Upload folder using huggingface_hub

16_128_e5_3e-5/checkpoint-1044/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

16_128_e5_3e-5/checkpoint-1044/adapter_config.json

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

16_128_e5_3e-5/checkpoint-1044/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0c140408b75fb2ad00f251f4e61e5b4cc93f41afb115bd598e5e5626fc24745d
+size 791751704

16_128_e5_3e-5/checkpoint-1044/latest

@@ -0,0 +1 @@
+global_step1044

16_128_e5_3e-5/checkpoint-1044/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b52f2c0dc5bf5f6eb28039a494683917e110a26fdb808bcc09498ac74d6d6e39
+size 15920

16_128_e5_3e-5/checkpoint-1044/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0013be0cf52b192b2481f4fd6f2c031b48d504ffeebb9b1a6fc3e44d851c6b3f
+size 15920

16_128_e5_3e-5/checkpoint-1044/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4af235b8eeb8a7cd7209387dd75851976e056d81b712694809fc2736e9a5364b
+size 15920

16_128_e5_3e-5/checkpoint-1044/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28e19a5bc291ab37e46efeb04e428b710e2b875a73c75f260ae0f22f4a4c9f8f
+size 15920

16_128_e5_3e-5/checkpoint-1044/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8505b629f365d0d8e3b9af31f9adc8234338a04e6245c139542e49d59390637a
+size 15920

16_128_e5_3e-5/checkpoint-1044/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:843a660ac9cc482b40a058411dd25d0a91053587e965bc8d02f02bab6787d1c0
+size 15920

16_128_e5_3e-5/checkpoint-1044/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:22ea985b52d59e24620de21208f5f77129d6537bb8c7a8bdc94d7ea5a017f0ef
+size 15920

16_128_e5_3e-5/checkpoint-1044/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:225ee92fe1d4f596314703e46ea23d2d45a800126fb62ec8dcb1ba65bd874869
+size 15920

16_128_e5_3e-5/checkpoint-1044/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ca1bc56815f4442db226b29d45e6db346d531976fd323c44f6128718e548e84e
+size 1064

16_128_e5_3e-5/checkpoint-1044/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
+  }
+}

16_128_e5_3e-5/checkpoint-1044/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
+}

