0034672021865845,
+      "learning_rate": 6.683800135868487e-06,
+      "loss": 0.1193,
+      "step": 1150
+    },
+    {
+      "epoch": 3.532924961715161,
+      "grad_norm": 1.0432562828063965,
+      "learning_rate": 6.557960920520089e-06,
+      "loss": 0.1337,
+      "step": 1155
+    },
+    {
+      "epoch": 3.548238897396631,
+      "grad_norm": 0.9052329659461975,
+      "learning_rate": 6.432985360582641e-06,
+      "loss": 0.1091,
+      "step": 1160
+    },
+    {
+      "epoch": 3.5635528330781012,
+      "grad_norm": 0.9641390442848206,
+      "learning_rate": 6.308886241571299e-06,
+      "loss": 0.1101,
+      "step": 1165
+    },
+    {
+      "epoch": 3.578866768759571,
+      "grad_norm": 1.187159776687622,
+      "learning_rate": 6.18567625933769e-06,
+      "loss": 0.122,
+      "step": 1170
+    },
+    {
+      "epoch": 3.594180704441041,
+      "grad_norm": 0.9154672622680664,
+      "learning_rate": 6.063368018771096e-06,
+      "loss": 0.1059,
+      "step": 1175
+    },
+    {
+      "epoch": 3.6094946401225116,
+      "grad_norm": 0.9932954907417297,
+      "learning_rate": 5.941974032508896e-06,
+      "loss": 0.1183,
+      "step": 1180
+    },
+    {
+      "epoch": 3.6248085758039816,
+      "grad_norm": 0.9219599366188049,
+      "learning_rate": 5.821506719656485e-06,
+      "loss": 0.1058,
+      "step": 1185
+    },
+    {
+      "epoch": 3.6401225114854516,
+      "grad_norm": 0.8277255296707153,
+      "learning_rate": 5.70197840451675e-06,
+      "loss": 0.1076,
+      "step": 1190
+    },
+    {
+      "epoch": 3.655436447166922,
+      "grad_norm": 0.8804378509521484,
+      "learning_rate": 5.5834013153292396e-06,
+      "loss": 0.113,
+      "step": 1195
+    },
+    {
+      "epoch": 3.670750382848392,
+      "grad_norm": 0.9384981989860535,
+      "learning_rate": 5.465787583019156e-06,
+      "loss": 0.101,
+      "step": 1200
+    },
+    {
+      "epoch": 3.686064318529862,
+      "grad_norm": 1.1971505880355835,
+      "learning_rate": 5.349149239956339e-06,
+      "loss": 0.1061,
+      "step": 1205
+    },
+    {
+      "epoch": 3.701378254211332,
+      "grad_norm": 1.0984221696853638,
+      "learning_rate": 5.23349821872427e-06,
+      "loss": 0.1276,
+      "step": 1210
+    },
+    {
+      "epoch": 3.7166921898928025,
+      "grad_norm": 0.9261698126792908,
+      "learning_rate": 5.1188463508993396e-06,
+      "loss": 0.1094,
+      "step": 1215
+    },
+    {
+      "epoch": 3.7320061255742725,
+      "grad_norm": 0.9417651891708374,
+      "learning_rate": 5.0052053658404285e-06,
+      "loss": 0.1085,
+      "step": 1220
+    },
+    {
+      "epoch": 3.747320061255743,
+      "grad_norm": 0.9424726963043213,
+      "learning_rate": 4.892586889488935e-06,
+      "loss": 0.1183,
+      "step": 1225
+    },
+    {
+      "epoch": 3.762633996937213,
+      "grad_norm": 0.9524731636047363,
+      "learning_rate": 4.7810024431794e-06,
+      "loss": 0.1177,
+      "step": 1230
+    },
+    {
+      "epoch": 3.777947932618683,
+      "grad_norm": 0.8342095017433167,
+      "learning_rate": 4.670463442460842e-06,
+      "loss": 0.1,
+      "step": 1235
+    },
+    {
