Upload folder using huggingface_hub

31_128_e5_3e-5/checkpoint-1188/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

31_128_e5_3e-5/checkpoint-1188/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": [
+    "k_proj",
+    "q_proj",
+    "up_proj",
+    "o_proj",
+    "gate_proj",
+    "v_proj",
+    "down_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

31_128_e5_3e-5/checkpoint-1188/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dc495cbaf8a9fff850e877744b0bea9cdee7630166d10b9bd91e9fda3e2bdab4
+size 791751704

31_128_e5_3e-5/checkpoint-1188/latest

@@ -0,0 +1 @@
+global_step1188

31_128_e5_3e-5/checkpoint-1188/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2c888ff64f2088b8c242abd6562e8580cda783a1c1b0e97f1b220df5f762e4b5
+size 15920

31_128_e5_3e-5/checkpoint-1188/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:db3da393e1b4cf694eb33cfb732a361e9aa084c29772ec70e91b3dae912372f2
+size 15920

31_128_e5_3e-5/checkpoint-1188/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c4568af5560ed64d68465a21aea1690696f6a2405c4371e558c080c079d69ad3
+size 15920

31_128_e5_3e-5/checkpoint-1188/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28d5dbee841e4dcc4642811d755f4e7e0ce33dc0b39597c7dfd09969500eef31
+size 15920

31_128_e5_3e-5/checkpoint-1188/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:10f39eae90a3bc188603f1de6469922b9c36ccb22835d5889bd97b73d2b464e9
+size 15920

31_128_e5_3e-5/checkpoint-1188/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6fb21a7bdb0caa5c794e0c780dd1bcc26593fe9a55658bb6dfbdfd7a28324b58
+size 15920

31_128_e5_3e-5/checkpoint-1188/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d99ed02b034b59705fd05492d88396024e15a4a400394a22cee99c3299eef3fd
+size 15920

31_128_e5_3e-5/checkpoint-1188/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:be148e481c6b028785dbb89097dcf737389cbb554bbadc1fb64c5979537e2eab
+size 15920

31_128_e5_3e-5/checkpoint-1188/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ef7d4d8796189a49531f8e3979c0108e1230d3e9e6e42fc003aa38248f89708d
+size 1064

31_128_e5_3e-5/checkpoint-1188/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
+  }
+}

31_128_e5_3e-5/checkpoint-1188/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
+}