16_128_e5_3e-5/checkpoint-1044/trainer_state.json

@@ -0,0 +1,1490 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 1044,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.014367816091954023,
+      "grad_norm": 0.9956693053245544,
+      "learning_rate": 1.3793103448275862e-06,
+      "loss": 1.3425,
+      "step": 5
+    },
+    {
+      "epoch": 0.028735632183908046,
+      "grad_norm": 0.9614667296409607,
+      "learning_rate": 3.103448275862069e-06,
+      "loss": 1.361,
+      "step": 10
+    },
+    {
+      "epoch": 0.04310344827586207,
+      "grad_norm": 0.6255717277526855,
+      "learning_rate": 4.827586206896552e-06,
+      "loss": 1.3075,
+      "step": 15
+    },
+    {
+      "epoch": 0.05747126436781609,
+      "grad_norm": 0.6244459748268127,
+      "learning_rate": 6.551724137931035e-06,
+      "loss": 1.3881,
+      "step": 20
+    },
+    {
+      "epoch": 0.07183908045977011,
+      "grad_norm": 0.7454733848571777,
+      "learning_rate": 8.275862068965517e-06,
+      "loss": 1.2883,
+      "step": 25
+    },
+    {
+      "epoch": 0.08620689655172414,
+      "grad_norm": 0.6252493262290955,
+      "learning_rate": 9.999999999999999e-06,
+      "loss": 1.3055,
+      "step": 30
+    },
+    {
+      "epoch": 0.10057471264367816,
+      "grad_norm": 0.6072434782981873,
+      "learning_rate": 1.1724137931034483e-05,
+      "loss": 1.2972,
+      "step": 35
+    },
+    {
+      "epoch": 0.11494252873563218,
+      "grad_norm": 0.634437620639801,
+      "learning_rate": 1.3448275862068966e-05,
+      "loss": 1.2811,
+      "step": 40
+    },
+    {
+      "epoch": 0.12931034482758622,
+      "grad_norm": 0.4772552251815796,
+      "learning_rate": 1.517241379310345e-05,
+      "loss": 1.2304,
+      "step": 45
+    },
+    {
+      "epoch": 0.14367816091954022,
+      "grad_norm": 0.4635521173477173,
+      "learning_rate": 1.6896551724137932e-05,
+      "loss": 1.2537,
+      "step": 50
+    },
+    {
+      "epoch": 0.15804597701149425,
+      "grad_norm": 0.55922931432724,
+      "learning_rate": 1.8620689655172415e-05,
+      "loss": 1.2188,
+      "step": 55
+    },
+    {
+      "epoch": 0.1724137931034483,
+      "grad_norm": 0.44469261169433594,
+      "learning_rate": 2.0344827586206894e-05,
+      "loss": 1.2065,
+      "step": 60
+    },
+    {
+      "epoch": 0.1867816091954023,
+      "grad_norm": 0.48425620794296265,
+      "learning_rate": 2.206896551724138e-05,
+      "loss": 1.2016,
+      "step": 65
+    },
+    {
+      "epoch": 0.20114942528735633,
+      "grad_norm": 0.5222848653793335,
+      "learning_rate": 2.3793103448275862e-05,
+      "loss": 1.2123,
+      "step": 70
+    },
+    {
+      "epoch": 0.21551724137931033,
+      "grad_norm": 0.48502737283706665,
+      "learning_rate": 2.5517241379310345e-05,
+      "loss": 1.1739,
+      "step": 75
+    },
+    {
+      "epoch": 0.22988505747126436,
+      "grad_norm": 0.549568772315979,
+      "learning_rate": 2.7241379310344827e-05,
+      "loss": 1.2135,
+      "step": 80
+    },
+    {
+      "epoch": 0.2442528735632184,
+      "grad_norm": 0.538798451423645,
+      "learning_rate": 2.8965517241379313e-05,
+      "loss": 1.1525,
+      "step": 85
+    },
+    {
+      "epoch": 0.25862068965517243,
+      "grad_norm": 0.5352644920349121,
+      "learning_rate": 2.9999891638559405e-05,
+      "loss": 1.1424,
+      "step": 90
+    },
+    {
+      "epoch": 0.27298850574712646,
+      "grad_norm": 0.5414754152297974,
+      "learning_rate": 2.9998672590332875e-05,
+      "loss": 1.1321,
+      "step": 95
+    },
+    {
+      "epoch": 0.28735632183908044,
+      "grad_norm": 0.550433337688446,
+      "learning_rate": 2.999609915252684e-05,
+      "loss": 1.1292,
+      "step": 100
+    },
+    {
+      "epoch": 0.3017241379310345,
+      "grad_norm": 0.5901126265525818,
+      "learning_rate": 2.999217155752434e-05,
+      "loss": 1.1459,
+      "step": 105
+    },
+    {
+      "epoch": 0.3160919540229885,
+      "grad_norm": 0.6439724564552307,
+      "learning_rate": 2.9986890159989696e-05,
+      "loss": 1.0742,
+      "step": 110
+    },
+    {
+      "epoch": 0.33045977011494254,
+      "grad_norm": 0.5696046352386475,
+      "learning_rate": 2.9980255436836418e-05,
+      "loss": 1.1454,
+      "step": 115
+    },
+    {
+      "epoch": 0.3448275862068966,
+      "grad_norm": 0.6889880299568176,
+      "learning_rate": 2.9972267987184195e-05,
+      "loss": 1.086,
+      "step": 120
+    },
+    {
+      "epoch": 0.35919540229885055,
+      "grad_norm": 0.5843938589096069,
+      "learning_rate": 2.996292853230475e-05,
+      "loss": 1.0829,
+      "step": 125
+    },
+    {
+      "epoch": 0.3735632183908046,
+      "grad_norm": 0.6964264512062073,
+      "learning_rate": 2.9952237915556746e-05,
+      "loss": 1.0979,
+      "step": 130
+    },
+    {
+      "epoch": 0.3879310344827586,
+      "grad_norm": 0.741144597530365,
+      "learning_rate": 2.9940197102309603e-05,
+      "loss": 1.1022,
+      "step": 135
+    },
+    {
+      "epoch": 0.40229885057471265,
+      "grad_norm": 0.6913484930992126,
+      "learning_rate": 2.9926807179856342e-05,
+      "loss": 1.061,
+      "step": 140
+    },
+    {
+      "epoch": 0.4166666666666667,
+      "grad_norm": 0.6852656006813049,
+      "learning_rate": 2.9912069357315394e-05,
+      "loss": 1.0238,
+      "step": 145
+    },
+    {
+      "epoch": 0.43103448275862066,
+      "grad_norm": 0.7292410731315613,
+      "learning_rate": 2.989598496552141e-05,
+      "loss": 1.009,
+      "step": 150
+    },
+    {
+      "epoch": 0.4454022988505747,
+      "grad_norm": 0.6415665745735168,
+      "learning_rate": 2.9878555456905102e-05,
+      "loss": 1.0434,
+      "step": 155
+    },
+    {
+      "epoch": 0.45977011494252873,
+      "grad_norm": 0.7307606935501099,
+      "learning_rate": 2.985978240536207e-05,
+      "loss": 1.0383,
+      "step": 160
+    },
+    {
+      "epoch": 0.47413793103448276,
+      "grad_norm": 0.6481092572212219,
+      "learning_rate": 2.983966750611069e-05,
+      "loss": 1.0478,
+      "step": 165
+    },
+    {
+      "epoch": 0.4885057471264368,
+      "grad_norm": 0.694217324256897,
+      "learning_rate": 2.981821257553903e-05,
+      "loss": 1.0037,
+      "step": 170
+    },
+    {
+      "epoch": 0.5028735632183908,
+      "grad_norm": 0.7326360940933228,
+      "learning_rate": 2.9795419551040836e-05,
+      "loss": 1.0133,
+      "step": 175
+    },
+    {
+      "epoch": 0.5172413793103449,
+      "grad_norm": 0.8161800503730774,
+      "learning_rate": 2.9771290490840567e-05,
+      "loss": 0.9865,
+      "step": 180
+    },
+    {
+      "epoch": 0.5316091954022989,
+      "grad_norm": 0.8219459056854248,
+      "learning_rate": 2.9745827573807555e-05,
+      "loss": 0.9889,
+      "step": 185
+    },
+    {
+      "epoch": 0.5459770114942529,
+      "grad_norm": 0.7902458906173706,
+      "learning_rate": 2.9719033099259242e-05,
+      "loss": 0.9483,
+      "step": 190
+    },
+    {
+      "epoch": 0.5603448275862069,
+      "grad_norm": 0.7876456379890442,
+      "learning_rate": 2.969090948675355e-05,
+      "loss": 0.9768,
+      "step": 195
+    },
+    {
+      "epoch": 0.5747126436781609,
+      "grad_norm": 0.8362347483634949,
+      "learning_rate": 2.96614592758704e-05,
+      "loss": 0.9041,
+      "step": 200
+    },
+    {
+      "epoch": 0.5890804597701149,
+      "grad_norm": 1.062076210975647,
+      "learning_rate": 2.963068512598238e-05,
+      "loss": 0.9232,
+      "step": 205
+    },
+    {
+      "epoch": 0.603448275862069,
+      "grad_norm": 0.8888565897941589,
+      "learning_rate": 2.9598589816014606e-05,
+      "loss": 0.9191,
+      "step": 210
+    },
+    {
+      "epoch": 0.617816091954023,
+      "grad_norm": 0.8549149632453918,
+      "learning_rate": 2.9565176244193768e-05,
+      "loss": 0.8662,
+      "step": 215
+    },
+    {
+      "epoch": 0.632183908045977,
+      "grad_norm": 0.7843875288963318,
+      "learning_rate": 2.9530447427786454e-05,
+      "loss": 0.8909,
+      "step": 220
+    },
+    {
+      "epoch": 0.646551724137931,
+      "grad_norm": 0.9624897241592407,
+      "learning_rate": 2.949440650282664e-05,
+      "loss": 0.8995,
+      "step": 225
+    },
+    {
+      "epoch": 0.6609195402298851,
+      "grad_norm": 0.8020894527435303,
+      "learning_rate": 2.9457056723832563e-05,
+      "loss": 0.8472,
+      "step": 230
+    },
+    {
+      "epoch": 0.6752873563218391,
+      "grad_norm": 1.180816411972046,
+      "learning_rate": 2.9418401463512772e-05,
+      "loss": 0.9324,
+      "step": 235
+    },
+    {
+      "epoch": 0.6896551724137931,
+      "grad_norm": 0.985790491104126,
+      "learning_rate": 2.937844421246162e-05,
+      "loss": 0.8657,
+      "step": 240
+    },
+    {
+      "epoch": 0.7040229885057471,
+      "grad_norm": 0.9070732593536377,
+      "learning_rate": 2.9337188578844035e-05,
+      "loss": 0.8351,
+      "step": 245
+    },
+    {
+      "epoch": 0.7183908045977011,
+      "grad_norm": 0.945988118648529,
+      "learning_rate": 2.929463828806971e-05,
+      "loss": 0.8672,
+      "step": 250
+    },
+    {
+      "epoch": 0.7327586206896551,
+      "grad_norm": 0.9512550234794617,
+      "learning_rate": 2.9250797182456704e-05,
+      "loss": 0.8258,
+      "step": 255
+    },
+    {
+      "epoch": 0.7471264367816092,
+      "grad_norm": 1.025450587272644,
+      "learning_rate": 2.920566922088445e-05,
+      "loss": 0.8214,
+      "step": 260
+    },
+    {
+      "epoch": 0.7614942528735632,
+      "grad_norm": 0.9240112900733948,
+      "learning_rate": 2.9159258478436297e-05,
+      "loss": 0.7967,
+      "step": 265
+    },
+    {
+      "epoch": 0.7758620689655172,
+      "grad_norm": 1.0366438627243042,
+      "learning_rate": 2.911156914603152e-05,
+      "loss": 0.7778,
+      "step": 270
+    },
+    {
+      "epoch": 0.7902298850574713,
+      "grad_norm": 0.942286491394043,
+      "learning_rate": 2.9062605530046854e-05,
+      "loss": 0.745,
+      "step": 275
+    },
+    {
+      "epoch": 0.8045977011494253,
+      "grad_norm": 1.0067769289016724,
+      "learning_rate": 2.901237205192767e-05,
+      "loss": 0.7587,
+      "step": 280
+    },
+    {
+      "epoch": 0.8189655172413793,
+      "grad_norm": 0.9403071999549866,
+      "learning_rate": 2.8960873247788655e-05,
+      "loss": 0.7969,
+      "step": 285
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 0.9700637459754944,
+      "learning_rate": 2.8908113768004266e-05,
+      "loss": 0.8114,
+      "step": 290
+    },
+    {
+      "epoch": 0.8477011494252874,
+      "grad_norm": 0.9859459400177002,
+      "learning_rate": 2.8854098376788742e-05,
+      "loss": 0.7941,
+      "step": 295
+    },
+    {
+      "epoch": 0.8620689655172413,
+      "grad_norm": 0.8560702800750732,
+      "learning_rate": 2.8798831951765933e-05,
+      "loss": 0.8226,
+      "step": 300
+    },
+    {
+      "epoch": 0.8764367816091954,
+      "grad_norm": 0.9503181576728821,
+      "learning_rate": 2.8742319483528805e-05,
+      "loss": 0.7818,
+      "step": 305
+    },
+    {
+      "epoch": 0.8908045977011494,
+      "grad_norm": 0.9771649837493896,
+      "learning_rate": 2.8684566075188835e-05,
+      "loss": 0.777,
+      "step": 310
+    },
+    {
+      "epoch": 0.9051724137931034,
+      "grad_norm": 1.0014532804489136,
+      "learning_rate": 2.8625576941915162e-05,
+      "loss": 0.7499,
+      "step": 315
+    },
+    {
+      "epoch": 0.9195402298850575,
+      "grad_norm": 1.0961740016937256,
+      "learning_rate": 2.8565357410463664e-05,
+      "loss": 0.7755,
+      "step": 320
+    },
+    {
+      "epoch": 0.9339080459770115,
+      "grad_norm": 1.0916162729263306,
+      "learning_rate": 2.8503912918695952e-05,
+      "loss": 0.6966,
+      "step": 325
+    },
+    {
+      "epoch": 0.9482758620689655,
+      "grad_norm": 1.1250373125076294,
+      "learning_rate": 2.8441249015088322e-05,
+      "loss": 0.6778,
+      "step": 330
+    },
+    {
+      "epoch": 0.9626436781609196,
+      "grad_norm": 1.1927026510238647,
+      "learning_rate": 2.8377371358230733e-05,
+      "loss": 0.7595,
+      "step": 335
+    },
+    {
+      "epoch": 0.9770114942528736,
+      "grad_norm": 0.9869332313537598,
+      "learning_rate": 2.8312285716315823e-05,
+      "loss": 0.7361,
+      "step": 340
+    },
+    {
+      "epoch": 0.9913793103448276,
+      "grad_norm": 1.10604989528656,
+      "learning_rate": 2.8245997966618037e-05,
+      "loss": 0.6662,
+      "step": 345
+    },
+    {
+      "epoch": 1.0057471264367817,
+      "grad_norm": 1.1478843688964844,
+      "learning_rate": 2.8178514094962917e-05,
+      "loss": 0.6208,
+      "step": 350
+    },
+    {
+      "epoch": 1.0201149425287357,
+      "grad_norm": 1.15924072265625,
+      "learning_rate": 2.8109840195186565e-05,
+      "loss": 0.6334,
+      "step": 355
+    },
+    {
+      "epoch": 1.0344827586206897,
+      "grad_norm": 1.1480649709701538,
+      "learning_rate": 2.8039982468585372e-05,
+      "loss": 0.6328,
+      "step": 360
+    },
+    {
+      "epoch": 1.0488505747126438,
+      "grad_norm": 1.1573606729507446,
+      "learning_rate": 2.796894722335605e-05,
+      "loss": 0.5882,
+      "step": 365
+    },
+    {
+      "epoch": 1.0632183908045978,
+      "grad_norm": 1.1750164031982422,
+      "learning_rate": 2.7896740874025962e-05,
+      "loss": 0.6313,
+      "step": 370
+    },
+    {
+      "epoch": 1.0775862068965518,
+      "grad_norm": 1.1878570318222046,
+      "learning_rate": 2.782336994087393e-05,
+      "loss": 0.5848,
+      "step": 375
+    },
+    {
+      "epoch": 1.0919540229885056,
+      "grad_norm": 1.1398131847381592,
+      "learning_rate": 2.7748841049341407e-05,
+      "loss": 0.5686,
+      "step": 380
+    },
+    {
+      "epoch": 1.1063218390804597,
+      "grad_norm": 1.0254820585250854,
+      "learning_rate": 2.7673160929434242e-05,
+      "loss": 0.5807,
+      "step": 385
+    },
+    {
+      "epoch": 1.1206896551724137,
+      "grad_norm": 1.2164369821548462,
+      "learning_rate": 2.759633641511491e-05,
+      "loss": 0.6266,
+      "step": 390
+    },
+    {
+      "epoch": 1.1350574712643677,
+      "grad_norm": 1.0185494422912598,
+      "learning_rate": 2.7518374443685432e-05,
+      "loss": 0.5731,
+      "step": 395
+    },
+    {
+      "epoch": 1.1494252873563218,
+      "grad_norm": 1.2007052898406982,
+      "learning_rate": 2.7439282055160924e-05,
+      "loss": 0.5687,
+      "step": 400
+    },
+    {
+      "epoch": 1.1637931034482758,
+      "grad_norm": 1.2135870456695557,
+      "learning_rate": 2.735906639163387e-05,
+      "loss": 0.5538,
+      "step": 405
+    },
+    {
+      "epoch": 1.1781609195402298,
+      "grad_norm": 1.0088754892349243,
+      "learning_rate": 2.7277734696629197e-05,
+      "loss": 0.5804,
+      "step": 410
+    },
+    {
+      "epoch": 1.1925287356321839,
+      "grad_norm": 1.2317101955413818,
+      "learning_rate": 2.719529431445018e-05,
+      "loss": 0.5416,
+      "step": 415
+    },
+    {
+      "epoch": 1.206896551724138,
+      "grad_norm": 1.1315484046936035,
+      "learning_rate": 2.7111752689515243e-05,
+      "loss": 0.5089,
+      "step": 420
+    },
+    {
+      "epoch": 1.221264367816092,
+      "grad_norm": 1.205450415611267,
+      "learning_rate": 2.7027117365685727e-05,
+      "loss": 0.5906,
+      "step": 425
+    },
+    {
+      "epoch": 1.235632183908046,
+      "grad_norm": 1.1637941598892212,
+      "learning_rate": 2.6941395985584656e-05,
+      "loss": 0.5385,
+      "step": 430
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.1773250102996826,
+      "learning_rate": 2.6854596289906646e-05,
+      "loss": 0.5534,
+      "step": 435
+    },
+    {
+      "epoch": 1.264367816091954,
+      "grad_norm": 1.093550205230713,
+      "learning_rate": 2.6766726116718863e-05,
+      "loss": 0.5846,
+      "step": 440
+    },
+    {
+      "epoch": 1.278735632183908,
+      "grad_norm": 1.2478264570236206,
+      "learning_rate": 2.6677793400753285e-05,
+      "loss": 0.5546,
+      "step": 445
+    },
+    {
+      "epoch": 1.293103448275862,
+      "grad_norm": 1.2445526123046875,
+      "learning_rate": 2.6587806172690153e-05,
+      "loss": 0.5603,
+      "step": 450
+    },
+    {
+      "epoch": 1.3074712643678161,
+      "grad_norm": 1.061118483543396,
+      "learning_rate": 2.6496772558432844e-05,
+      "loss": 0.4761,
+      "step": 455
+    },
+    {
+      "epoch": 1.3218390804597702,
+      "grad_norm": 1.376889705657959,
+      "learning_rate": 2.6404700778374034e-05,
+      "loss": 0.5313,
+      "step": 460
+    },
+    {
+      "epoch": 1.3362068965517242,
+      "grad_norm": 1.097720742225647,
+      "learning_rate": 2.6311599146653446e-05,
+      "loss": 0.4878,
+      "step": 465
+    },
+    {
+      "epoch": 1.3505747126436782,
+      "grad_norm": 1.3083029985427856,
+      "learning_rate": 2.6217476070407048e-05,
+      "loss": 0.4992,
+      "step": 470
+    },
+    {
+      "epoch": 1.3649425287356323,
+      "grad_norm": 1.0825142860412598,
+      "learning_rate": 2.6122340049007895e-05,
+      "loss": 0.5441,
+      "step": 475
+    },
+    {
+      "epoch": 1.3793103448275863,
+      "grad_norm": 1.1726360321044922,
+      "learning_rate": 2.602619967329862e-05,
+      "loss": 0.5023,
+      "step": 480
+    },
+    {
+      "epoch": 1.3936781609195403,
+      "grad_norm": 1.739815354347229,
+      "learning_rate": 2.5929063624815698e-05,
+      "loss": 0.5557,
+      "step": 485
+    },
+    {
+      "epoch": 1.4080459770114944,
+      "grad_norm": 1.263541579246521,
+      "learning_rate": 2.583094067500547e-05,
+      "loss": 0.5174,
+      "step": 490
+    },
+    {
+      "epoch": 1.4224137931034484,
+      "grad_norm": 1.1408122777938843,
+      "learning_rate": 2.573183968443209e-05,
+      "loss": 0.5416,
+      "step": 495
+    },
+    {
+      "epoch": 1.4367816091954024,
+      "grad_norm": 1.0248563289642334,
+      "learning_rate": 2.5631769601977408e-05,
+      "loss": 0.5255,
+      "step": 500
+    },
+    {
+      "epoch": 1.4511494252873562,
+      "grad_norm": 1.1455802917480469,
+      "learning_rate": 2.553073946403289e-05,
+      "loss": 0.5121,
+      "step": 505
+    },
+    {
+      "epoch": 1.4655172413793103,
+      "grad_norm": 1.0702999830245972,
+      "learning_rate": 2.5428758393683605e-05,
+      "loss": 0.4928,
+      "step": 510
+    },
+    {
+      "epoch": 1.4798850574712643,
+      "grad_norm": 1.3014647960662842,
+      "learning_rate": 2.5325835599884427e-05,
+      "loss": 0.487,
+      "step": 515
+    },
+    {
+      "epoch": 1.4942528735632183,
+      "grad_norm": 1.1569771766662598,
+      "learning_rate": 2.5221980376628445e-05,
+      "loss": 0.5036,
+      "step": 520
+    },