+      "epoch": 3.793261868300153,
+      "grad_norm": 0.9804592132568359,
+      "learning_rate": 4.560981195928869e-06,
+      "loss": 0.1109,
+      "step": 1240
+    },
+    {
+      "epoch": 3.8085758039816233,
+      "grad_norm": 1.1521655321121216,
+      "learning_rate": 4.452566904068778e-06,
+      "loss": 0.1081,
+      "step": 1245
+    },
+    {
+      "epoch": 3.8238897396630933,
+      "grad_norm": 1.014464259147644,
+      "learning_rate": 4.345231658109707e-06,
+      "loss": 0.1126,
+      "step": 1250
+    },
+    {
+      "epoch": 3.8392036753445637,
+      "grad_norm": 1.0303317308425903,
+      "learning_rate": 4.238986438889942e-06,
+      "loss": 0.1125,
+      "step": 1255
+    },
+    {
+      "epoch": 3.8545176110260337,
+      "grad_norm": 1.0285574197769165,
+      "learning_rate": 4.133842115733537e-06,
+      "loss": 0.1163,
+      "step": 1260
+    },
+    {
+      "epoch": 3.8698315467075037,
+      "grad_norm": 0.921039342880249,
+      "learning_rate": 4.029809445338359e-06,
+      "loss": 0.0999,
+      "step": 1265
+    },
+    {
+      "epoch": 3.8851454823889737,
+      "grad_norm": 0.8448448777198792,
+      "learning_rate": 3.926899070675577e-06,
+      "loss": 0.1109,
+      "step": 1270
+    },
+    {
+      "epoch": 3.900459418070444,
+      "grad_norm": 0.9851784706115723,
+      "learning_rate": 3.825121519900913e-06,
+      "loss": 0.1054,
+      "step": 1275
+    },
+    {
+      "epoch": 3.915773353751914,
+      "grad_norm": 1.327043890953064,
+      "learning_rate": 3.7244872052775165e-06,
+      "loss": 0.1034,
+      "step": 1280
+    },
+    {
+      "epoch": 3.9310872894333846,
+      "grad_norm": 0.8687987923622131,
+      "learning_rate": 3.6250064221107638e-06,
+      "loss": 0.0915,
+      "step": 1285
+    },
+    {
+      "epoch": 3.9464012251148546,
+      "grad_norm": 0.9403600692749023,
+      "learning_rate": 3.5266893476950096e-06,
+      "loss": 0.1012,
+      "step": 1290
+    },
+    {
+      "epoch": 3.9617151607963246,
+      "grad_norm": 0.7973408699035645,
+      "learning_rate": 3.4295460402724148e-06,
+      "loss": 0.0837,
+      "step": 1295
+    },
+    {
+      "epoch": 3.9770290964777946,
+      "grad_norm": 0.8510920405387878,
+      "learning_rate": 3.3335864380039123e-06,
+      "loss": 0.088,
+      "step": 1300
+    },
+    {
+      "epoch": 3.992343032159265,
+      "grad_norm": 0.9783728122711182,
+      "learning_rate": 3.238820357952539e-06,
+      "loss": 0.0981,
+      "step": 1305
+    },
+    {
+      "epoch": 4.006125574272588,
+      "grad_norm": 0.7199473977088928,
+      "learning_rate": 3.1452574950790776e-06,
+      "loss": 0.0949,
+      "step": 1310
+    },
+    {
+      "epoch": 4.0214395099540585,
+      "grad_norm": 0.8278849124908447,
+      "learning_rate": 3.0529074212502323e-06,
+      "loss": 0.1102,
+      "step": 1315
+    },
+    {
+      "epoch": 4.0367534456355285,
+      "grad_norm": 0.7794520258903503,
+      "learning_rate": 2.961779584259384e-06,
+      "loss": 0.0804,
+      "step": 1320
+    },
+    {
+      "epoch": 4.0520673813169985,
+      "grad_norm": 0.796407163143158,
+      "learning_rate": 2.8718833068600575e-06,