31_128_e5_3e-5/checkpoint-1188/trainer_state.json

@@ -0,0 +1,1693 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 1188,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.012626262626262626,
+      "grad_norm": 1.2214584350585938,
+      "learning_rate": 1.2121212121212122e-06,
+      "loss": 1.3078,
+      "step": 5
+    },
+    {
+      "epoch": 0.025252525252525252,
+      "grad_norm": 0.9447234272956848,
+      "learning_rate": 2.7272727272727272e-06,
+      "loss": 1.3299,
+      "step": 10
+    },
+    {
+      "epoch": 0.03787878787878788,
+      "grad_norm": 0.6590636968612671,
+      "learning_rate": 4.242424242424242e-06,
+      "loss": 1.2777,
+      "step": 15
+    },
+    {
+      "epoch": 0.050505050505050504,
+      "grad_norm": 0.6515881419181824,
+      "learning_rate": 5.757575757575758e-06,
+      "loss": 1.2685,
+      "step": 20
+    },
+    {
+      "epoch": 0.06313131313131314,
+      "grad_norm": 0.5434719324111938,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 1.2521,
+      "step": 25
+    },
+    {
+      "epoch": 0.07575757575757576,
+      "grad_norm": 0.5539897084236145,
+      "learning_rate": 8.787878787878788e-06,
+      "loss": 1.1989,
+      "step": 30
+    },
+    {
+      "epoch": 0.08838383838383838,
+      "grad_norm": 0.5708086490631104,
+      "learning_rate": 1.0303030303030302e-05,
+      "loss": 1.1534,
+      "step": 35
+    },
+    {
+      "epoch": 0.10101010101010101,
+      "grad_norm": 0.5267711281776428,
+      "learning_rate": 1.1818181818181819e-05,
+      "loss": 1.2464,
+      "step": 40
+    },
+    {
+      "epoch": 0.11363636363636363,
+      "grad_norm": 0.5095251798629761,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.1572,
+      "step": 45
+    },
+    {
+      "epoch": 0.12626262626262627,
+      "grad_norm": 0.4897612929344177,
+      "learning_rate": 1.484848484848485e-05,
+      "loss": 1.1698,
+      "step": 50
+    },
+    {
+      "epoch": 0.1388888888888889,
+      "grad_norm": 0.4447384476661682,
+      "learning_rate": 1.6363636363636363e-05,
+      "loss": 1.0932,
+      "step": 55
+    },
+    {
+      "epoch": 0.15151515151515152,
+      "grad_norm": 0.5444252490997314,
+      "learning_rate": 1.7878787878787877e-05,
+      "loss": 1.1421,
+      "step": 60
+    },
+    {
+      "epoch": 0.16414141414141414,
+      "grad_norm": 0.4492619037628174,
+      "learning_rate": 1.9393939393939395e-05,
+      "loss": 1.1182,
+      "step": 65
+    },
+    {
+      "epoch": 0.17676767676767677,
+      "grad_norm": 0.541714072227478,
+      "learning_rate": 2.090909090909091e-05,
+      "loss": 1.1767,
+      "step": 70
+    },
+    {
+      "epoch": 0.1893939393939394,
+      "grad_norm": 0.43421369791030884,
+      "learning_rate": 2.2424242424242424e-05,
+      "loss": 1.0917,
+      "step": 75
+    },
+    {
+      "epoch": 0.20202020202020202,
+      "grad_norm": 0.4997319281101227,
+      "learning_rate": 2.3939393939393942e-05,
+      "loss": 1.1036,
+      "step": 80
+    },
+    {
+      "epoch": 0.21464646464646464,
+      "grad_norm": 0.4616036117076874,
+      "learning_rate": 2.5454545454545457e-05,
+      "loss": 1.1136,
+      "step": 85
+    },
+    {
+      "epoch": 0.22727272727272727,
+      "grad_norm": 0.5620744228363037,
+      "learning_rate": 2.696969696969697e-05,
+      "loss": 1.1174,
+      "step": 90
+    },
+    {
+      "epoch": 0.2398989898989899,
+      "grad_norm": 0.48610466718673706,
+      "learning_rate": 2.8484848484848486e-05,
+      "loss": 1.1173,
+      "step": 95
+    },
+    {
+      "epoch": 0.25252525252525254,
+      "grad_norm": 0.8980438113212585,
+      "learning_rate": 3e-05,
+      "loss": 1.0533,
+      "step": 100
+    },
+    {
+      "epoch": 0.26515151515151514,
+      "grad_norm": 0.5355147123336792,
+      "learning_rate": 2.9999476976861445e-05,
+      "loss": 1.0703,
+      "step": 105
+    },
+    {
+      "epoch": 0.2777777777777778,
+      "grad_norm": 0.6903505325317383,
+      "learning_rate": 2.999790794391954e-05,
+      "loss": 1.0783,
+      "step": 110
+    },
+    {
+      "epoch": 0.2904040404040404,
+      "grad_norm": 0.5622875094413757,
+      "learning_rate": 2.999529301059302e-05,
+      "loss": 1.0516,
+      "step": 115
+    },
+    {
+      "epoch": 0.30303030303030304,
+      "grad_norm": 0.5711376070976257,
+      "learning_rate": 2.9991632359237976e-05,
+      "loss": 1.016,
+      "step": 120
+    },
+    {
+      "epoch": 0.31565656565656564,
+      "grad_norm": 0.6692824959754944,