+    {
+      "epoch": 1.5086206896551724,
+      "grad_norm": 1.3842607736587524,
+      "learning_rate": 2.5117202102107707e-05,
+      "loss": 0.478,
+      "step": 525
+    },
+    {
+      "epoch": 1.5229885057471264,
+      "grad_norm": 1.1965588331222534,
+      "learning_rate": 2.5011510237866398e-05,
+      "loss": 0.4697,
+      "step": 530
+    },
+    {
+      "epoch": 1.5373563218390804,
+      "grad_norm": 1.211634635925293,
+      "learning_rate": 2.4904914327946398e-05,
+      "loss": 0.5162,
+      "step": 535
+    },
+    {
+      "epoch": 1.5517241379310345,
+      "grad_norm": 1.0299171209335327,
+      "learning_rate": 2.4797423998025505e-05,
+      "loss": 0.4793,
+      "step": 540
+    },
+    {
+      "epoch": 1.5660919540229885,
+      "grad_norm": 1.1820342540740967,
+      "learning_rate": 2.4689048954548193e-05,
+      "loss": 0.487,
+      "step": 545
+    },
+    {
+      "epoch": 1.5804597701149425,
+      "grad_norm": 1.2619454860687256,
+      "learning_rate": 2.4579798983849127e-05,
+      "loss": 0.5021,
+      "step": 550
+    },
+    {
+      "epoch": 1.5948275862068966,
+      "grad_norm": 1.279123306274414,
+      "learning_rate": 2.446968395126946e-05,
+      "loss": 0.5236,
+      "step": 555
+    },
+    {
+      "epoch": 1.6091954022988506,
+      "grad_norm": 1.0806742906570435,
+      "learning_rate": 2.4358713800265978e-05,
+      "loss": 0.4738,
+      "step": 560
+    },
+    {
+      "epoch": 1.6235632183908046,
+      "grad_norm": 1.1664921045303345,
+      "learning_rate": 2.4246898551513182e-05,
+      "loss": 0.4352,
+      "step": 565
+    },
+    {
+      "epoch": 1.6379310344827587,
+      "grad_norm": 1.0607774257659912,
+      "learning_rate": 2.4134248301998455e-05,
+      "loss": 0.4097,
+      "step": 570
+    },
+    {
+      "epoch": 1.6522988505747125,
+      "grad_norm": 1.155291199684143,
+      "learning_rate": 2.402077322411026e-05,
+      "loss": 0.4855,
+      "step": 575
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 1.2184966802597046,
+      "learning_rate": 2.3906483564719586e-05,
+      "loss": 0.4937,
+      "step": 580
+    },
+    {
+      "epoch": 1.6810344827586206,
+      "grad_norm": 1.0277451276779175,
+      "learning_rate": 2.379138964425464e-05,
+      "loss": 0.3834,
+      "step": 585
+    },
+    {
+      "epoch": 1.6954022988505746,
+      "grad_norm": 1.153117299079895,
+      "learning_rate": 2.3675501855768913e-05,
+      "loss": 0.404,
+      "step": 590
+    },
+    {
+      "epoch": 1.7097701149425286,
+      "grad_norm": 1.2163881063461304,
+      "learning_rate": 2.3558830664002688e-05,
+      "loss": 0.4134,
+      "step": 595
+    },
+    {
+      "epoch": 1.7241379310344827,
+      "grad_norm": 1.148512363433838,
+      "learning_rate": 2.344138660443804e-05,
+      "loss": 0.4225,
+      "step": 600
+    },
+    {
+      "epoch": 1.7385057471264367,
+      "grad_norm": 1.2566574811935425,
+      "learning_rate": 2.3323180282347505e-05,
+      "loss": 0.4337,
+      "step": 605
+    },
+    {
+      "epoch": 1.7528735632183907,
+      "grad_norm": 1.3375940322875977,
+      "learning_rate": 2.320422237183641e-05,
+      "loss": 0.4509,
+      "step": 610
+    },
+    {
+      "epoch": 1.7672413793103448,
+      "grad_norm": 1.1347380876541138,
+      "learning_rate": 2.3084523614878962e-05,
+      "loss": 0.3973,
+      "step": 615
+    },
+    {
+      "epoch": 1.7816091954022988,
+      "grad_norm": 1.21699857711792,
+      "learning_rate": 2.2964094820348302e-05,
+      "loss": 0.4534,
+      "step": 620
+    },
+    {
+      "epoch": 1.7959770114942528,
+      "grad_norm": 1.137501835823059,
+      "learning_rate": 2.284294686304041e-05,
+      "loss": 0.4236,
+      "step": 625
+    },
+    {
+      "epoch": 1.8103448275862069,
+      "grad_norm": 1.1551192998886108,
+      "learning_rate": 2.2721090682692114e-05,
+      "loss": 0.4328,
+      "step": 630
+    },
+    {
+      "epoch": 1.8247126436781609,
+      "grad_norm": 1.149267315864563,
+      "learning_rate": 2.2598537282993232e-05,
+      "loss": 0.4095,
+      "step": 635
+    },
+    {
+      "epoch": 1.839080459770115,
+      "grad_norm": 1.2572345733642578,
+      "learning_rate": 2.247529773059293e-05,
+      "loss": 0.3723,
+      "step": 640
+    },
+    {
+      "epoch": 1.853448275862069,
+      "grad_norm": 1.1943402290344238,
+      "learning_rate": 2.235138315410039e-05,
+      "loss": 0.3944,
+      "step": 645
+    },
+    {
+      "epoch": 1.867816091954023,
+      "grad_norm": 1.2040687799453735,
+      "learning_rate": 2.2226804743079897e-05,
+      "loss": 0.3892,
+      "step": 650
+    },
+    {
+      "epoch": 1.882183908045977,
+      "grad_norm": 1.1246652603149414,
+      "learning_rate": 2.2101573747040418e-05,
+      "loss": 0.4065,
+      "step": 655
+    },
+    {
+      "epoch": 1.896551724137931,
+      "grad_norm": 1.3335580825805664,
+      "learning_rate": 2.1975701474419752e-05,
+      "loss": 0.3986,
+      "step": 660
+    },
+    {
+      "epoch": 1.910919540229885,
+      "grad_norm": 1.1872365474700928,
+      "learning_rate": 2.1849199291563375e-05,
+      "loss": 0.3806,
+      "step": 665
+    },
+    {
+      "epoch": 1.9252873563218391,
+      "grad_norm": 1.035778284072876,
+      "learning_rate": 2.1722078621698056e-05,
+      "loss": 0.3427,
+      "step": 670
+    },
+    {
+      "epoch": 1.9396551724137931,
+      "grad_norm": 1.2677980661392212,
+      "learning_rate": 2.1594350943900334e-05,
+      "loss": 0.4134,
+      "step": 675
+    },
+    {
+      "epoch": 1.9540229885057472,
+      "grad_norm": 2.563685655593872,
+      "learning_rate": 2.1466027792059938e-05,
+      "loss": 0.3871,
+      "step": 680
+    },
+    {
+      "epoch": 1.9683908045977012,
+      "grad_norm": 1.1063146591186523,
+      "learning_rate": 2.133712075383829e-05,
+      "loss": 0.3482,
+      "step": 685
+    },
+    {
+      "epoch": 1.9827586206896552,
+      "grad_norm": 1.1051040887832642,
+      "learning_rate": 2.1207641469622116e-05,
+      "loss": 0.3415,
+      "step": 690
+    },
+    {
+      "epoch": 1.9971264367816093,
+      "grad_norm": 1.1961126327514648,
+      "learning_rate": 2.1077601631472304e-05,
+      "loss": 0.378,
+      "step": 695
+    },
+    {
+      "epoch": 2.0114942528735633,
+      "grad_norm": 1.4363032579421997,
+      "learning_rate": 2.0947012982068133e-05,
+      "loss": 0.3195,
+      "step": 700
+    },
+    {
+      "epoch": 2.0258620689655173,
+      "grad_norm": 1.0249502658843994,
+      "learning_rate": 2.0815887313646867e-05,
+      "loss": 0.2616,
+      "step": 705
+    },
+    {
+      "epoch": 2.0402298850574714,
+      "grad_norm": 1.139730453491211,
+      "learning_rate": 2.0684236466938943e-05,
+      "loss": 0.2989,
+      "step": 710
+    },
+    {
+      "epoch": 2.0545977011494254,
+      "grad_norm": 1.1467976570129395,
+      "learning_rate": 2.055207233009872e-05,
+      "loss": 0.2754,
+      "step": 715
+    },
+    {
+      "epoch": 2.0689655172413794,
+      "grad_norm": 1.3604381084442139,
+      "learning_rate": 2.0419406837630962e-05,
+      "loss": 0.2883,
+      "step": 720
+    },
+    {
+      "epoch": 2.0833333333333335,
+      "grad_norm": 1.2879091501235962,
+      "learning_rate": 2.0286251969313194e-05,
+      "loss": 0.3149,
+      "step": 725
+    },
+    {
+      "epoch": 2.0977011494252875,
+      "grad_norm": 1.4303884506225586,
+      "learning_rate": 2.0152619749113864e-05,
+      "loss": 0.2978,
+      "step": 730
+    },
+    {
+      "epoch": 2.1120689655172415,
+      "grad_norm": 1.280913233757019,
+      "learning_rate": 2.00185222441066e-05,
+      "loss": 0.27,
+      "step": 735
+    },
+    {
+      "epoch": 2.1264367816091956,
+      "grad_norm": 1.3666342496871948,
+      "learning_rate": 1.9883971563380562e-05,
+      "loss": 0.3314,
+      "step": 740
+    },
+    {
+      "epoch": 2.1408045977011496,
+      "grad_norm": 1.1915347576141357,
+      "learning_rate": 1.974897985694693e-05,
+      "loss": 0.275,
+      "step": 745
+    },
+    {
+      "epoch": 2.1551724137931036,
+      "grad_norm": 1.2179591655731201,
+      "learning_rate": 1.96135593146418e-05,
+      "loss": 0.2529,
+      "step": 750
+    },
+    {
+      "epoch": 2.1695402298850572,
+      "grad_norm": 1.0577384233474731,
+      "learning_rate": 1.9477722165025422e-05,
+      "loss": 0.2589,
+      "step": 755
+    },
+    {
+      "epoch": 2.1839080459770113,
+      "grad_norm": 1.1857539415359497,
+      "learning_rate": 1.9341480674277935e-05,
+      "loss": 0.2922,
+      "step": 760
+    },
+    {
+      "epoch": 2.1982758620689653,
+      "grad_norm": 1.3017208576202393,
+      "learning_rate": 1.9204847145091755e-05,
+      "loss": 0.2778,
+      "step": 765
+    },
+    {
+      "epoch": 2.2126436781609193,
+      "grad_norm": 1.17992103099823,
+      "learning_rate": 1.9067833915560587e-05,
+      "loss": 0.2908,
+      "step": 770
+    },
+    {
+      "epoch": 2.2270114942528734,
+      "grad_norm": 1.127840280532837,
+      "learning_rate": 1.8930453358065356e-05,
+      "loss": 0.2796,
+      "step": 775
+    },
+    {
+      "epoch": 2.2413793103448274,
+      "grad_norm": 1.0646225214004517,
+      "learning_rate": 1.8792717878156902e-05,
+      "loss": 0.2868,
+      "step": 780
+    },
+    {
+      "epoch": 2.2557471264367814,
+      "grad_norm": 1.36182701587677,
+      "learning_rate": 1.8654639913435786e-05,
+      "loss": 0.2752,
+      "step": 785
+    },
+    {
+      "epoch": 2.2701149425287355,
+      "grad_norm": 1.0714491605758667,
+      "learning_rate": 1.8516231932429175e-05,
+      "loss": 0.2668,
+      "step": 790
+    },
+    {
+      "epoch": 2.2844827586206895,
+      "grad_norm": 1.2853004932403564,
+      "learning_rate": 1.8377506433464914e-05,
+      "loss": 0.2463,
+      "step": 795
+    },
+    {
+      "epoch": 2.2988505747126435,
+      "grad_norm": 1.011826515197754,
+      "learning_rate": 1.823847594354291e-05,
+      "loss": 0.239,
+      "step": 800
+    },
+    {
+      "epoch": 2.3132183908045976,
+      "grad_norm": 1.3612297773361206,
+      "learning_rate": 1.8099153017203943e-05,
+      "loss": 0.2804,
+      "step": 805
+    },
+    {
+      "epoch": 2.3275862068965516,
+      "grad_norm": 1.0135477781295776,
+      "learning_rate": 1.7959550235396002e-05,
+      "loss": 0.3225,
+      "step": 810
+    },
+    {
+      "epoch": 2.3419540229885056,
+      "grad_norm": 1.096671223640442,
+      "learning_rate": 1.781968020433819e-05,
+      "loss": 0.2386,
+      "step": 815
+    },
+    {
+      "epoch": 2.3563218390804597,
+      "grad_norm": 1.1345715522766113,
+      "learning_rate": 1.7679555554382382e-05,
+      "loss": 0.2524,
+      "step": 820
+    },
+    {
+      "epoch": 2.3706896551724137,
+      "grad_norm": 1.0843546390533447,
+      "learning_rate": 1.7539188938872708e-05,
+      "loss": 0.221,
+      "step": 825
+    },
+    {
+      "epoch": 2.3850574712643677,
+      "grad_norm": 1.3394324779510498,
+      "learning_rate": 1.7398593033002945e-05,
+      "loss": 0.284,
+      "step": 830
+    },
+    {
+      "epoch": 2.3994252873563218,
+      "grad_norm": 1.4917807579040527,
+      "learning_rate": 1.7257780532671903e-05,
+      "loss": 0.2559,
+      "step": 835
+    },
+    {
+      "epoch": 2.413793103448276,
+      "grad_norm": 1.0991957187652588,
+      "learning_rate": 1.7116764153337056e-05,
+      "loss": 0.2438,
+      "step": 840
+    },
+    {
+      "epoch": 2.42816091954023,
+      "grad_norm": 1.4984471797943115,
+      "learning_rate": 1.697555662886625e-05,
+      "loss": 0.2584,
+      "step": 845
+    },
+    {
+      "epoch": 2.442528735632184,
+      "grad_norm": 1.1856584548950195,
+      "learning_rate": 1.683417071038785e-05,
+      "loss": 0.2481,
+      "step": 850
+    },
+    {
+      "epoch": 2.456896551724138,
+      "grad_norm": 1.3198546171188354,
+      "learning_rate": 1.6692619165139334e-05,
+      "loss": 0.266,
+      "step": 855
+    },
+    {
+      "epoch": 2.471264367816092,
+      "grad_norm": 1.1649495363235474,
+      "learning_rate": 1.6550914775314372e-05,
+      "loss": 0.2795,
+      "step": 860
+    },
+    {
+      "epoch": 2.485632183908046,
+      "grad_norm": 1.236920714378357,
+      "learning_rate": 1.6409070336908588e-05,
+      "loss": 0.2235,
+      "step": 865
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.1417492628097534,
+      "learning_rate": 1.62670986585641e-05,
+      "loss": 0.2742,
+      "step": 870
+    },
+    {
+      "epoch": 2.514367816091954,
+      "grad_norm": 1.3992029428482056,
+      "learning_rate": 1.6125012560412856e-05,
+      "loss": 0.2246,
+      "step": 875
+    },
+    {
+      "epoch": 2.528735632183908,
+      "grad_norm": 1.2300302982330322,
+      "learning_rate": 1.5982824872918994e-05,
+      "loss": 0.2788,
+      "step": 880
+    },
+    {
+      "epoch": 2.543103448275862,
+      "grad_norm": 1.0692559480667114,
+      "learning_rate": 1.5840548435720226e-05,
+      "loss": 0.218,
+      "step": 885
+    },
+    {
+      "epoch": 2.557471264367816,
+      "grad_norm": 1.056885838508606,
+      "learning_rate": 1.569819609646842e-05,
+      "loss": 0.2386,
+      "step": 890
+    },
+    {
+      "epoch": 2.57183908045977,
+      "grad_norm": 1.2882906198501587,
+      "learning_rate": 1.5555780709669455e-05,
+      "loss": 0.2371,
+      "step": 895
+    },
+    {
+      "epoch": 2.586206896551724,
+      "grad_norm": 1.0865809917449951,
+      "learning_rate": 1.5413315135522434e-05,
+      "loss": 0.249,
+      "step": 900
+    },
+    {
+      "epoch": 2.6005747126436782,
+      "grad_norm": 1.100043535232544,
+      "learning_rate": 1.5270812238758407e-05,
+      "loss": 0.248,
+      "step": 905
+    },
+    {
+      "epoch": 2.6149425287356323,
+      "grad_norm": 1.1947323083877563,
+      "learning_rate": 1.5128284887478698e-05,
+      "loss": 0.2366,
+      "step": 910
+    },
+    {
+      "epoch": 2.6293103448275863,
+      "grad_norm": 1.1760265827178955,
+      "learning_rate": 1.4985745951992865e-05,
+      "loss": 0.2149,
+      "step": 915
+    },
+    {
+      "epoch": 2.6436781609195403,
+      "grad_norm": 1.2085545063018799,
+      "learning_rate": 1.4843208303656542e-05,
+      "loss": 0.2301,
+      "step": 920
+    },
+    {
+      "epoch": 2.6580459770114944,
+      "grad_norm": 1.1468777656555176,
+      "learning_rate": 1.4700684813709132e-05,
+      "loss": 0.2535,
+      "step": 925
+    },
+    {
+      "epoch": 2.6724137931034484,
+      "grad_norm": 1.318094253540039,
+      "learning_rate": 1.455818835211152e-05,
+      "loss": 0.2112,
+      "step": 930
+    },
+    {
+      "epoch": 2.6867816091954024,
+      "grad_norm": 1.2293264865875244,
+      "learning_rate": 1.441573178638392e-05,
+      "loss": 0.2199,
+      "step": 935
+    },
+    {
+      "epoch": 2.7011494252873565,
+      "grad_norm": 1.1567370891571045,
+      "learning_rate": 1.427332798044392e-05,
+      "loss": 0.2001,
+      "step": 940
+    },
+    {
+      "epoch": 2.7155172413793105,
+      "grad_norm": 1.162387490272522,
+      "learning_rate": 1.413098979344487e-05,
+      "loss": 0.1811,
+      "step": 945
+    },
+    {
+      "epoch": 2.7298850574712645,
+      "grad_norm": 1.17710542678833,
+      "learning_rate": 1.3988730078614684e-05,
+      "loss": 0.2425,
+      "step": 950
+    },
+    {
+      "epoch": 2.7442528735632186,
+      "grad_norm": 1.2836874723434448,
+      "learning_rate": 1.3846561682095192e-05,
+      "loss": 0.2217,
+      "step": 955
+    },
+    {
+      "epoch": 2.7586206896551726,
+      "grad_norm": 1.2637510299682617,
+      "learning_rate": 1.3704497441782117e-05,
+      "loss": 0.2311,
+      "step": 960
+    },
+    {
+      "epoch": 2.7729885057471266,
+      "grad_norm": 1.1894924640655518,
+      "learning_rate": 1.3562550186165829e-05,
+      "loss": 0.2246,
+      "step": 965
+    },
+    {
+      "epoch": 2.7873563218390807,
+      "grad_norm": 1.1791173219680786,
+      "learning_rate": 1.3420732733172882e-05,
+      "loss": 0.1856,
+      "step": 970
+    },
+    {
+      "epoch": 2.8017241379310347,
+      "grad_norm": 1.2785968780517578,
+      "learning_rate": 1.327905788900859e-05,
+      "loss": 0.2254,
+      "step": 975
+    },
+    {
+      "epoch": 2.8160919540229887,
+      "grad_norm": 1.2465283870697021,
+      "learning_rate": 1.313753844700059e-05,
+      "loss": 0.2233,
+      "step": 980
+    },
+    {
+      "epoch": 2.8304597701149428,
+      "grad_norm": 1.0080101490020752,
+      "learning_rate": 1.299618718644361e-05,
+      "loss": 0.1771,
+      "step": 985
+    },
+    {
+      "epoch": 2.844827586206897,
+      "grad_norm": 1.207362174987793,
+      "learning_rate": 1.285501687144548e-05,
+      "loss": 0.2265,
+      "step": 990
+    },
+    {
+      "epoch": 2.859195402298851,
+      "grad_norm": 1.1175612211227417,
+      "learning_rate": 1.2714040249774554e-05,
+      "loss": 0.19,
+      "step": 995
+    },
+    {
+      "epoch": 2.873563218390805,
+      "grad_norm": 1.1890356540679932,
+      "learning_rate": 1.2573270051708529e-05,
+      "loss": 0.1878,
+      "step": 1000
+    },
+    {
+      "epoch": 2.887931034482759,
+      "grad_norm": 1.3065342903137207,
+      "learning_rate": 1.2432718988884941e-05,
+      "loss": 0.1981,
+      "step": 1005
+    },
+    {
+      "epoch": 2.9022988505747125,
+      "grad_norm": 1.240831732749939,
+      "learning_rate": 1.2292399753153262e-05,
+      "loss": 0.2053,
+      "step": 1010
+    },
+    {
+      "epoch": 2.9166666666666665,
+      "grad_norm": 1.2138749361038208,
+      "learning_rate": 1.215232501542883e-05,
+      "loss": 0.1898,
+      "step": 1015
+    },
+    {
+      "epoch": 2.9310344827586206,
+      "grad_norm": 1.136871576309204,
+      "learning_rate": 1.201250742454868e-05,
+      "loss": 0.2288,
+      "step": 1020
+    },
+    {
+      "epoch": 2.9454022988505746,
+      "grad_norm": 1.0746241807937622,
+      "learning_rate": 1.1872959606129304e-05,
+      "loss": 0.2167,
+      "step": 1025
+    },
+    {
+      "epoch": 2.9597701149425286,
+      "grad_norm": 1.154268503189087,
+      "learning_rate": 1.1733694161426581e-05,
+      "loss": 0.2056,
+      "step": 1030
+    },
+    {
+      "epoch": 2.9741379310344827,
+      "grad_norm": 1.2475097179412842,
+      "learning_rate": 1.1594723666197858e-05,
+      "loss": 0.2068,
+      "step": 1035
+    },
+    {
+      "epoch": 2.9885057471264367,
+      "grad_norm": 1.196179986000061,
+      "learning_rate": 1.1456060669566362e-05,
+      "loss": 0.1926,
+      "step": 1040
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1740,
+  "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": 1.3909175239683604e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