+      "loss": 0.0765,
+      "step": 1325
+    },
+    {
+      "epoch": 4.0673813169984685,
+      "grad_norm": 0.7140238285064697,
+      "learning_rate": 2.783227785812129e-06,
+      "loss": 0.0735,
+      "step": 1330
+    },
+    {
+      "epoch": 4.0826952526799385,
+      "grad_norm": 0.6903073787689209,
+      "learning_rate": 2.6958220909410112e-06,
+      "loss": 0.0781,
+      "step": 1335
+    },
+    {
+      "epoch": 4.0980091883614085,
+      "grad_norm": 0.7919360399246216,
+      "learning_rate": 2.6096751642097244e-06,
+      "loss": 0.0821,
+      "step": 1340
+    },
+    {
+      "epoch": 4.113323124042879,
+      "grad_norm": 0.7220079898834229,
+      "learning_rate": 2.524795818804126e-06,
+      "loss": 0.0747,
+      "step": 1345
+    },
+    {
+      "epoch": 4.128637059724349,
+      "grad_norm": 0.8579390645027161,
+      "learning_rate": 2.441192738231271e-06,
+      "loss": 0.0836,
+      "step": 1350
+    },
+    {
+      "epoch": 4.143950995405819,
+      "grad_norm": 0.731046199798584,
+      "learning_rate": 2.3588744754310744e-06,
+      "loss": 0.0775,
+      "step": 1355
+    },
+    {
+      "epoch": 4.159264931087289,
+      "grad_norm": 0.8168486952781677,
+      "learning_rate": 2.2778494519012667e-06,
+      "loss": 0.085,
+      "step": 1360
+    },
+    {
+      "epoch": 4.174578866768759,
+      "grad_norm": 0.777069628238678,
+      "learning_rate": 2.1981259568358956e-06,
+      "loss": 0.0773,
+      "step": 1365
+    },
+    {
+      "epoch": 4.189892802450229,
+      "grad_norm": 0.9429895877838135,
+      "learning_rate": 2.1197121462772665e-06,
+      "loss": 0.0788,
+      "step": 1370
+    },
+    {
+      "epoch": 4.2052067381317,
+      "grad_norm": 0.7801960706710815,
+      "learning_rate": 2.04261604228156e-06,
+      "loss": 0.079,
+      "step": 1375
+    },
+    {
+      "epoch": 4.22052067381317,
+      "grad_norm": 0.8091967105865479,
+      "learning_rate": 1.9668455320981406e-06,
+      "loss": 0.0741,
+      "step": 1380
+    },
+    {
+      "epoch": 4.23583460949464,
+      "grad_norm": 0.7686964273452759,
+      "learning_rate": 1.8924083673626686e-06,
+      "loss": 0.0821,
+      "step": 1385
+    },
+    {
+      "epoch": 4.25114854517611,
+      "grad_norm": 0.6906757950782776,
+      "learning_rate": 1.8193121633040472e-06,
+      "loss": 0.0804,
+      "step": 1390
+    },
+    {
+      "epoch": 4.26646248085758,
+      "grad_norm": 0.8286918997764587,
+      "learning_rate": 1.7475643979653926e-06,
+      "loss": 0.0828,
+      "step": 1395
+    },
+    {
+      "epoch": 4.28177641653905,
+      "grad_norm": 0.7227317690849304,
+      "learning_rate": 1.677172411438969e-06,
+      "loss": 0.0761,
+      "step": 1400
+    },
+    {
+      "epoch": 4.297090352220521,
+      "grad_norm": 0.7851067781448364,
+      "learning_rate": 1.6081434051152805e-06,
+      "loss": 0.0816,
+      "step": 1405
+    },
+    {
+      "epoch": 4.312404287901991,
+      "grad_norm": 0.7805889844894409,
+      "learning_rate": 1.5404844409463298e-06,
+      "loss": 0.0821,
+      "step": 1410
+    },
+    {
+      "epoch": 4.327718223583461,
+      "grad_norm": 0.7815679907798767,