+      "learning_rate": 2.9986926245135113e-05,
+      "loss": 1.0827,
+      "step": 125
+    },
+    {
+      "epoch": 0.3282828282828283,
+      "grad_norm": 0.5953907370567322,
+      "learning_rate": 2.9981174996471984e-05,
+      "loss": 1.0046,
+      "step": 130
+    },
+    {
+      "epoch": 0.3409090909090909,
+      "grad_norm": 0.5587031841278076,
+      "learning_rate": 2.9974379014320066e-05,
+      "loss": 1.0265,
+      "step": 135
+    },
+    {
+      "epoch": 0.35353535353535354,
+      "grad_norm": 0.5270025730133057,
+      "learning_rate": 2.996653877260682e-05,
+      "loss": 1.0009,
+      "step": 140
+    },
+    {
+      "epoch": 0.3661616161616162,
+      "grad_norm": 0.6027882695198059,
+      "learning_rate": 2.9957654818082615e-05,
+      "loss": 1.0222,
+      "step": 145
+    },
+    {
+      "epoch": 0.3787878787878788,
+      "grad_norm": 0.5629182457923889,
+      "learning_rate": 2.9947727770282626e-05,
+      "loss": 0.9617,
+      "step": 150
+    },
+    {
+      "epoch": 0.39141414141414144,
+      "grad_norm": 0.6864057183265686,
+      "learning_rate": 2.9936758321483616e-05,
+      "loss": 0.9682,
+      "step": 155
+    },
+    {
+      "epoch": 0.40404040404040403,
+      "grad_norm": 0.623359739780426,
+      "learning_rate": 2.992474723665565e-05,
+      "loss": 0.9456,
+      "step": 160
+    },
+    {
+      "epoch": 0.4166666666666667,
+      "grad_norm": 0.6435149312019348,
+      "learning_rate": 2.9911695353408766e-05,
+      "loss": 0.965,
+      "step": 165
+    },
+    {
+      "epoch": 0.4292929292929293,
+      "grad_norm": 0.6187605857849121,
+      "learning_rate": 2.989760358193456e-05,
+      "loss": 0.9599,
+      "step": 170
+    },
+    {
+      "epoch": 0.44191919191919193,
+      "grad_norm": 0.7720528244972229,
+      "learning_rate": 2.9882472904942707e-05,
+      "loss": 0.9487,
+      "step": 175
+    },
+    {
+      "epoch": 0.45454545454545453,
+      "grad_norm": 0.7527753710746765,
+      "learning_rate": 2.9866304377592427e-05,
+      "loss": 0.9445,
+      "step": 180
+    },
+    {
+      "epoch": 0.4671717171717172,
+      "grad_norm": 0.7059020400047302,
+      "learning_rate": 2.9849099127418905e-05,
+      "loss": 0.8939,
+      "step": 185
+    },
+    {
+      "epoch": 0.4797979797979798,
+      "grad_norm": 0.6808345913887024,
+      "learning_rate": 2.9830858354254672e-05,
+      "loss": 0.8791,
+      "step": 190
+    },
+    {
+      "epoch": 0.49242424242424243,
+      "grad_norm": 0.7607003450393677,
+      "learning_rate": 2.9811583330145915e-05,
+      "loss": 0.9394,
+      "step": 195
+    },
+    {
+      "epoch": 0.5050505050505051,
+      "grad_norm": 0.6947716474533081,
+      "learning_rate": 2.9791275399263785e-05,
+      "loss": 0.8957,
+      "step": 200
+    },
+    {
+      "epoch": 0.5176767676767676,
+      "grad_norm": 0.7062451839447021,
+      "learning_rate": 2.976993597781065e-05,
+      "loss": 0.9051,
+      "step": 205
+    },
+    {
+      "epoch": 0.5303030303030303,
+      "grad_norm": 0.7581499814987183,
+      "learning_rate": 2.9747566553921326e-05,
+      "loss": 0.8459,
+      "step": 210
+    },
+    {
+      "epoch": 0.5429292929292929,
+      "grad_norm": 1.0111594200134277,
+      "learning_rate": 2.9724168687559325e-05,
+      "loss": 0.8905,
+      "step": 215
+    },
+    {
+      "epoch": 0.5555555555555556,
+      "grad_norm": 0.915945291519165,
+      "learning_rate": 2.969974401040805e-05,
+      "loss": 0.8879,
+      "step": 220
+    },
+    {
+      "epoch": 0.5681818181818182,
+      "grad_norm": 0.7350049018859863,
+      "learning_rate": 2.9674294225757e-05,
+      "loss": 0.8922,
+      "step": 225
+    },
+    {
+      "epoch": 0.5808080808080808,
+      "grad_norm": 0.7845703363418579,
+      "learning_rate": 2.964782110838301e-05,
+      "loss": 0.8542,
+      "step": 230
+    },
+    {
+      "epoch": 0.5934343434343434,
+      "grad_norm": 0.8170814514160156,
+      "learning_rate": 2.9620326504426476e-05,
+      "loss": 0.8349,
+      "step": 235
+    },
+    {
+      "epoch": 0.6060606060606061,
+      "grad_norm": 0.7793628573417664,
+      "learning_rate": 2.9591812331262598e-05,
+      "loss": 0.8179,
+      "step": 240
+    },
+    {
+      "epoch": 0.6186868686868687,
+      "grad_norm": 0.7981735467910767,
+      "learning_rate": 2.9562280577367694e-05,
+      "loss": 0.8412,
+      "step": 245
+    },
+    {
+      "epoch": 0.6313131313131313,
+      "grad_norm": 0.7543655633926392,