16_128_e5_3e-5/checkpoint-1044/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7a8472bd637c6342cf14f5f02b2d5bc446377f37283692b5cde3421a32d36359
+size 7736

16_128_e5_3e-5/checkpoint-1044/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)

16_128_e5_3e-5/checkpoint-1392/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

16_128_e5_3e-5/checkpoint-1392/adapter_config.json

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

16_128_e5_3e-5/checkpoint-1392/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:00160af9aaa7337b28bfd4dd2d06cd0156756f6cf875b82dd51692c8b1ce2704
+size 791751704

16_128_e5_3e-5/checkpoint-1392/latest

@@ -0,0 +1 @@
+global_step1392

16_128_e5_3e-5/checkpoint-1392/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8207d2d435ecbe83ca9e9a23809eb0f4c7aec72acf6842f76785feff3270097b
+size 15920

16_128_e5_3e-5/checkpoint-1392/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:01878a71b286a50d8cef9c0fb6d75a975410f1cb77b5437b4e9b4d886dfce443
+size 15920

16_128_e5_3e-5/checkpoint-1392/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:354d871f4b20740544e4fea0d79776c38cadbdbf9476700e21ba92367ceb0bd7
+size 15920

16_128_e5_3e-5/checkpoint-1392/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:02114b3435436b198a09e6031a80f1cfc358bf97bc13f2644dc4398255f4fb35
+size 15920

16_128_e5_3e-5/checkpoint-1392/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:89912764b57d2e3d3292aa7167ebc1041639fce758b1d037d02dbf2ea23d5504
+size 15920

16_128_e5_3e-5/checkpoint-1392/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8b7060f7ab86f81cfc4cee774d06c6c83bcd83e57b0cfac3e6e4deb9c8f2e0a9
+size 15920

16_128_e5_3e-5/checkpoint-1392/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9acf875edadfa5e15cafe7c4ec3fe1a275a38a53606d76a3ff1a4ecbf6f16ccd
+size 15920

16_128_e5_3e-5/checkpoint-1392/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4dc50a3775f84a2a2f9836e7ff88f279daa82c16f241b6d113cd2bcbd17c4cf4
+size 15920

16_128_e5_3e-5/checkpoint-1392/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ca3ee0c09275fa923615e19d6e8020c66b730b58b13d944386dba7e457fa16f6
+size 1064

16_128_e5_3e-5/checkpoint-1392/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
+  }
+}

16_128_e5_3e-5/checkpoint-1392/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
+}