+      "learning_rate": 1.4742024407231709e-06,
+      "loss": 0.08,
+      "step": 1415
+    },
+    {
+      "epoch": 4.343032159264931,
+      "grad_norm": 0.7986721992492676,
+      "learning_rate": 1.4093041853677474e-06,
+      "loss": 0.0849,
+      "step": 1420
+    },
+    {
+      "epoch": 4.358346094946401,
+      "grad_norm": 0.8983235955238342,
+      "learning_rate": 1.3457963142392172e-06,
+      "loss": 0.0749,
+      "step": 1425
+    },
+    {
+      "epoch": 4.373660030627871,
+      "grad_norm": 1.0158237218856812,
+      "learning_rate": 1.2836853244546875e-06,
+      "loss": 0.0846,
+      "step": 1430
+    },
+    {
+      "epoch": 4.388973966309342,
+      "grad_norm": 0.7490912675857544,
+      "learning_rate": 1.222977570224545e-06,
+      "loss": 0.0861,
+      "step": 1435
+    },
+    {
+      "epoch": 4.404287901990812,
+      "grad_norm": 0.7817654013633728,
+      "learning_rate": 1.163679262202389e-06,
+      "loss": 0.0836,
+      "step": 1440
+    },
+    {
+      "epoch": 4.419601837672282,
+      "grad_norm": 0.7339651584625244,
+      "learning_rate": 1.1057964668496729e-06,
+      "loss": 0.0828,
+      "step": 1445
+    },
+    {
+      "epoch": 4.434915773353752,
+      "grad_norm": 0.7097742557525635,
+      "learning_rate": 1.0493351058150446e-06,
+      "loss": 0.0708,
+      "step": 1450
+    },
+    {
+      "epoch": 4.450229709035222,
+      "grad_norm": 0.776528537273407,
+      "learning_rate": 9.943009553285837e-07,
+      "loss": 0.0778,
+      "step": 1455
+    },
+    {
+      "epoch": 4.465543644716692,
+      "grad_norm": 0.7682094573974609,
+      "learning_rate": 9.406996456108263e-07,
+      "loss": 0.0794,
+      "step": 1460
+    },
+    {
+      "epoch": 4.480857580398163,
+      "grad_norm": 0.7762686610221863,
+      "learning_rate": 8.88536660296802e-07,
+      "loss": 0.0746,
+      "step": 1465
+    },
+    {
+      "epoch": 4.496171516079633,
+      "grad_norm": 0.8416975736618042,
+      "learning_rate": 8.378173358750102e-07,
+      "loss": 0.0796,
+      "step": 1470
+    },
+    {
+      "epoch": 4.511485451761103,
+      "grad_norm": 0.8884332776069641,
+      "learning_rate": 7.885468611415036e-07,
+      "loss": 0.0835,
+      "step": 1475
+    },
+    {
+      "epoch": 4.526799387442573,
+      "grad_norm": 0.7939807176589966,
+      "learning_rate": 7.4073027666902e-07,
+      "loss": 0.0837,
+      "step": 1480
+    },
+    {
+      "epoch": 4.542113323124043,
+      "grad_norm": 0.79791259765625,
+      "learning_rate": 6.943724742913404e-07,
+      "loss": 0.073,
+      "step": 1485
+    },
+    {
+      "epoch": 4.557427258805513,
+      "grad_norm": 0.8921972513198853,
+      "learning_rate": 6.494781966028173e-07,
+      "loss": 0.0778,
+      "step": 1490
+    },
+    {
+      "epoch": 4.572741194486984,
+      "grad_norm": 0.8343205451965332,
+      "learning_rate": 6.060520364731903e-07,
+      "loss": 0.0866,
+      "step": 1495
+    },
+    {
+      "epoch": 4.588055130168454,
+      "grad_norm": 0.7539684176445007,
+      "learning_rate": 5.640984365777202e-07,
+      "loss": 0.0744,
+      "step": 1500
+    },
+    {