+      "learning_rate": 2.953173330218051e-05,
+      "loss": 0.8018,
+      "step": 250
+    },
+    {
+      "epoch": 0.6439393939393939,
+      "grad_norm": 0.8550294041633606,
+      "learning_rate": 2.950017263595861e-05,
+      "loss": 0.8431,
+      "step": 255
+    },
+    {
+      "epoch": 0.6565656565656566,
+      "grad_norm": 0.8755800724029541,
+      "learning_rate": 2.9467600779629825e-05,
+      "loss": 0.8049,
+      "step": 260
+    },
+    {
+      "epoch": 0.6691919191919192,
+      "grad_norm": 0.8669186234474182,
+      "learning_rate": 2.9434020004638757e-05,
+      "loss": 0.8233,
+      "step": 265
+    },
+    {
+      "epoch": 0.6818181818181818,
+      "grad_norm": 0.7797918915748596,
+      "learning_rate": 2.9399432652788383e-05,
+      "loss": 0.8118,
+      "step": 270
+    },
+    {
+      "epoch": 0.6944444444444444,
+      "grad_norm": 0.9602000117301941,
+      "learning_rate": 2.936384113607674e-05,
+      "loss": 0.809,
+      "step": 275
+    },
+    {
+      "epoch": 0.7070707070707071,
+      "grad_norm": 0.7648482918739319,
+      "learning_rate": 2.9327247936528742e-05,
+      "loss": 0.7763,
+      "step": 280
+    },
+    {
+      "epoch": 0.7196969696969697,
+      "grad_norm": 0.8220114707946777,
+      "learning_rate": 2.9289655606023056e-05,
+      "loss": 0.8279,
+      "step": 285
+    },
+    {
+      "epoch": 0.7323232323232324,
+      "grad_norm": 0.8677099347114563,
+      "learning_rate": 2.925106676611418e-05,
+      "loss": 0.7877,
+      "step": 290
+    },
+    {
+      "epoch": 0.7449494949494949,
+      "grad_norm": 0.912298858165741,
+      "learning_rate": 2.92114841078496e-05,
+      "loss": 0.7746,
+      "step": 295
+    },
+    {
+      "epoch": 0.7575757575757576,
+      "grad_norm": 0.9250658750534058,
+      "learning_rate": 2.917091039158214e-05,
+      "loss": 0.7766,
+      "step": 300
+    },
+    {
+      "epoch": 0.7702020202020202,
+      "grad_norm": 0.9757301807403564,
+      "learning_rate": 2.9129348446777456e-05,
+      "loss": 0.7492,
+      "step": 305
+    },
+    {
+      "epoch": 0.7828282828282829,
+      "grad_norm": 0.9662628769874573,
+      "learning_rate": 2.9086801171816717e-05,
+      "loss": 0.764,
+      "step": 310
+    },
+    {
+      "epoch": 0.7954545454545454,
+      "grad_norm": 0.8361748456954956,
+      "learning_rate": 2.90432715337945e-05,
+      "loss": 0.6907,
+      "step": 315
+    },
+    {
+      "epoch": 0.8080808080808081,
+      "grad_norm": 0.8114555478096008,
+      "learning_rate": 2.8998762568311857e-05,
+      "loss": 0.6749,
+      "step": 320
+    },
+    {
+      "epoch": 0.8207070707070707,
+      "grad_norm": 0.9168850779533386,
+      "learning_rate": 2.8953277379264633e-05,
+      "loss": 0.7378,
+      "step": 325
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 0.9979681968688965,
+      "learning_rate": 2.8906819138627002e-05,
+      "loss": 0.7289,
+      "step": 330
+    },
+    {
+      "epoch": 0.8459595959595959,
+      "grad_norm": 0.9025216698646545,
+      "learning_rate": 2.885939108623028e-05,
+      "loss": 0.7326,
+      "step": 335
+    },
+    {
+      "epoch": 0.8585858585858586,
+      "grad_norm": 0.8616014122962952,
+      "learning_rate": 2.8810996529536967e-05,
+      "loss": 0.6868,
+      "step": 340
+    },
+    {
+      "epoch": 0.8712121212121212,
+      "grad_norm": 0.9671257734298706,
+      "learning_rate": 2.8761638843410133e-05,
+      "loss": 0.7142,
+      "step": 345
+    },
+    {
+      "epoch": 0.8838383838383839,
+      "grad_norm": 1.1255998611450195,
+      "learning_rate": 2.8711321469878024e-05,
+      "loss": 0.7196,
+      "step": 350
+    },
+    {
+      "epoch": 0.8964646464646465,
+      "grad_norm": 0.9464638829231262,
+      "learning_rate": 2.866004791789406e-05,
+      "loss": 0.6761,
+      "step": 355
+    },
+    {
+      "epoch": 0.9090909090909091,
+      "grad_norm": 1.016979455947876,
+      "learning_rate": 2.8607821763092116e-05,
+      "loss": 0.7013,
+      "step": 360
+    },
+    {
+      "epoch": 0.9217171717171717,
+      "grad_norm": 1.0915886163711548,
+      "learning_rate": 2.8554646647537178e-05,
+      "loss": 0.6671,
+      "step": 365
+    },
+    {
+      "epoch": 0.9343434343434344,
+      "grad_norm": 1.0145480632781982,
+      "learning_rate": 2.8500526279471362e-05,
+      "loss": 0.6824,
+      "step": 370
+    },
+    {
+      "epoch": 0.946969696969697,
+      "grad_norm": 0.9449098706245422,
+      "learning_rate": 2.8445464433055303e-05,
+      "loss": 0.7025,