16_128_e5_3e-5/checkpoint-1392/trainer_state.json

@@ -0,0 +1,1980 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 1392,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.014367816091954023,
+      "grad_norm": 0.9956693053245544,
+      "learning_rate": 1.3793103448275862e-06,
+      "loss": 1.3425,
+      "step": 5
+    },
+    {
+      "epoch": 0.028735632183908046,
+      "grad_norm": 0.9614667296409607,
+      "learning_rate": 3.103448275862069e-06,
+      "loss": 1.361,
+      "step": 10
+    },
+    {
+      "epoch": 0.04310344827586207,
+      "grad_norm": 0.6255717277526855,
+      "learning_rate": 4.827586206896552e-06,
+      "loss": 1.3075,
+      "step": 15
+    },
+    {
+      "epoch": 0.05747126436781609,
+      "grad_norm": 0.6244459748268127,
+      "learning_rate": 6.551724137931035e-06,
+      "loss": 1.3881,
+      "step": 20
+    },
+    {
+      "epoch": 0.07183908045977011,
+      "grad_norm": 0.7454733848571777,
+      "learning_rate": 8.275862068965517e-06,
+      "loss": 1.2883,
+      "step": 25
+    },
+    {
+      "epoch": 0.08620689655172414,
+      "grad_norm": 0.6252493262290955,
+      "learning_rate": 9.999999999999999e-06,
+      "loss": 1.3055,
+      "step": 30
+    },
+    {
+      "epoch": 0.10057471264367816,
+      "grad_norm": 0.6072434782981873,
+      "learning_rate": 1.1724137931034483e-05,
+      "loss": 1.2972,
+      "step": 35
+    },
+    {
+      "epoch": 0.11494252873563218,
+      "grad_norm": 0.634437620639801,
+      "learning_rate": 1.3448275862068966e-05,
+      "loss": 1.2811,
+      "step": 40
+    },
+    {
+      "epoch": 0.12931034482758622,
+      "grad_norm": 0.4772552251815796,
+      "learning_rate": 1.517241379310345e-05,
+      "loss": 1.2304,
+      "step": 45
+    },
+    {
+      "epoch": 0.14367816091954022,
+      "grad_norm": 0.4635521173477173,
+      "learning_rate": 1.6896551724137932e-05,
+      "loss": 1.2537,
+      "step": 50
+    },
+    {
+      "epoch": 0.15804597701149425,
+      "grad_norm": 0.55922931432724,
+      "learning_rate": 1.8620689655172415e-05,
+      "loss": 1.2188,
+      "step": 55
+    },
+    {
+      "epoch": 0.1724137931034483,
+      "grad_norm": 0.44469261169433594,
+      "learning_rate": 2.0344827586206894e-05,
+      "loss": 1.2065,
+      "step": 60
+    },
+    {
+      "epoch": 0.1867816091954023,
+      "grad_norm": 0.48425620794296265,
+      "learning_rate": 2.206896551724138e-05,
+      "loss": 1.2016,
+      "step": 65
+    },
+    {
+      "epoch": 0.20114942528735633,
+      "grad_norm": 0.5222848653793335,
+      "learning_rate": 2.3793103448275862e-05,
+      "loss": 1.2123,
+      "step": 70
+    },
+    {
+      "epoch": 0.21551724137931033,
+      "grad_norm": 0.48502737283706665,
+      "learning_rate": 2.5517241379310345e-05,
+      "loss": 1.1739,
+      "step": 75
+    },
+    {
+      "epoch": 0.22988505747126436,
+      "grad_norm": 0.549568772315979,
+      "learning_rate": 2.7241379310344827e-05,
+      "loss": 1.2135,
+      "step": 80
+    },
+    {
+      "epoch": 0.2442528735632184,
+      "grad_norm": 0.538798451423645,
+      "learning_rate": 2.8965517241379313e-05,
+      "loss": 1.1525,
+      "step": 85
+    },
+    {
+      "epoch": 0.25862068965517243,
+      "grad_norm": 0.5352644920349121,
+      "learning_rate": 2.9999891638559405e-05,
+      "loss": 1.1424,
+      "step": 90
+    },
+    {
+      "epoch": 0.27298850574712646,
+      "grad_norm": 0.5414754152297974,
+      "learning_rate": 2.9998672590332875e-05,
+      "loss": 1.1321,
+      "step": 95
+    },
+    {
+      "epoch": 0.28735632183908044,
+      "grad_norm": 0.550433337688446,
+      "learning_rate": 2.999609915252684e-05,
+      "loss": 1.1292,
+      "step": 100
+    },
+    {
+      "epoch": 0.3017241379310345,
+      "grad_norm": 0.5901126265525818,
+      "learning_rate": 2.999217155752434e-05,
+      "loss": 1.1459,
+      "step": 105
+    },
+    {
+      "epoch": 0.3160919540229885,
+      "grad_norm": 0.6439724564552307,
+      "learning_rate": 2.9986890159989696e-05,
+      "loss": 1.0742,
+      "step": 110
+    },
+    {
+      "epoch": 0.33045977011494254,
+      "grad_norm": 0.5696046352386475,
+      "learning_rate": 2.9980255436836418e-05,
+      "loss": 1.1454,
+      "step": 115
+    },
+    {
+      "epoch": 0.3448275862068966,
+      "grad_norm": 0.6889880299568176,
+      "learning_rate": 2.9972267987184195e-05,
+      "loss": 1.086,
+      "step": 120
+    },
+    {
+      "epoch": 0.35919540229885055,
+      "grad_norm": 0.5843938589096069,
+      "learning_rate": 2.996292853230475e-05,
+      "loss": 1.0829,
+      "step": 125
+    },
+    {
+      "epoch": 0.3735632183908046,
+      "grad_norm": 0.6964264512062073,
+      "learning_rate": 2.9952237915556746e-05,
+      "loss": 1.0979,
+      "step": 130
+    },
+    {
+      "epoch": 0.3879310344827586,
+      "grad_norm": 0.741144597530365,
+      "learning_rate": 2.9940197102309603e-05,
+      "loss": 1.1022,
+      "step": 135
+    },
+    {
+      "epoch": 0.40229885057471265,
+      "grad_norm": 0.6913484930992126,
+      "learning_rate": 2.9926807179856342e-05,
+      "loss": 1.061,
+      "step": 140
+    },
+    {
+      "epoch": 0.4166666666666667,
+      "grad_norm": 0.6852656006813049,
+      "learning_rate": 2.9912069357315394e-05,
+      "loss": 1.0238,
+      "step": 145
+    },
+    {
+      "epoch": 0.43103448275862066,
+      "grad_norm": 0.7292410731315613,
+      "learning_rate": 2.989598496552141e-05,
+      "loss": 1.009,
+      "step": 150
+    },
+    {
+      "epoch": 0.4454022988505747,
+      "grad_norm": 0.6415665745735168,
+      "learning_rate": 2.9878555456905102e-05,
+      "loss": 1.0434,
+      "step": 155
+    },
+    {
+      "epoch": 0.45977011494252873,
+      "grad_norm": 0.7307606935501099,
+      "learning_rate": 2.985978240536207e-05,
+      "loss": 1.0383,
+      "step": 160
+    },
+    {
+      "epoch": 0.47413793103448276,
+      "grad_norm": 0.6481092572212219,
+      "learning_rate": 2.983966750611069e-05,
+      "loss": 1.0478,
+      "step": 165
+    },
+    {
+      "epoch": 0.4885057471264368,
+      "grad_norm": 0.694217324256897,
+      "learning_rate": 2.981821257553903e-05,
+      "loss": 1.0037,
+      "step": 170
+    },
+    {
+      "epoch": 0.5028735632183908,
+      "grad_norm": 0.7326360940933228,
+      "learning_rate": 2.9795419551040836e-05,
+      "loss": 1.0133,
+      "step": 175
+    },
+    {
+      "epoch": 0.5172413793103449,
+      "grad_norm": 0.8161800503730774,
+      "learning_rate": 2.9771290490840567e-05,
+      "loss": 0.9865,
+      "step": 180
+    },
+    {
+      "epoch": 0.5316091954022989,
+      "grad_norm": 0.8219459056854248,
+      "learning_rate": 2.9745827573807555e-05,
+      "loss": 0.9889,
+      "step": 185
+    },
+    {
+      "epoch": 0.5459770114942529,
+      "grad_norm": 0.7902458906173706,
+      "learning_rate": 2.9719033099259242e-05,
+      "loss": 0.9483,
+      "step": 190
+    },
+    {
+      "epoch": 0.5603448275862069,
+      "grad_norm": 0.7876456379890442,
+      "learning_rate": 2.969090948675355e-05,
+      "loss": 0.9768,
+      "step": 195
+    },
+    {
+      "epoch": 0.5747126436781609,
+      "grad_norm": 0.8362347483634949,
+      "learning_rate": 2.96614592758704e-05,
+      "loss": 0.9041,
+      "step": 200
+    },
+    {
+      "epoch": 0.5890804597701149,
+      "grad_norm": 1.062076210975647,
+      "learning_rate": 2.963068512598238e-05,
+      "loss": 0.9232,
+      "step": 205
+    },
+    {
+      "epoch": 0.603448275862069,
+      "grad_norm": 0.8888565897941589,
+      "learning_rate": 2.9598589816014606e-05,
+      "loss": 0.9191,
+      "step": 210
+    },
+    {
+      "epoch": 0.617816091954023,
+      "grad_norm": 0.8549149632453918,
+      "learning_rate": 2.9565176244193768e-05,
+      "loss": 0.8662,
+      "step": 215
+    },
+    {
+      "epoch": 0.632183908045977,
+      "grad_norm": 0.7843875288963318,
+      "learning_rate": 2.9530447427786454e-05,
+      "loss": 0.8909,
+      "step": 220
+    },
+    {
+      "epoch": 0.646551724137931,
+      "grad_norm": 0.9624897241592407,
+      "learning_rate": 2.949440650282664e-05,
+      "loss": 0.8995,
+      "step": 225
+    },
+    {
+      "epoch": 0.6609195402298851,
+      "grad_norm": 0.8020894527435303,
+      "learning_rate": 2.9457056723832563e-05,
+      "loss": 0.8472,
+      "step": 230
+    },
+    {
+      "epoch": 0.6752873563218391,
+      "grad_norm": 1.180816411972046,
+      "learning_rate": 2.9418401463512772e-05,
+      "loss": 0.9324,
+      "step": 235
+    },
+    {
+      "epoch": 0.6896551724137931,
+      "grad_norm": 0.985790491104126,
+      "learning_rate": 2.937844421246162e-05,
+      "loss": 0.8657,
+      "step": 240
+    },
+    {
+      "epoch": 0.7040229885057471,
+      "grad_norm": 0.9070732593536377,
+      "learning_rate": 2.9337188578844035e-05,
+      "loss": 0.8351,
+      "step": 245
+    },
+    {
+      "epoch": 0.7183908045977011,
+      "grad_norm": 0.945988118648529,
+      "learning_rate": 2.929463828806971e-05,
+      "loss": 0.8672,
+      "step": 250
+    },
+    {
+      "epoch": 0.7327586206896551,
+      "grad_norm": 0.9512550234794617,
+      "learning_rate": 2.9250797182456704e-05,
+      "loss": 0.8258,
+      "step": 255
+    },
+    {
+      "epoch": 0.7471264367816092,
+      "grad_norm": 1.025450587272644,
+      "learning_rate": 2.920566922088445e-05,
+      "loss": 0.8214,
+      "step": 260
+    },
+    {
+      "epoch": 0.7614942528735632,
+      "grad_norm": 0.9240112900733948,
+      "learning_rate": 2.9159258478436297e-05,
+      "loss": 0.7967,
+      "step": 265
+    },
+    {
+      "epoch": 0.7758620689655172,
+      "grad_norm": 1.0366438627243042,
+      "learning_rate": 2.911156914603152e-05,
+      "loss": 0.7778,
+      "step": 270
+    },
+    {
+      "epoch": 0.7902298850574713,
+      "grad_norm": 0.942286491394043,
+      "learning_rate": 2.9062605530046854e-05,
+      "loss": 0.745,
+      "step": 275
+    },
+    {
+      "epoch": 0.8045977011494253,
+      "grad_norm": 1.0067769289016724,
+      "learning_rate": 2.901237205192767e-05,
+      "loss": 0.7587,
+      "step": 280
+    },
+    {
+      "epoch": 0.8189655172413793,
+      "grad_norm": 0.9403071999549866,
+      "learning_rate": 2.8960873247788655e-05,
+      "loss": 0.7969,
+      "step": 285
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 0.9700637459754944,
+      "learning_rate": 2.8908113768004266e-05,
+      "loss": 0.8114,
+      "step": 290
+    },
+    {
+      "epoch": 0.8477011494252874,
+      "grad_norm": 0.9859459400177002,
+      "learning_rate": 2.8854098376788742e-05,
+      "loss": 0.7941,
+      "step": 295
+    },
+    {
+      "epoch": 0.8620689655172413,
+      "grad_norm": 0.8560702800750732,
+      "learning_rate": 2.8798831951765933e-05,
+      "loss": 0.8226,
+      "step": 300
+    },
+    {
+      "epoch": 0.8764367816091954,
+      "grad_norm": 0.9503181576728821,
+      "learning_rate": 2.8742319483528805e-05,
+      "loss": 0.7818,
+      "step": 305
+    },
+    {
+      "epoch": 0.8908045977011494,
+      "grad_norm": 0.9771649837493896,
+      "learning_rate": 2.8684566075188835e-05,
+      "loss": 0.777,
+      "step": 310
+    },
+    {
+      "epoch": 0.9051724137931034,
+      "grad_norm": 1.0014532804489136,
+      "learning_rate": 2.8625576941915162e-05,
+      "loss": 0.7499,
+      "step": 315
+    },
+    {
+      "epoch": 0.9195402298850575,
+      "grad_norm": 1.0961740016937256,
+      "learning_rate": 2.8565357410463664e-05,
+      "loss": 0.7755,
+      "step": 320
+    },
+    {
+      "epoch": 0.9339080459770115,
+      "grad_norm": 1.0916162729263306,
+      "learning_rate": 2.8503912918695952e-05,
+      "loss": 0.6966,
+      "step": 325
+    },
+    {
+      "epoch": 0.9482758620689655,
+      "grad_norm": 1.1250373125076294,
+      "learning_rate": 2.8441249015088322e-05,
+      "loss": 0.6778,
+      "step": 330
+    },
+    {
+      "epoch": 0.9626436781609196,
+      "grad_norm": 1.1927026510238647,
+      "learning_rate": 2.8377371358230733e-05,
+      "loss": 0.7595,
+      "step": 335
+    },
+    {
+      "epoch": 0.9770114942528736,
+      "grad_norm": 0.9869332313537598,
+      "learning_rate": 2.8312285716315823e-05,
+      "loss": 0.7361,
+      "step": 340
+    },
+    {
+      "epoch": 0.9913793103448276,
+      "grad_norm": 1.10604989528656,
+      "learning_rate": 2.8245997966618037e-05,
+      "loss": 0.6662,
+      "step": 345
+    },
+    {
+      "epoch": 1.0057471264367817,
+      "grad_norm": 1.1478843688964844,
+      "learning_rate": 2.8178514094962917e-05,
+      "loss": 0.6208,
+      "step": 350
+    },
+    {
+      "epoch": 1.0201149425287357,
+      "grad_norm": 1.15924072265625,
+      "learning_rate": 2.8109840195186565e-05,