+      "epoch": 4.603369065849924,
+      "grad_norm": 0.7956764101982117,
+      "learning_rate": 5.236216889426898e-07,
+      "loss": 0.0721,
+      "step": 1505
+    },
+    {
+      "epoch": 4.6186830015313936,
+      "grad_norm": 0.8460783958435059,
+      "learning_rate": 4.846259345062914e-07,
+      "loss": 0.0836,
+      "step": 1510
+    },
+    {
+      "epoch": 4.6339969372128635,
+      "grad_norm": 0.7418811321258545,
+      "learning_rate": 4.4711516269502094e-07,
+      "loss": 0.0786,
+      "step": 1515
+    },
+    {
+      "epoch": 4.6493108728943335,
+      "grad_norm": 0.8257374167442322,
+      "learning_rate": 4.110932110155158e-07,
+      "loss": 0.0794,
+      "step": 1520
+    },
+    {
+      "epoch": 4.6646248085758035,
+      "grad_norm": 0.6829298734664917,
+      "learning_rate": 3.7656376466198073e-07,
+      "loss": 0.0701,
+      "step": 1525
+    },
+    {
+      "epoch": 4.679938744257274,
+      "grad_norm": 0.7458698749542236,
+      "learning_rate": 3.435303561391695e-07,
+      "loss": 0.0753,
+      "step": 1530
+    },
+    {
+      "epoch": 4.695252679938744,
+      "grad_norm": 0.7856422662734985,
+      "learning_rate": 3.1199636490099713e-07,
+      "loss": 0.0759,
+      "step": 1535
+    },
+    {
+      "epoch": 4.710566615620214,
+      "grad_norm": 0.7594045996665955,
+      "learning_rate": 2.819650170048066e-07,
+      "loss": 0.0659,
+      "step": 1540
+    },
+    {
+      "epoch": 4.725880551301684,
+      "grad_norm": 0.7138618230819702,
+      "learning_rate": 2.53439384781331e-07,
+      "loss": 0.0861,
+      "step": 1545
+    },
+    {
+      "epoch": 4.741194486983154,
+      "grad_norm": 0.6646899580955505,
+      "learning_rate": 2.264223865203785e-07,
+      "loss": 0.0803,
+      "step": 1550
+    },
+    {
+      "epoch": 4.756508422664625,
+      "grad_norm": 0.8440792560577393,
+      "learning_rate": 2.0091678617228437e-07,
+      "loss": 0.0765,
+      "step": 1555
+    },
+    {
+      "epoch": 4.771822358346095,
+      "grad_norm": 0.7916809916496277,
+      "learning_rate": 1.7692519306514386e-07,
+      "loss": 0.0779,
+      "step": 1560
+    },
+    {
+      "epoch": 4.787136294027565,
+      "grad_norm": 0.7673787474632263,
+      "learning_rate": 1.544500616378691e-07,
+      "loss": 0.0745,
+      "step": 1565
+    },
+    {
+      "epoch": 4.802450229709035,
+      "grad_norm": 0.6981791257858276,
+      "learning_rate": 1.3349369118908328e-07,
+      "loss": 0.0757,
+      "step": 1570
+    },
+    {
+      "epoch": 4.817764165390505,
+      "grad_norm": 0.7295683026313782,
+      "learning_rate": 1.1405822564190871e-07,
+      "loss": 0.0706,
+      "step": 1575
+    },
+    {
+      "epoch": 4.833078101071975,
+      "grad_norm": 0.7721351385116577,
+      "learning_rate": 9.614565332461412e-08,
+      "loss": 0.0761,
+      "step": 1580
+    },
+    {
+      "epoch": 4.848392036753445,
+      "grad_norm": 0.7022418975830078,
+      "learning_rate": 7.975780676721722e-08,
+      "loss": 0.0784,
+      "step": 1585
+    },
+    {
+      "epoch": 4.863705972434916,
+      "grad_norm": 0.7544510364532471,
+      "learning_rate": 6.489636251399811e-08,