+      "step": 375
+    },
+    {
+      "epoch": 0.9595959595959596,
+      "grad_norm": 1.0181083679199219,
+      "learning_rate": 2.8389464948104962e-05,
+      "loss": 0.6374,
+      "step": 380
+    },
+    {
+      "epoch": 0.9722222222222222,
+      "grad_norm": 1.1749608516693115,
+      "learning_rate": 2.8332531729823853e-05,
+      "loss": 0.6567,
+      "step": 385
+    },
+    {
+      "epoch": 0.9848484848484849,
+      "grad_norm": 1.0283434391021729,
+      "learning_rate": 2.8274668748530718e-05,
+      "loss": 0.7205,
+      "step": 390
+    },
+    {
+      "epoch": 0.9974747474747475,
+      "grad_norm": 0.9649919867515564,
+      "learning_rate": 2.8215880039382625e-05,
+      "loss": 0.671,
+      "step": 395
+    },
+    {
+      "epoch": 1.0101010101010102,
+      "grad_norm": 0.9317898154258728,
+      "learning_rate": 2.8156169702093608e-05,
+      "loss": 0.592,
+      "step": 400
+    },
+    {
+      "epoch": 1.0227272727272727,
+      "grad_norm": 1.047213077545166,
+      "learning_rate": 2.809554190064873e-05,
+      "loss": 0.5692,
+      "step": 405
+    },
+    {
+      "epoch": 1.0353535353535352,
+      "grad_norm": 0.9808226823806763,
+      "learning_rate": 2.803400086301372e-05,
+      "loss": 0.549,
+      "step": 410
+    },
+    {
+      "epoch": 1.047979797979798,
+      "grad_norm": 1.1340376138687134,
+      "learning_rate": 2.7971550880840138e-05,
+      "loss": 0.5457,
+      "step": 415
+    },
+    {
+      "epoch": 1.0606060606060606,
+      "grad_norm": 0.9728055000305176,
+      "learning_rate": 2.7908196309166074e-05,
+      "loss": 0.5686,
+      "step": 420
+    },
+    {
+      "epoch": 1.0732323232323233,
+      "grad_norm": 0.9779664278030396,
+      "learning_rate": 2.7843941566112442e-05,
+      "loss": 0.5344,
+      "step": 425
+    },
+    {
+      "epoch": 1.0858585858585859,
+      "grad_norm": 0.9450783133506775,
+      "learning_rate": 2.7778791132574908e-05,
+      "loss": 0.5322,
+      "step": 430
+    },
+    {
+      "epoch": 1.0984848484848484,
+      "grad_norm": 0.994297981262207,
+      "learning_rate": 2.7712749551911355e-05,
+      "loss": 0.5255,
+      "step": 435
+    },
+    {
+      "epoch": 1.1111111111111112,
+      "grad_norm": 0.9942148327827454,
+      "learning_rate": 2.7645821429625092e-05,
+      "loss": 0.5137,
+      "step": 440
+    },
+    {
+      "epoch": 1.1237373737373737,
+      "grad_norm": 1.0680190324783325,
+      "learning_rate": 2.757801143304367e-05,
+      "loss": 0.5647,
+      "step": 445
+    },
+    {
+      "epoch": 1.1363636363636362,
+      "grad_norm": 1.0500057935714722,
+      "learning_rate": 2.750932429099338e-05,
+      "loss": 0.5349,
+      "step": 450
+    },
+    {
+      "epoch": 1.148989898989899,
+      "grad_norm": 1.1989120244979858,
+      "learning_rate": 2.7439764793469504e-05,
+      "loss": 0.5376,
+      "step": 455
+    },
+    {
+      "epoch": 1.1616161616161615,
+      "grad_norm": 1.0777108669281006,
+      "learning_rate": 2.7369337791302272e-05,
+      "loss": 0.5937,
+      "step": 460
+    },
+    {
+      "epoch": 1.1742424242424243,
+      "grad_norm": 1.0498294830322266,
+      "learning_rate": 2.729804819581858e-05,
+      "loss": 0.5608,
+      "step": 465
+    },
+    {
+      "epoch": 1.1868686868686869,
+      "grad_norm": 0.9939706325531006,
+      "learning_rate": 2.7225900978499487e-05,
+      "loss": 0.4959,
+      "step": 470
+    },
+    {
+      "epoch": 1.1994949494949494,
+      "grad_norm": 1.0636131763458252,
+      "learning_rate": 2.715290117063354e-05,
+      "loss": 0.5269,
+      "step": 475
+    },
+    {
+      "epoch": 1.2121212121212122,
+      "grad_norm": 1.0121461153030396,
+      "learning_rate": 2.707905386296588e-05,
+      "loss": 0.4775,
+      "step": 480
+    },
+    {
+      "epoch": 1.2247474747474747,
+      "grad_norm": 1.0765471458435059,
+      "learning_rate": 2.700436420534326e-05,
+      "loss": 0.4939,
+      "step": 485
+    },
+    {
+      "epoch": 1.2373737373737375,
+      "grad_norm": 1.1116187572479248,
+      "learning_rate": 2.6928837406354905e-05,
+      "loss": 0.5349,
+      "step": 490
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.9618929624557495,
+      "learning_rate": 2.6852478732969272e-05,
+      "loss": 0.5085,
+      "step": 495
+    },
+    {
+      "epoch": 1.2626262626262625,
+      "grad_norm": 1.0578511953353882,
+      "learning_rate": 2.677529351016676e-05,
+      "loss": 0.4741,
+      "step": 500
+    },
+    {