+      "loss": 0.6334,
+      "step": 355
+    },
+    {
+      "epoch": 1.0344827586206897,
+      "grad_norm": 1.1480649709701538,
+      "learning_rate": 2.8039982468585372e-05,
+      "loss": 0.6328,
+      "step": 360
+    },
+    {
+      "epoch": 1.0488505747126438,
+      "grad_norm": 1.1573606729507446,
+      "learning_rate": 2.796894722335605e-05,
+      "loss": 0.5882,
+      "step": 365
+    },
+    {
+      "epoch": 1.0632183908045978,
+      "grad_norm": 1.1750164031982422,
+      "learning_rate": 2.7896740874025962e-05,
+      "loss": 0.6313,
+      "step": 370
+    },
+    {
+      "epoch": 1.0775862068965518,
+      "grad_norm": 1.1878570318222046,
+      "learning_rate": 2.782336994087393e-05,
+      "loss": 0.5848,
+      "step": 375
+    },
+    {
+      "epoch": 1.0919540229885056,
+      "grad_norm": 1.1398131847381592,
+      "learning_rate": 2.7748841049341407e-05,
+      "loss": 0.5686,
+      "step": 380
+    },
+    {
+      "epoch": 1.1063218390804597,
+      "grad_norm": 1.0254820585250854,
+      "learning_rate": 2.7673160929434242e-05,
+      "loss": 0.5807,
+      "step": 385
+    },
+    {
+      "epoch": 1.1206896551724137,
+      "grad_norm": 1.2164369821548462,
+      "learning_rate": 2.759633641511491e-05,
+      "loss": 0.6266,
+      "step": 390
+    },
+    {
+      "epoch": 1.1350574712643677,
+      "grad_norm": 1.0185494422912598,
+      "learning_rate": 2.7518374443685432e-05,
+      "loss": 0.5731,
+      "step": 395
+    },
+    {
+      "epoch": 1.1494252873563218,
+      "grad_norm": 1.2007052898406982,
+      "learning_rate": 2.7439282055160924e-05,
+      "loss": 0.5687,
+      "step": 400
+    },
+    {
+      "epoch": 1.1637931034482758,
+      "grad_norm": 1.2135870456695557,
+      "learning_rate": 2.735906639163387e-05,
+      "loss": 0.5538,
+      "step": 405
+    },
+    {
+      "epoch": 1.1781609195402298,
+      "grad_norm": 1.0088754892349243,
+      "learning_rate": 2.7277734696629197e-05,
+      "loss": 0.5804,
+      "step": 410
+    },
+    {
+      "epoch": 1.1925287356321839,
+      "grad_norm": 1.2317101955413818,
+      "learning_rate": 2.719529431445018e-05,
+      "loss": 0.5416,
+      "step": 415
+    },
+    {
+      "epoch": 1.206896551724138,
+      "grad_norm": 1.1315484046936035,
+      "learning_rate": 2.7111752689515243e-05,
+      "loss": 0.5089,
+      "step": 420
+    },
+    {
+      "epoch": 1.221264367816092,
+      "grad_norm": 1.205450415611267,
+      "learning_rate": 2.7027117365685727e-05,
+      "loss": 0.5906,
+      "step": 425
+    },
+    {
+      "epoch": 1.235632183908046,
+      "grad_norm": 1.1637941598892212,
+      "learning_rate": 2.6941395985584656e-05,
+      "loss": 0.5385,
+      "step": 430
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.1773250102996826,
+      "learning_rate": 2.6854596289906646e-05,
+      "loss": 0.5534,
+      "step": 435
+    },
+    {
+      "epoch": 1.264367816091954,
+      "grad_norm": 1.093550205230713,
+      "learning_rate": 2.6766726116718863e-05,
+      "loss": 0.5846,
+      "step": 440
+    },
+    {
+      "epoch": 1.278735632183908,
+      "grad_norm": 1.2478264570236206,
+      "learning_rate": 2.6677793400753285e-05,
+      "loss": 0.5546,
+      "step": 445
+    },
+    {
+      "epoch": 1.293103448275862,
+      "grad_norm": 1.2445526123046875,
+      "learning_rate": 2.6587806172690153e-05,
+      "loss": 0.5603,
+      "step": 450
+    },
+    {
+      "epoch": 1.3074712643678161,
+      "grad_norm": 1.061118483543396,
+      "learning_rate": 2.6496772558432844e-05,
+      "loss": 0.4761,
+      "step": 455
+    },
+    {
+      "epoch": 1.3218390804597702,
+      "grad_norm": 1.376889705657959,
+      "learning_rate": 2.6404700778374034e-05,
+      "loss": 0.5313,
+      "step": 460
+    },
+    {
+      "epoch": 1.3362068965517242,
+      "grad_norm": 1.097720742225647,
+      "learning_rate": 2.6311599146653446e-05,
+      "loss": 0.4878,
+      "step": 465
+    },
+    {
+      "epoch": 1.3505747126436782,
+      "grad_norm": 1.3083029985427856,
+      "learning_rate": 2.6217476070407048e-05,
+      "loss": 0.4992,
+      "step": 470
+    },
+    {
+      "epoch": 1.3649425287356323,
+      "grad_norm": 1.0825142860412598,
+      "learning_rate": 2.6122340049007895e-05,
+      "loss": 0.5441,
+      "step": 475
+    },
+    {
+      "epoch": 1.3793103448275863,
+      "grad_norm": 1.1726360321044922,
+      "learning_rate": 2.602619967329862e-05,
+      "loss": 0.5023,
+      "step": 480
+    },
+    {
+      "epoch": 1.3936781609195403,
+      "grad_norm": 1.739815354347229,
+      "learning_rate": 2.5929063624815698e-05,
+      "loss": 0.5557,
+      "step": 485
+    },
+    {
+      "epoch": 1.4080459770114944,
+      "grad_norm": 1.263541579246521,
+      "learning_rate": 2.583094067500547e-05,
+      "loss": 0.5174,
+      "step": 490
+    },
+    {
+      "epoch": 1.4224137931034484,
+      "grad_norm": 1.1408122777938843,
+      "learning_rate": 2.573183968443209e-05,
+      "loss": 0.5416,
+      "step": 495
+    },
+    {
+      "epoch": 1.4367816091954024,
+      "grad_norm": 1.0248563289642334,
+      "learning_rate": 2.5631769601977408e-05,
+      "loss": 0.5255,
+      "step": 500
+    },
+    {
+      "epoch": 1.4511494252873562,
+      "grad_norm": 1.1455802917480469,
+      "learning_rate": 2.553073946403289e-05,
+      "loss": 0.5121,
+      "step": 505
+    },
+    {
+      "epoch": 1.4655172413793103,
+      "grad_norm": 1.0702999830245972,
+      "learning_rate": 2.5428758393683605e-05,
+      "loss": 0.4928,
+      "step": 510
+    },
+    {
+      "epoch": 1.4798850574712643,
+      "grad_norm": 1.3014647960662842,
+      "learning_rate": 2.5325835599884427e-05,
+      "loss": 0.487,
+      "step": 515
+    },
+    {
+      "epoch": 1.4942528735632183,
+      "grad_norm": 1.1569771766662598,
+      "learning_rate": 2.5221980376628445e-05,
+      "loss": 0.5036,
+      "step": 520
+    },
+    {
+      "epoch": 1.5086206896551724,
+      "grad_norm": 1.3842607736587524,
+      "learning_rate": 2.5117202102107707e-05,
+      "loss": 0.478,
+      "step": 525
+    },
+    {
+      "epoch": 1.5229885057471264,
+      "grad_norm": 1.1965588331222534,
+      "learning_rate": 2.5011510237866398e-05,
+      "loss": 0.4697,
+      "step": 530
+    },
+    {
+      "epoch": 1.5373563218390804,
+      "grad_norm": 1.211634635925293,
+      "learning_rate": 2.4904914327946398e-05,
+      "loss": 0.5162,
+      "step": 535
+    },
+    {
+      "epoch": 1.5517241379310345,
+      "grad_norm": 1.0299171209335327,
+      "learning_rate": 2.4797423998025505e-05,
+      "loss": 0.4793,
+      "step": 540
+    },
+    {
+      "epoch": 1.5660919540229885,
+      "grad_norm": 1.1820342540740967,
+      "learning_rate": 2.4689048954548193e-05,
+      "loss": 0.487,
+      "step": 545
+    },
+    {
+      "epoch": 1.5804597701149425,
+      "grad_norm": 1.2619454860687256,
+      "learning_rate": 2.4579798983849127e-05,
+      "loss": 0.5021,
+      "step": 550
+    },
+    {
+      "epoch": 1.5948275862068966,
+      "grad_norm": 1.279123306274414,
+      "learning_rate": 2.446968395126946e-05,
+      "loss": 0.5236,
+      "step": 555
+    },
+    {
+      "epoch": 1.6091954022988506,
+      "grad_norm": 1.0806742906570435,
+      "learning_rate": 2.4358713800265978e-05,
+      "loss": 0.4738,
+      "step": 560
+    },
+    {
+      "epoch": 1.6235632183908046,
+      "grad_norm": 1.1664921045303345,
+      "learning_rate": 2.4246898551513182e-05,
+      "loss": 0.4352,
+      "step": 565
+    },
+    {
+      "epoch": 1.6379310344827587,
+      "grad_norm": 1.0607774257659912,
+      "learning_rate": 2.4134248301998455e-05,
+      "loss": 0.4097,
+      "step": 570
+    },
+    {
+      "epoch": 1.6522988505747125,
+      "grad_norm": 1.155291199684143,
+      "learning_rate": 2.402077322411026e-05,
+      "loss": 0.4855,
+      "step": 575
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 1.2184966802597046,
+      "learning_rate": 2.3906483564719586e-05,
+      "loss": 0.4937,
+      "step": 580
+    },
+    {
+      "epoch": 1.6810344827586206,
+      "grad_norm": 1.0277451276779175,
+      "learning_rate": 2.379138964425464e-05,
+      "loss": 0.3834,
+      "step": 585
+    },
+    {
+      "epoch": 1.6954022988505746,
+      "grad_norm": 1.153117299079895,
+      "learning_rate": 2.3675501855768913e-05,
+      "loss": 0.404,
+      "step": 590
+    },
+    {
+      "epoch": 1.7097701149425286,
+      "grad_norm": 1.2163881063461304,
+      "learning_rate": 2.3558830664002688e-05,
+      "loss": 0.4134,
+      "step": 595
+    },
+    {
+      "epoch": 1.7241379310344827,
+      "grad_norm": 1.148512363433838,
+      "learning_rate": 2.344138660443804e-05,
+      "loss": 0.4225,
+      "step": 600
+    },
+    {
+      "epoch": 1.7385057471264367,
+      "grad_norm": 1.2566574811935425,
+      "learning_rate": 2.3323180282347505e-05,
+      "loss": 0.4337,
+      "step": 605
+    },
+    {
+      "epoch": 1.7528735632183907,
+      "grad_norm": 1.3375940322875977,
+      "learning_rate": 2.320422237183641e-05,
+      "loss": 0.4509,
+      "step": 610
+    },
+    {
+      "epoch": 1.7672413793103448,
+      "grad_norm": 1.1347380876541138,
+      "learning_rate": 2.3084523614878962e-05,
+      "loss": 0.3973,
+      "step": 615
+    },
+    {
+      "epoch": 1.7816091954022988,
+      "grad_norm": 1.21699857711792,
+      "learning_rate": 2.2964094820348302e-05,
+      "loss": 0.4534,
+      "step": 620
+    },
+    {
+      "epoch": 1.7959770114942528,
+      "grad_norm": 1.137501835823059,
+      "learning_rate": 2.284294686304041e-05,
+      "loss": 0.4236,
+      "step": 625
+    },
+    {
+      "epoch": 1.8103448275862069,
+      "grad_norm": 1.1551192998886108,
+      "learning_rate": 2.2721090682692114e-05,
+      "loss": 0.4328,
+      "step": 630
+    },
+    {
+      "epoch": 1.8247126436781609,
+      "grad_norm": 1.149267315864563,
+      "learning_rate": 2.2598537282993232e-05,
+      "loss": 0.4095,
+      "step": 635
+    },
+    {
+      "epoch": 1.839080459770115,
+      "grad_norm": 1.2572345733642578,
+      "learning_rate": 2.247529773059293e-05,
+      "loss": 0.3723,
+      "step": 640
+    },
+    {
+      "epoch": 1.853448275862069,
+      "grad_norm": 1.1943402290344238,
+      "learning_rate": 2.235138315410039e-05,
+      "loss": 0.3944,
+      "step": 645
+    },
+    {
+      "epoch": 1.867816091954023,
+      "grad_norm": 1.2040687799453735,
+      "learning_rate": 2.2226804743079897e-05,
+      "loss": 0.3892,
+      "step": 650
+    },
+    {
+      "epoch": 1.882183908045977,
+      "grad_norm": 1.1246652603149414,
+      "learning_rate": 2.2101573747040418e-05,
+      "loss": 0.4065,
+      "step": 655
+    },
+    {
+      "epoch": 1.896551724137931,
+      "grad_norm": 1.3335580825805664,
+      "learning_rate": 2.1975701474419752e-05,
+      "loss": 0.3986,
+      "step": 660
+    },
+    {
+      "epoch": 1.910919540229885,
+      "grad_norm": 1.1872365474700928,
+      "learning_rate": 2.1849199291563375e-05,
+      "loss": 0.3806,
+      "step": 665
+    },
+    {
+      "epoch": 1.9252873563218391,
+      "grad_norm": 1.035778284072876,
+      "learning_rate": 2.1722078621698056e-05,
+      "loss": 0.3427,
+      "step": 670
+    },
+    {
+      "epoch": 1.9396551724137931,
+      "grad_norm": 1.2677980661392212,
+      "learning_rate": 2.1594350943900334e-05,
+      "loss": 0.4134,
+      "step": 675
+    },
+    {
+      "epoch": 1.9540229885057472,
+      "grad_norm": 2.563685655593872,
+      "learning_rate": 2.1466027792059938e-05,
+      "loss": 0.3871,
+      "step": 680
+    },
+    {
+      "epoch": 1.9683908045977012,
+      "grad_norm": 1.1063146591186523,
+      "learning_rate": 2.133712075383829e-05,
+      "loss": 0.3482,
+      "step": 685
+    },
+    {
+      "epoch": 1.9827586206896552,
+      "grad_norm": 1.1051040887832642,
+      "learning_rate": 2.1207641469622116e-05,
+      "loss": 0.3415,
+      "step": 690
+    },
+    {
+      "epoch": 1.9971264367816093,
+      "grad_norm": 1.1961126327514648,
+      "learning_rate": 2.1077601631472304e-05,
+      "loss": 0.378,
+      "step": 695
+    },
+    {
+      "epoch": 2.0114942528735633,
+      "grad_norm": 1.4363032579421997,
+      "learning_rate": 2.0947012982068133e-05,
+      "loss": 0.3195,
+      "step": 700
+    },
+    {
+      "epoch": 2.0258620689655173,
+      "grad_norm": 1.0249502658843994,
+      "learning_rate": 2.0815887313646867e-05,
+      "loss": 0.2616,
+      "step": 705
+    },
+    {
+      "epoch": 2.0402298850574714,
+      "grad_norm": 1.139730453491211,
+      "learning_rate": 2.0684236466938943e-05,
+      "loss": 0.2989,
+      "step": 710
+    },
+    {
+      "epoch": 2.0545977011494254,