+      "loss": 0.0785,
+      "step": 1590
+    },
+    {
+      "epoch": 4.879019908116386,
+      "grad_norm": 0.6989184021949768,
+      "learning_rate": 5.156284095199304e-08,
+      "loss": 0.0713,
+      "step": 1595
+    },
+    {
+      "epoch": 4.894333843797856,
+      "grad_norm": 0.8224272727966309,
+      "learning_rate": 3.9758606155438916e-08,
+      "loss": 0.079,
+      "step": 1600
+    },
+    {
+      "epoch": 4.909647779479326,
+      "grad_norm": 0.7936996817588806,
+      "learning_rate": 2.9484865746236545e-08,
+      "loss": 0.0824,
+      "step": 1605
+    },
+    {
+      "epoch": 4.924961715160796,
+      "grad_norm": 0.6947341561317444,
+      "learning_rate": 2.0742670770394488e-08,
+      "loss": 0.0797,
+      "step": 1610
+    },
+    {
+      "epoch": 4.940275650842267,
+      "grad_norm": 0.8294661045074463,
+      "learning_rate": 1.3532915590510064e-08,
+      "loss": 0.0745,
+      "step": 1615
+    },
+    {
+      "epoch": 4.955589586523737,
+      "grad_norm": 0.7988950610160828,
+      "learning_rate": 7.856337794269198e-09,
+      "loss": 0.0858,
+      "step": 1620
+    },
+    {
+      "epoch": 4.970903522205207,
+      "grad_norm": 0.7842313051223755,
+      "learning_rate": 3.713518118990122e-09,
+      "loss": 0.0731,
+      "step": 1625
+    },
+    {
+      "epoch": 4.986217457886677,
+      "grad_norm": 0.6295393109321594,
+      "learning_rate": 1.1048803922109052e-09,
+      "loss": 0.0802,
+      "step": 1630
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.8685061931610107,
+      "learning_rate": 3.069148832413049e-11,
+      "loss": 0.0693,
+      "step": 1635
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1635,
+  "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": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.364024017842602e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

41_128_e5_3e-5/checkpoint-1635/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ccd25f5c530b5e7648b66503aed21679838fee027eb7c7df40e963df785d384d
+size 7736

41_128_e5_3e-5/checkpoint-1635/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)

41_128_e5_3e-5/checkpoint-327/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

41_128_e5_3e-5/checkpoint-327/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": [
+    "gate_proj",
+    "o_proj",
+    "up_proj",
+    "v_proj",
+    "k_proj",
+    "q_proj",
+    "down_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

41_128_e5_3e-5/checkpoint-327/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6af6f2436b08339f106da1daab77542d4fe39c098a2d0a1af235e1fc5ab3a541
+size 791751704

41_128_e5_3e-5/checkpoint-327/latest

@@ -0,0 +1 @@
+global_step327

41_128_e5_3e-5/checkpoint-327/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c0fd8e8d9373e3ee2068cfdf33598e523f9c1060fbc6357b460a42fa0892d954
+size 15920

41_128_e5_3e-5/checkpoint-327/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e02561d325e5fa63bdeb12d99527ea903c3b9fea569e75a6a6c8fd072ab8244b
+size 15920

41_128_e5_3e-5/checkpoint-327/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6caccff63f20ee144511aee268f0a13d249761414633d6b87ae23794e8422f69
+size 15920