+      "epoch": 1.2752525252525253,
+      "grad_norm": 0.9444258809089661,
+      "learning_rate": 2.6697287120568364e-05,
+      "loss": 0.5153,
+      "step": 505
+    },
+    {
+      "epoch": 1.2878787878787878,
+      "grad_norm": 1.1680361032485962,
+      "learning_rate": 2.6618465004060324e-05,
+      "loss": 0.5033,
+      "step": 510
+    },
+    {
+      "epoch": 1.3005050505050506,
+      "grad_norm": 1.0985764265060425,
+      "learning_rate": 2.653883265741473e-05,
+      "loss": 0.4917,
+      "step": 515
+    },
+    {
+      "epoch": 1.3131313131313131,
+      "grad_norm": 1.1677299737930298,
+      "learning_rate": 2.645839563390624e-05,
+      "loss": 0.5526,
+      "step": 520
+    },
+    {
+      "epoch": 1.3257575757575757,
+      "grad_norm": 1.122022271156311,
+      "learning_rate": 2.637715954292478e-05,
+      "loss": 0.5197,
+      "step": 525
+    },
+    {
+      "epoch": 1.3383838383838385,
+      "grad_norm": 1.02968430519104,
+      "learning_rate": 2.6295130049584388e-05,
+      "loss": 0.4619,
+      "step": 530
+    },
+    {
+      "epoch": 1.351010101010101,
+      "grad_norm": 0.9885948300361633,
+      "learning_rate": 2.6212312874328136e-05,
+      "loss": 0.4599,
+      "step": 535
+    },
+    {
+      "epoch": 1.3636363636363638,
+      "grad_norm": 1.1250592470169067,
+      "learning_rate": 2.6128713792529225e-05,
+      "loss": 0.4805,
+      "step": 540
+    },
+    {
+      "epoch": 1.3762626262626263,
+      "grad_norm": 1.128479242324829,
+      "learning_rate": 2.6044338634088198e-05,
+      "loss": 0.4858,
+      "step": 545
+    },
+    {
+      "epoch": 1.3888888888888888,
+      "grad_norm": 1.0660187005996704,
+      "learning_rate": 2.595919328302641e-05,
+      "loss": 0.4568,
+      "step": 550
+    },
+    {
+      "epoch": 1.4015151515151514,
+      "grad_norm": 1.115667700767517,
+      "learning_rate": 2.5873283677075704e-05,
+      "loss": 0.4843,
+      "step": 555
+    },
+    {
+      "epoch": 1.4141414141414141,
+      "grad_norm": 1.0488578081130981,
+      "learning_rate": 2.5786615807264306e-05,
+      "loss": 0.4969,
+      "step": 560
+    },
+    {
+      "epoch": 1.4267676767676767,
+      "grad_norm": 0.9793870449066162,
+      "learning_rate": 2.569919571749905e-05,
+      "loss": 0.486,
+      "step": 565
+    },
+    {
+      "epoch": 1.4393939393939394,
+      "grad_norm": 1.2411023378372192,
+      "learning_rate": 2.5611029504143905e-05,
+      "loss": 0.415,
+      "step": 570
+    },
+    {
+      "epoch": 1.452020202020202,
+      "grad_norm": 1.0087097883224487,
+      "learning_rate": 2.552212331559482e-05,
+      "loss": 0.4396,
+      "step": 575
+    },
+    {
+      "epoch": 1.4646464646464645,
+      "grad_norm": 1.0943728685379028,
+      "learning_rate": 2.5432483351850962e-05,
+      "loss": 0.4658,
+      "step": 580
+    },
+    {
+      "epoch": 1.4772727272727273,
+      "grad_norm": 1.065333604812622,
+      "learning_rate": 2.5342115864082355e-05,
+      "loss": 0.4314,
+      "step": 585
+    },
+    {
+      "epoch": 1.4898989898989898,
+      "grad_norm": 1.1207003593444824,
+      "learning_rate": 2.5251027154193945e-05,
+      "loss": 0.4763,
+      "step": 590
+    },
+    {
+      "epoch": 1.5025252525252526,
+      "grad_norm": 1.0626108646392822,
+      "learning_rate": 2.5159223574386117e-05,
+      "loss": 0.408,
+      "step": 595
+    },
+    {
+      "epoch": 1.5151515151515151,
+      "grad_norm": 1.0062793493270874,
+      "learning_rate": 2.5066711526711736e-05,
+      "loss": 0.4337,
+      "step": 600
+    },
+    {
+      "epoch": 1.5277777777777777,
+      "grad_norm": 1.0512022972106934,
+      "learning_rate": 2.497349746262967e-05,
+      "loss": 0.4004,
+      "step": 605
+    },
+    {
+      "epoch": 1.5404040404040404,
+      "grad_norm": 1.0447345972061157,
+      "learning_rate": 2.4879587882554903e-05,
+      "loss": 0.4434,
+      "step": 610
+    },
+    {
+      "epoch": 1.553030303030303,
+      "grad_norm": 1.0264374017715454,
+      "learning_rate": 2.4784989335405203e-05,
+      "loss": 0.4561,
+      "step": 615
+    },
+    {
+      "epoch": 1.5656565656565657,
+      "grad_norm": 1.2251687049865723,
+      "learning_rate": 2.468970841814445e-05,
+      "loss": 0.4333,
+      "step": 620
+    },
+    {
+      "epoch": 1.5782828282828283,
+      "grad_norm": 1.082527995109558,
+      "learning_rate": 2.4593751775322553e-05,
+      "loss": 0.4615,
+      "step": 625
+    },
+    {
+      "epoch": 1.5909090909090908,