+      "grad_norm": 1.1467976570129395,
+      "learning_rate": 2.055207233009872e-05,
+      "loss": 0.2754,
+      "step": 715
+    },
+    {
+      "epoch": 2.0689655172413794,
+      "grad_norm": 1.3604381084442139,
+      "learning_rate": 2.0419406837630962e-05,
+      "loss": 0.2883,
+      "step": 720
+    },
+    {
+      "epoch": 2.0833333333333335,
+      "grad_norm": 1.2879091501235962,
+      "learning_rate": 2.0286251969313194e-05,
+      "loss": 0.3149,
+      "step": 725
+    },
+    {
+      "epoch": 2.0977011494252875,
+      "grad_norm": 1.4303884506225586,
+      "learning_rate": 2.0152619749113864e-05,
+      "loss": 0.2978,
+      "step": 730
+    },
+    {
+      "epoch": 2.1120689655172415,
+      "grad_norm": 1.280913233757019,
+      "learning_rate": 2.00185222441066e-05,
+      "loss": 0.27,
+      "step": 735
+    },
+    {
+      "epoch": 2.1264367816091956,
+      "grad_norm": 1.3666342496871948,
+      "learning_rate": 1.9883971563380562e-05,
+      "loss": 0.3314,
+      "step": 740
+    },
+    {
+      "epoch": 2.1408045977011496,
+      "grad_norm": 1.1915347576141357,
+      "learning_rate": 1.974897985694693e-05,
+      "loss": 0.275,
+      "step": 745
+    },
+    {
+      "epoch": 2.1551724137931036,
+      "grad_norm": 1.2179591655731201,
+      "learning_rate": 1.96135593146418e-05,
+      "loss": 0.2529,
+      "step": 750
+    },
+    {
+      "epoch": 2.1695402298850572,
+      "grad_norm": 1.0577384233474731,
+      "learning_rate": 1.9477722165025422e-05,
+      "loss": 0.2589,
+      "step": 755
+    },
+    {
+      "epoch": 2.1839080459770113,
+      "grad_norm": 1.1857539415359497,
+      "learning_rate": 1.9341480674277935e-05,
+      "loss": 0.2922,
+      "step": 760
+    },
+    {
+      "epoch": 2.1982758620689653,
+      "grad_norm": 1.3017208576202393,
+      "learning_rate": 1.9204847145091755e-05,
+      "loss": 0.2778,
+      "step": 765
+    },
+    {
+      "epoch": 2.2126436781609193,
+      "grad_norm": 1.17992103099823,
+      "learning_rate": 1.9067833915560587e-05,
+      "loss": 0.2908,
+      "step": 770
+    },
+    {
+      "epoch": 2.2270114942528734,
+      "grad_norm": 1.127840280532837,
+      "learning_rate": 1.8930453358065356e-05,
+      "loss": 0.2796,
+      "step": 775
+    },
+    {
+      "epoch": 2.2413793103448274,
+      "grad_norm": 1.0646225214004517,
+      "learning_rate": 1.8792717878156902e-05,
+      "loss": 0.2868,
+      "step": 780
+    },
+    {
+      "epoch": 2.2557471264367814,
+      "grad_norm": 1.36182701587677,
+      "learning_rate": 1.8654639913435786e-05,
+      "loss": 0.2752,
+      "step": 785
+    },
+    {
+      "epoch": 2.2701149425287355,
+      "grad_norm": 1.0714491605758667,
+      "learning_rate": 1.8516231932429175e-05,
+      "loss": 0.2668,
+      "step": 790
+    },
+    {
+      "epoch": 2.2844827586206895,
+      "grad_norm": 1.2853004932403564,
+      "learning_rate": 1.8377506433464914e-05,
+      "loss": 0.2463,
+      "step": 795
+    },
+    {
+      "epoch": 2.2988505747126435,
+      "grad_norm": 1.011826515197754,
+      "learning_rate": 1.823847594354291e-05,
+      "loss": 0.239,
+      "step": 800
+    },
+    {
+      "epoch": 2.3132183908045976,
+      "grad_norm": 1.3612297773361206,
+      "learning_rate": 1.8099153017203943e-05,
+      "loss": 0.2804,
+      "step": 805
+    },
+    {
+      "epoch": 2.3275862068965516,
+      "grad_norm": 1.0135477781295776,
+      "learning_rate": 1.7959550235396002e-05,
+      "loss": 0.3225,
+      "step": 810
+    },
+    {
+      "epoch": 2.3419540229885056,
+      "grad_norm": 1.096671223640442,
+      "learning_rate": 1.781968020433819e-05,
+      "loss": 0.2386,
+      "step": 815
+    },
+    {
+      "epoch": 2.3563218390804597,
+      "grad_norm": 1.1345715522766113,
+      "learning_rate": 1.7679555554382382e-05,
+      "loss": 0.2524,
+      "step": 820
+    },
+    {
+      "epoch": 2.3706896551724137,
+      "grad_norm": 1.0843546390533447,
+      "learning_rate": 1.7539188938872708e-05,
+      "loss": 0.221,
+      "step": 825
+    },
+    {
+      "epoch": 2.3850574712643677,
+      "grad_norm": 1.3394324779510498,
+      "learning_rate": 1.7398593033002945e-05,
+      "loss": 0.284,
+      "step": 830
+    },
+    {
+      "epoch": 2.3994252873563218,
+      "grad_norm": 1.4917807579040527,
+      "learning_rate": 1.7257780532671903e-05,
+      "loss": 0.2559,
+      "step": 835
+    },
+    {
+      "epoch": 2.413793103448276,
+      "grad_norm": 1.0991957187652588,
+      "learning_rate": 1.7116764153337056e-05,
+      "loss": 0.2438,
+      "step": 840
+    },
+    {
+      "epoch": 2.42816091954023,
+      "grad_norm": 1.4984471797943115,
+      "learning_rate": 1.697555662886625e-05,
+      "loss": 0.2584,
+      "step": 845
+    },
+    {
+      "epoch": 2.442528735632184,
+      "grad_norm": 1.1856584548950195,
+      "learning_rate": 1.683417071038785e-05,
+      "loss": 0.2481,
+      "step": 850
+    },
+    {
+      "epoch": 2.456896551724138,
+      "grad_norm": 1.3198546171188354,
+      "learning_rate": 1.6692619165139334e-05,
+      "loss": 0.266,
+      "step": 855
+    },
+    {
+      "epoch": 2.471264367816092,
+      "grad_norm": 1.1649495363235474,
+      "learning_rate": 1.6550914775314372e-05,
+      "loss": 0.2795,
+      "step": 860
+    },
+    {
+      "epoch": 2.485632183908046,
+      "grad_norm": 1.236920714378357,
+      "learning_rate": 1.6409070336908588e-05,
+      "loss": 0.2235,
+      "step": 865
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.1417492628097534,
+      "learning_rate": 1.62670986585641e-05,
+      "loss": 0.2742,
+      "step": 870
+    },
+    {
+      "epoch": 2.514367816091954,
+      "grad_norm": 1.3992029428482056,
+      "learning_rate": 1.6125012560412856e-05,
+      "loss": 0.2246,
+      "step": 875
+    },
+    {
+      "epoch": 2.528735632183908,
+      "grad_norm": 1.2300302982330322,
+      "learning_rate": 1.5982824872918994e-05,
+      "loss": 0.2788,
+      "step": 880
+    },
+    {
+      "epoch": 2.543103448275862,
+      "grad_norm": 1.0692559480667114,
+      "learning_rate": 1.5840548435720226e-05,
+      "loss": 0.218,
+      "step": 885
+    },
+    {
+      "epoch": 2.557471264367816,
+      "grad_norm": 1.056885838508606,
+      "learning_rate": 1.569819609646842e-05,
+      "loss": 0.2386,
+      "step": 890
+    },
+    {
+      "epoch": 2.57183908045977,
+      "grad_norm": 1.2882906198501587,
+      "learning_rate": 1.5555780709669455e-05,
+      "loss": 0.2371,
+      "step": 895
+    },
+    {
+      "epoch": 2.586206896551724,
+      "grad_norm": 1.0865809917449951,
+      "learning_rate": 1.5413315135522434e-05,
+      "loss": 0.249,
+      "step": 900
+    },
+    {
+      "epoch": 2.6005747126436782,
+      "grad_norm": 1.100043535232544,
+      "learning_rate": 1.5270812238758407e-05,
+      "loss": 0.248,
+      "step": 905
+    },
+    {
+      "epoch": 2.6149425287356323,
+      "grad_norm": 1.1947323083877563,
+      "learning_rate": 1.5128284887478698e-05,
+      "loss": 0.2366,
+      "step": 910
+    },
+    {
+      "epoch": 2.6293103448275863,
+      "grad_norm": 1.1760265827178955,
+      "learning_rate": 1.4985745951992865e-05,
+      "loss": 0.2149,
+      "step": 915
+    },
+    {
+      "epoch": 2.6436781609195403,
+      "grad_norm": 1.2085545063018799,
+      "learning_rate": 1.4843208303656542e-05,
+      "loss": 0.2301,
+      "step": 920
+    },
+    {
+      "epoch": 2.6580459770114944,
+      "grad_norm": 1.1468777656555176,
+      "learning_rate": 1.4700684813709132e-05,
+      "loss": 0.2535,
+      "step": 925
+    },
+    {
+      "epoch": 2.6724137931034484,
+      "grad_norm": 1.318094253540039,
+      "learning_rate": 1.455818835211152e-05,
+      "loss": 0.2112,
+      "step": 930
+    },
+    {
+      "epoch": 2.6867816091954024,
+      "grad_norm": 1.2293264865875244,
+      "learning_rate": 1.441573178638392e-05,
+      "loss": 0.2199,
+      "step": 935
+    },
+    {
+      "epoch": 2.7011494252873565,
+      "grad_norm": 1.1567370891571045,
+      "learning_rate": 1.427332798044392e-05,
+      "loss": 0.2001,
+      "step": 940
+    },
+    {
+      "epoch": 2.7155172413793105,
+      "grad_norm": 1.162387490272522,
+      "learning_rate": 1.413098979344487e-05,
+      "loss": 0.1811,
+      "step": 945
+    },
+    {
+      "epoch": 2.7298850574712645,
+      "grad_norm": 1.17710542678833,
+      "learning_rate": 1.3988730078614684e-05,
+      "loss": 0.2425,
+      "step": 950
+    },
+    {
+      "epoch": 2.7442528735632186,
+      "grad_norm": 1.2836874723434448,
+      "learning_rate": 1.3846561682095192e-05,
+      "loss": 0.2217,
+      "step": 955
+    },
+    {
+      "epoch": 2.7586206896551726,
+      "grad_norm": 1.2637510299682617,
+      "learning_rate": 1.3704497441782117e-05,
+      "loss": 0.2311,
+      "step": 960
+    },
+    {
+      "epoch": 2.7729885057471266,
+      "grad_norm": 1.1894924640655518,
+      "learning_rate": 1.3562550186165829e-05,
+      "loss": 0.2246,
+      "step": 965
+    },
+    {
+      "epoch": 2.7873563218390807,
+      "grad_norm": 1.1791173219680786,
+      "learning_rate": 1.3420732733172882e-05,
+      "loss": 0.1856,
+      "step": 970
+    },
+    {
+      "epoch": 2.8017241379310347,
+      "grad_norm": 1.2785968780517578,
+      "learning_rate": 1.327905788900859e-05,
+      "loss": 0.2254,
+      "step": 975
+    },
+    {
+      "epoch": 2.8160919540229887,
+      "grad_norm": 1.2465283870697021,
+      "learning_rate": 1.313753844700059e-05,
+      "loss": 0.2233,
+      "step": 980
+    },
+    {
+      "epoch": 2.8304597701149428,
+      "grad_norm": 1.0080101490020752,
+      "learning_rate": 1.299618718644361e-05,
+      "loss": 0.1771,
+      "step": 985
+    },
+    {
+      "epoch": 2.844827586206897,
+      "grad_norm": 1.207362174987793,
+      "learning_rate": 1.285501687144548e-05,
+      "loss": 0.2265,
+      "step": 990
+    },
+    {
+      "epoch": 2.859195402298851,
+      "grad_norm": 1.1175612211227417,
+      "learning_rate": 1.2714040249774554e-05,
+      "loss": 0.19,
+      "step": 995
+    },
+    {
+      "epoch": 2.873563218390805,
+      "grad_norm": 1.1890356540679932,
+      "learning_rate": 1.2573270051708529e-05,
+      "loss": 0.1878,
+      "step": 1000
+    },
+    {
+      "epoch": 2.887931034482759,
+      "grad_norm": 1.3065342903137207,
+      "learning_rate": 1.2432718988884941e-05,
+      "loss": 0.1981,
+      "step": 1005
+    },
+    {
+      "epoch": 2.9022988505747125,
+      "grad_norm": 1.240831732749939,
+      "learning_rate": 1.2292399753153262e-05,
+      "loss": 0.2053,
+      "step": 1010
+    },
+    {
+      "epoch": 2.9166666666666665,
+      "grad_norm": 1.2138749361038208,
+      "learning_rate": 1.215232501542883e-05,
+      "loss": 0.1898,
+      "step": 1015
+    },
+    {
+      "epoch": 2.9310344827586206,
+      "grad_norm": 1.136871576309204,
+      "learning_rate": 1.201250742454868e-05,
+      "loss": 0.2288,
+      "step": 1020
+    },
+    {
+      "epoch": 2.9454022988505746,
+      "grad_norm": 1.0746241807937622,
+      "learning_rate": 1.1872959606129304e-05,
+      "loss": 0.2167,
+      "step": 1025
+    },
+    {
+      "epoch": 2.9597701149425286,
+      "grad_norm": 1.154268503189087,
+      "learning_rate": 1.1733694161426581e-05,
+      "loss": 0.2056,
+      "step": 1030
+    },
+    {
+      "epoch": 2.9741379310344827,
+      "grad_norm": 1.2475097179412842,
+      "learning_rate": 1.1594723666197858e-05,
+      "loss": 0.2068,
+      "step": 1035
+    },
+    {
+      "epoch": 2.9885057471264367,
+      "grad_norm": 1.196179986000061,
+      "learning_rate": 1.1456060669566362e-05,
+      "loss": 0.1926,
+      "step": 1040
+    },
+    {
+      "epoch": 3.0028735632183907,
+      "grad_norm": 1.0581117868423462,
+      "learning_rate": 1.1317717692888014e-05,
+      "loss": 0.1754,
+      "step": 1045
+    },
+    {
+      "epoch": 3.0172413793103448,
+      "grad_norm": 1.2780325412750244,
+      "learning_rate": 1.1179707228620711e-05,
+      "loss": 0.1565,
+      "step": 1050
+    },
+    {
+      "epoch": 3.031609195402299,
+      "grad_norm": 1.061492681503296,
+      "learning_rate": 1.1042041739196292e-05,
+      "loss": 0.1421,
+      "step": 1055
+    },
+    {
+      "epoch": 3.045977011494253,
+      "grad_norm": 1.2962989807128906,
+      "learning_rate": 1.0904733655895134e-05,
+      "loss": 0.1725,
+      "step": 1060
+    },
+    {
+      "epoch": 3.060344827586207,
+      "grad_norm": 1.0233877897262573,
+      "learning_rate": 1.0767795377723613e-05,
+      "loss": 0.1417,
+      "step": 1065
+    },
+    {
+      "epoch": 3.074712643678161,
+      "grad_norm": 1.0480644702911377,
+      "learning_rate": 1.0631239270294487e-05,
+      "loss": 0.1588,