+      "grad_norm": 1.155479073524475,
+      "learning_rate": 2.4497126098612115e-05,
+      "loss": 0.4304,
+      "step": 630
+    },
+    {
+      "epoch": 1.6035353535353534,
+      "grad_norm": 1.0841268301010132,
+      "learning_rate": 2.4399838126341768e-05,
+      "loss": 0.4345,
+      "step": 635
+    },
+    {
+      "epoch": 1.6161616161616161,
+      "grad_norm": 1.0636403560638428,
+      "learning_rate": 2.430189464302625e-05,
+      "loss": 0.3878,
+      "step": 640
+    },
+    {
+      "epoch": 1.628787878787879,
+      "grad_norm": 1.0887062549591064,
+      "learning_rate": 2.4203302478893307e-05,
+      "loss": 0.4154,
+      "step": 645
+    },
+    {
+      "epoch": 1.6414141414141414,
+      "grad_norm": 1.071465015411377,
+      "learning_rate": 2.410406850940735e-05,
+      "loss": 0.4158,
+      "step": 650
+    },
+    {
+      "epoch": 1.654040404040404,
+      "grad_norm": 1.0511163473129272,
+      "learning_rate": 2.4004199654790008e-05,
+      "loss": 0.3878,
+      "step": 655
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 1.1062862873077393,
+      "learning_rate": 2.3903702879537513e-05,
+      "loss": 0.4166,
+      "step": 660
+    },
+    {
+      "epoch": 1.6792929292929293,
+      "grad_norm": 1.101647138595581,
+      "learning_rate": 2.3802585191935044e-05,
+      "loss": 0.3954,
+      "step": 665
+    },
+    {
+      "epoch": 1.691919191919192,
+      "grad_norm": 1.1077150106430054,
+      "learning_rate": 2.3700853643567973e-05,
+      "loss": 0.413,
+      "step": 670
+    },
+    {
+      "epoch": 1.7045454545454546,
+      "grad_norm": 1.1404215097427368,
+      "learning_rate": 2.3598515328830135e-05,
+      "loss": 0.4299,
+      "step": 675
+    },
+    {
+      "epoch": 1.7171717171717171,
+      "grad_norm": 0.9413719773292542,
+      "learning_rate": 2.349557738442907e-05,
+      "loss": 0.361,
+      "step": 680
+    },
+    {
+      "epoch": 1.7297979797979797,
+      "grad_norm": 1.1357874870300293,
+      "learning_rate": 2.3392046988888345e-05,
+      "loss": 0.3782,
+      "step": 685
+    },
+    {
+      "epoch": 1.7424242424242424,
+      "grad_norm": 1.0469523668289185,
+      "learning_rate": 2.328793136204695e-05,
+      "loss": 0.3949,
+      "step": 690
+    },
+    {
+      "epoch": 1.7550505050505052,
+      "grad_norm": 1.0579326152801514,
+      "learning_rate": 2.3183237764555807e-05,
+      "loss": 0.4216,
+      "step": 695
+    },
+    {
+      "epoch": 1.7676767676767677,
+      "grad_norm": 1.1634066104888916,
+      "learning_rate": 2.307797349737144e-05,
+      "loss": 0.4091,
+      "step": 700
+    },
+    {
+      "epoch": 1.7803030303030303,
+      "grad_norm": 1.1639355421066284,
+      "learning_rate": 2.297214590124684e-05,
+      "loss": 0.4212,
+      "step": 705
+    },
+    {
+      "epoch": 1.7929292929292928,
+      "grad_norm": 0.9826981425285339,
+      "learning_rate": 2.2865762356219533e-05,
+      "loss": 0.3518,
+      "step": 710
+    },
+    {
+      "epoch": 1.8055555555555556,
+      "grad_norm": 1.0579651594161987,
+      "learning_rate": 2.2758830281096935e-05,
+      "loss": 0.3899,
+      "step": 715
+    },
+    {
+      "epoch": 1.8181818181818183,
+      "grad_norm": 1.061562418937683,
+      "learning_rate": 2.265135713293899e-05,
+      "loss": 0.3874,
+      "step": 720
+    },
+    {
+      "epoch": 1.8308080808080809,
+      "grad_norm": 1.1707763671875,
+      "learning_rate": 2.254335040653812e-05,
+      "loss": 0.376,
+      "step": 725
+    },
+    {
+      "epoch": 1.8434343434343434,
+      "grad_norm": 1.1716967821121216,
+      "learning_rate": 2.243481763389661e-05,
+      "loss": 0.3391,
+      "step": 730
+    },
+    {
+      "epoch": 1.856060606060606,
+      "grad_norm": 1.1718958616256714,
+      "learning_rate": 2.23257663837013e-05,
+      "loss": 0.361,
+      "step": 735
+    },
+    {
+      "epoch": 1.8686868686868687,
+      "grad_norm": 1.1531727313995361,
+      "learning_rate": 2.2216204260795813e-05,
+      "loss": 0.3678,
+      "step": 740
+    },
+    {
+      "epoch": 1.8813131313131313,
+      "grad_norm": 1.0929404497146606,
+      "learning_rate": 2.21061389056502e-05,
+      "loss": 0.3785,
+      "step": 745
+    },
+    {
+      "epoch": 1.893939393939394,
+      "grad_norm": 1.059478521347046,
+      "learning_rate": 2.199557799382813e-05,
+      "loss": 0.3319,
+      "step": 750
+    },
+    {
+      "epoch": 1.9065656565656566,
+      "grad_norm": 1.037405014038086,
+      "learning_rate": 2.1884529235451618e-05,