+      "step": 1070
+    },
+    {
+      "epoch": 3.089080459770115,
+      "grad_norm": 1.0096197128295898,
+      "learning_rate": 1.049507766471025e-05,
+      "loss": 0.1474,
+      "step": 1075
+    },
+    {
+      "epoch": 3.103448275862069,
+      "grad_norm": 0.950186550617218,
+      "learning_rate": 1.0359322856449619e-05,
+      "loss": 0.1244,
+      "step": 1080
+    },
+    {
+      "epoch": 3.117816091954023,
+      "grad_norm": 1.0260823965072632,
+      "learning_rate": 1.0223987104257262e-05,
+      "loss": 0.1303,
+      "step": 1085
+    },
+    {
+      "epoch": 3.132183908045977,
+      "grad_norm": 1.0128180980682373,
+      "learning_rate": 1.0089082629036825e-05,
+      "loss": 0.1388,
+      "step": 1090
+    },
+    {
+      "epoch": 3.146551724137931,
+      "grad_norm": 1.1141488552093506,
+      "learning_rate": 9.954621612747371e-06,
+      "loss": 0.1263,
+      "step": 1095
+    },
+    {
+      "epoch": 3.160919540229885,
+      "grad_norm": 1.087923288345337,
+      "learning_rate": 9.820616197303344e-06,
+      "loss": 0.1344,
+      "step": 1100
+    },
+    {
+      "epoch": 3.175287356321839,
+      "grad_norm": 1.0373666286468506,
+      "learning_rate": 9.687078483478135e-06,
+      "loss": 0.1338,
+      "step": 1105
+    },
+    {
+      "epoch": 3.189655172413793,
+      "grad_norm": 1.2134506702423096,
+      "learning_rate": 9.554020529811395e-06,
+      "loss": 0.1476,
+      "step": 1110
+    },
+    {
+      "epoch": 3.204022988505747,
+      "grad_norm": 0.9992417097091675,
+      "learning_rate": 9.421454351520118e-06,
+      "loss": 0.1518,
+      "step": 1115
+    },
+    {
+      "epoch": 3.218390804597701,
+      "grad_norm": 1.024229645729065,
+      "learning_rate": 9.289391919413686e-06,
+      "loss": 0.1568,
+      "step": 1120
+    },
+    {
+      "epoch": 3.2327586206896552,
+      "grad_norm": 0.9758639931678772,
+      "learning_rate": 9.157845158812881e-06,
+      "loss": 0.1359,
+      "step": 1125
+    },
+    {
+      "epoch": 3.2471264367816093,
+      "grad_norm": 1.1693804264068604,
+      "learning_rate": 9.026825948473015e-06,
+      "loss": 0.158,
+      "step": 1130
+    },
+    {
+      "epoch": 3.2614942528735633,
+      "grad_norm": 1.2794687747955322,
+      "learning_rate": 8.896346119511283e-06,
+      "loss": 0.137,
+      "step": 1135
+    },
+    {
+      "epoch": 3.2758620689655173,
+      "grad_norm": 1.1603080034255981,
+      "learning_rate": 8.766417454338408e-06,
+      "loss": 0.1462,
+      "step": 1140
+    },
+    {
+      "epoch": 3.2902298850574714,
+      "grad_norm": 0.9159196019172668,
+      "learning_rate": 8.637051685594658e-06,
+      "loss": 0.1451,
+      "step": 1145
+    },
+    {
+      "epoch": 3.3045977011494254,
+      "grad_norm": 1.1294188499450684,
+      "learning_rate": 8.508260495090427e-06,
+      "loss": 0.132,
+      "step": 1150
+    },
+    {
+      "epoch": 3.3189655172413794,
+      "grad_norm": 1.0812433958053589,
+      "learning_rate": 8.380055512751315e-06,
+      "loss": 0.1412,
+      "step": 1155
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 1.1534279584884644,
+      "learning_rate": 8.25244831556795e-06,
+      "loss": 0.1422,
+      "step": 1160
+    },
+    {
+      "epoch": 3.3477011494252875,
+      "grad_norm": 1.0479689836502075,
+      "learning_rate": 8.125450426550595e-06,
+      "loss": 0.1528,
+      "step": 1165
+    },
+    {
+      "epoch": 3.3620689655172415,
+      "grad_norm": 1.1747006177902222,
+      "learning_rate": 7.999073313688595e-06,
+      "loss": 0.1369,
+      "step": 1170
+    },
+    {
+      "epoch": 3.3764367816091956,
+      "grad_norm": 1.2122522592544556,
+      "learning_rate": 7.873328388914827e-06,
+      "loss": 0.1243,
+      "step": 1175
+    },
+    {
+      "epoch": 3.3908045977011496,
+      "grad_norm": 1.1704413890838623,
+      "learning_rate": 7.748227007075167e-06,
+      "loss": 0.1295,
+      "step": 1180
+    },
+    {
+      "epoch": 3.405172413793103,
+      "grad_norm": 1.1030290126800537,
+      "learning_rate": 7.623780464903189e-06,
+      "loss": 0.1115,
+      "step": 1185
+    },
+    {
+      "epoch": 3.4195402298850572,
+      "grad_norm": 1.128061294555664,
+      "learning_rate": 7.500000000000004e-06,
+      "loss": 0.1255,
+      "step": 1190
+    },
+    {
+      "epoch": 3.4339080459770113,
+      "grad_norm": 1.006690502166748,
+      "learning_rate": 7.37689678981953e-06,
+      "loss": 0.1312,
+      "step": 1195
+    },
+    {
+      "epoch": 3.4482758620689653,
+      "grad_norm": 1.1587826013565063,
+      "learning_rate": 7.25448195065918e-06,
+      "loss": 0.1289,
+      "step": 1200
+    },
+    {
+      "epoch": 3.4626436781609193,
+      "grad_norm": 1.2582279443740845,
+      "learning_rate": 7.132766536656005e-06,
+      "loss": 0.1409,
+      "step": 1205
+    },
+    {
+      "epoch": 3.4770114942528734,
+      "grad_norm": 0.9870187640190125,
+      "learning_rate": 7.011761538788527e-06,
+      "loss": 0.1359,
+      "step": 1210
+    },
+    {
+      "epoch": 3.4913793103448274,
+      "grad_norm": 0.8670295476913452,
+      "learning_rate": 6.8914778838842655e-06,
+      "loss": 0.1371,
+      "step": 1215
+    },
+    {
+      "epoch": 3.5057471264367814,
+      "grad_norm": 1.0406819581985474,
+      "learning_rate": 6.7719264336329885e-06,
+      "loss": 0.1306,
+      "step": 1220
+    },
+    {
+      "epoch": 3.5201149425287355,
+      "grad_norm": 1.1038048267364502,
+      "learning_rate": 6.653117983605936e-06,
+      "loss": 0.1385,
+      "step": 1225
+    },
+    {
+      "epoch": 3.5344827586206895,
+      "grad_norm": 1.0640748739242554,
+      "learning_rate": 6.535063262280964e-06,
+      "loss": 0.1336,
+      "step": 1230
+    },
+    {
+      "epoch": 3.5488505747126435,
+      "grad_norm": 1.0289545059204102,
+      "learning_rate": 6.417772930073744e-06,
+      "loss": 0.1423,
+      "step": 1235
+    },
+    {
+      "epoch": 3.5632183908045976,
+      "grad_norm": 1.013983964920044,
+      "learning_rate": 6.301257578375125e-06,
+      "loss": 0.1275,
+      "step": 1240
+    },
+    {
+      "epoch": 3.5775862068965516,
+      "grad_norm": 0.8542355298995972,
+      "learning_rate": 6.185527728594747e-06,
+      "loss": 0.1092,
+      "step": 1245
+    },
+    {
+      "epoch": 3.5919540229885056,
+      "grad_norm": 1.0594922304153442,
+      "learning_rate": 6.070593831210907e-06,
+      "loss": 0.1133,
+      "step": 1250
+    },
+    {
+      "epoch": 3.6063218390804597,
+      "grad_norm": 1.0956785678863525,
+      "learning_rate": 5.956466264826925e-06,
+      "loss": 0.126,
+      "step": 1255
+    },
+    {
+      "epoch": 3.6206896551724137,
+      "grad_norm": 0.9673142433166504,
+      "learning_rate": 5.843155335233904e-06,
+      "loss": 0.1035,
+      "step": 1260
+    },
+    {
+      "epoch": 3.6350574712643677,
+      "grad_norm": 0.8828128576278687,
+      "learning_rate": 5.730671274480139e-06,
+      "loss": 0.1033,
+      "step": 1265
+    },
+    {
+      "epoch": 3.6494252873563218,
+      "grad_norm": 0.9963325262069702,
+      "learning_rate": 5.619024239947146e-06,
+      "loss": 0.126,
+      "step": 1270
+    },
+    {
+      "epoch": 3.663793103448276,
+      "grad_norm": 1.107493281364441,
+      "learning_rate": 5.508224313432441e-06,
+      "loss": 0.1316,
+      "step": 1275
+    },
+    {
+      "epoch": 3.67816091954023,
+      "grad_norm": 0.8668447732925415,
+      "learning_rate": 5.3982815002391655e-06,
+      "loss": 0.1082,
+      "step": 1280
+    },
+    {
+      "epoch": 3.692528735632184,
+      "grad_norm": 0.9604552984237671,
+      "learning_rate": 5.289205728272587e-06,
+      "loss": 0.1114,
+      "step": 1285
+    },
+    {
+      "epoch": 3.706896551724138,
+      "grad_norm": 1.0154809951782227,
+      "learning_rate": 5.18100684714359e-06,
+      "loss": 0.1047,
+      "step": 1290
+    },
+    {
+      "epoch": 3.721264367816092,
+      "grad_norm": 0.9940729737281799,
+      "learning_rate": 5.0736946272792756e-06,
+      "loss": 0.0961,
+      "step": 1295
+    },
+    {
+      "epoch": 3.735632183908046,
+      "grad_norm": 0.8823755979537964,
+      "learning_rate": 4.9672787590406665e-06,
+      "loss": 0.1195,
+      "step": 1300
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.9927054643630981,
+      "learning_rate": 4.861768851847684e-06,
+      "loss": 0.1107,
+      "step": 1305
+    },
+    {
+      "epoch": 3.764367816091954,
+      "grad_norm": 1.2087199687957764,
+      "learning_rate": 4.757174433311376e-06,
+      "loss": 0.1282,
+      "step": 1310
+    },
+    {
+      "epoch": 3.778735632183908,
+      "grad_norm": 0.9635007977485657,
+      "learning_rate": 4.6535049483736095e-06,
+      "loss": 0.1295,
+      "step": 1315
+    },
+    {
+      "epoch": 3.793103448275862,
+      "grad_norm": 0.9805769920349121,
+      "learning_rate": 4.55076975845415e-06,
+      "loss": 0.1083,
+      "step": 1320
+    },
+    {
+      "epoch": 3.807471264367816,
+      "grad_norm": 1.0073679685592651,
+      "learning_rate": 4.448978140605339e-06,
+      "loss": 0.1269,
+      "step": 1325
+    },
+    {
+      "epoch": 3.82183908045977,
+      "grad_norm": 1.1844137907028198,
+      "learning_rate": 4.3481392866743605e-06,
+      "loss": 0.107,
+      "step": 1330
+    },
+    {
+      "epoch": 3.836206896551724,
+      "grad_norm": 0.8615861535072327,
+      "learning_rate": 4.248262302473233e-06,
+      "loss": 0.1046,
+      "step": 1335
+    },
+    {
+      "epoch": 3.8505747126436782,
+      "grad_norm": 1.0689043998718262,
+      "learning_rate": 4.149356206956512e-06,
+      "loss": 0.1223,
+      "step": 1340
+    },
+    {
+      "epoch": 3.8649425287356323,
+      "grad_norm": 1.1487630605697632,
+      "learning_rate": 4.051429931406912e-06,
+      "loss": 0.1225,
+      "step": 1345
+    },
+    {
+      "epoch": 3.8793103448275863,
+      "grad_norm": 0.9234219789505005,
+      "learning_rate": 3.9544923186287766e-06,
+      "loss": 0.0994,
+      "step": 1350
+    },
+    {
+      "epoch": 3.8936781609195403,
+      "grad_norm": 0.8749639391899109,
+      "learning_rate": 3.858552122149573e-06,
+      "loss": 0.1081,
+      "step": 1355
+    },
+    {
+      "epoch": 3.9080459770114944,
+      "grad_norm": 1.129578948020935,
+      "learning_rate": 3.7636180054294684e-06,
+      "loss": 0.1065,
+      "step": 1360
+    },
+    {
+      "epoch": 3.9224137931034484,
+      "grad_norm": 1.162880539894104,
+      "learning_rate": 3.6696985410789764e-06,
+      "loss": 0.1229,
+      "step": 1365
+    },
+    {
+      "epoch": 3.9367816091954024,
+      "grad_norm": 0.9815517067909241,
+      "learning_rate": 3.576802210084868e-06,
+      "loss": 0.1017,
+      "step": 1370
+    },
+    {
+      "epoch": 3.9511494252873565,
+      "grad_norm": 0.885273277759552,
+      "learning_rate": 3.484937401044325e-06,
+      "loss": 0.1061,
+      "step": 1375
+    },
+    {
+      "epoch": 3.9655172413793105,
+      "grad_norm": 0.9229301810264587,
+      "learning_rate": 3.394112409407455e-06,
+      "loss": 0.0948,
+      "step": 1380
+    },
+    {
+      "epoch": 3.9798850574712645,
+      "grad_norm": 1.038833498954773,
+      "learning_rate": 3.304335436728198e-06,
+      "loss": 0.1007,
+      "step": 1385
+    },
+    {
+      "epoch": 3.9942528735632186,
+      "grad_norm": 1.004219889640808,
+      "learning_rate": 3.2156145899237153e-06,
+      "loss": 0.1178,
+      "step": 1390
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1740,
+  "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": 1.8549951108114022e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

16_128_e5_3e-5/checkpoint-1392/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7a8472bd637c6342cf14f5f02b2d5bc446377f37283692b5cde3421a32d36359
+size 7736

16_128_e5_3e-5/checkpoint-1392/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)

16_128_e5_3e-5/checkpoint-1740/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

16_128_e5_3e-5/checkpoint-1740/adapter_config.json

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

16_128_e5_3e-5/checkpoint-1740/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fd4e30bb3d4fc85a6680a195d603978fab7b87ff2564e8f27fe7f8a9fbf5f7ae
+size 791751704

16_128_e5_3e-5/checkpoint-1740/latest

@@ -0,0 +1 @@
+global_step1740

16_128_e5_3e-5/checkpoint-1740/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc7ca694b5a0bf97c3c7a7abab3a6518e360cfded5bcf6bc703759d2e88d9987
+size 15920

16_128_e5_3e-5/checkpoint-1740/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a7dbab92a18caaefe173fc7fd2d6b281001066cf39f760ff0424373479576fa0
+size 15920

16_128_e5_3e-5/checkpoint-1740/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b5b59b71b0b7e61c269d3c22bcf33ca314d7552ab67b36f9e5bcac7d4e014728
+size 15920