+      "loss": 0.3493,
+      "step": 755
+    },
+    {
+      "epoch": 1.9191919191919191,
+      "grad_norm": 1.3604897260665894,
+      "learning_rate": 2.177300037466334e-05,
+      "loss": 0.3749,
+      "step": 760
+    },
+    {
+      "epoch": 1.9318181818181817,
+      "grad_norm": 0.9598345160484314,
+      "learning_rate": 2.1660999189086613e-05,
+      "loss": 0.3459,
+      "step": 765
+    },
+    {
+      "epoch": 1.9444444444444444,
+      "grad_norm": 1.1897592544555664,
+      "learning_rate": 2.1548533489282977e-05,
+      "loss": 0.361,
+      "step": 770
+    },
+    {
+      "epoch": 1.9570707070707072,
+      "grad_norm": 1.1726038455963135,
+      "learning_rate": 2.1435611118207546e-05,
+      "loss": 0.354,
+      "step": 775
+    },
+    {
+      "epoch": 1.9696969696969697,
+      "grad_norm": 1.1047810316085815,
+      "learning_rate": 2.1322239950662035e-05,
+      "loss": 0.3034,
+      "step": 780
+    },
+    {
+      "epoch": 1.9823232323232323,
+      "grad_norm": 1.1414453983306885,
+      "learning_rate": 2.120842789274563e-05,
+      "loss": 0.3174,
+      "step": 785
+    },
+    {
+      "epoch": 1.9949494949494948,
+      "grad_norm": 1.0688472986221313,
+      "learning_rate": 2.1094182881303636e-05,
+      "loss": 0.3343,
+      "step": 790
+    },
+    {
+      "epoch": 2.007575757575758,
+      "grad_norm": 1.1202057600021362,
+      "learning_rate": 2.0979512883373972e-05,
+      "loss": 0.2793,
+      "step": 795
+    },
+    {
+      "epoch": 2.0202020202020203,
+      "grad_norm": 1.4000380039215088,
+      "learning_rate": 2.08644258956316e-05,
+      "loss": 0.2709,
+      "step": 800
+    },
+    {
+      "epoch": 2.032828282828283,
+      "grad_norm": 0.9448375105857849,
+      "learning_rate": 2.0748929943830863e-05,
+      "loss": 0.2726,
+      "step": 805
+    },
+    {
+      "epoch": 2.0454545454545454,
+      "grad_norm": 1.2739609479904175,
+      "learning_rate": 2.0633033082245782e-05,
+      "loss": 0.2872,
+      "step": 810
+    },
+    {
+      "epoch": 2.058080808080808,
+      "grad_norm": 1.1996492147445679,
+      "learning_rate": 2.05167433931084e-05,
+      "loss": 0.311,
+      "step": 815
+    },
+    {
+      "epoch": 2.0707070707070705,
+      "grad_norm": 1.1399296522140503,
+      "learning_rate": 2.0400068986045142e-05,
+      "loss": 0.2853,
+      "step": 820
+    },
+    {
+      "epoch": 2.0833333333333335,
+      "grad_norm": 1.0565412044525146,
+      "learning_rate": 2.0283017997511283e-05,
+      "loss": 0.2629,
+      "step": 825
+    },
+    {
+      "epoch": 2.095959595959596,
+      "grad_norm": 1.0858699083328247,
+      "learning_rate": 2.016559859022355e-05,
+      "loss": 0.2907,
+      "step": 830
+    },
+    {
+      "epoch": 2.1085858585858586,
+      "grad_norm": 1.0091416835784912,
+      "learning_rate": 2.0047818952590854e-05,
+      "loss": 0.2576,
+      "step": 835
+    },
+    {
+      "epoch": 2.121212121212121,
+      "grad_norm": 1.1173107624053955,
+      "learning_rate": 1.99296872981433e-05,
+      "loss": 0.2398,
+      "step": 840
+    },
+    {
+      "epoch": 2.1338383838383836,
+      "grad_norm": 1.1509712934494019,
+      "learning_rate": 1.9811211864959374e-05,
+      "loss": 0.2555,
+      "step": 845
+    },
+    {
+      "epoch": 2.1464646464646466,
+      "grad_norm": 1.1491074562072754,
+      "learning_rate": 1.969240091509147e-05,
+      "loss": 0.2791,
+      "step": 850
+    },
+    {
+      "epoch": 2.159090909090909,
+      "grad_norm": 1.2488752603530884,
+      "learning_rate": 1.95732627339897e-05,
+      "loss": 0.236,
+      "step": 855
+    },
+    {
+      "epoch": 2.1717171717171717,
+      "grad_norm": 1.0945607423782349,
+      "learning_rate": 1.9453805629924126e-05,
+      "loss": 0.2854,
+      "step": 860
+    },
+    {
+      "epoch": 2.1843434343434343,
+      "grad_norm": 1.2910445928573608,
+      "learning_rate": 1.9334037933405337e-05,
+      "loss": 0.2543,
+      "step": 865
+    },
+    {
+      "epoch": 2.196969696969697,
+      "grad_norm": 1.092241883277893,
+      "learning_rate": 1.9213967996603542e-05,
+      "loss": 0.2535,
+      "step": 870
+    },
+    {
+      "epoch": 2.20959595959596,
+      "grad_norm": 1.2114317417144775,
+      "learning_rate": 1.9093604192766102e-05,
+      "loss": 0.2615,
+      "step": 875
+    },
+    {
+      "epoch": 2.2222222222222223,
+      "grad_norm": 0.9923803806304932,
+      "learning_rate": 1.8972954915633604e-05,
+      "loss": 0.2231,
+      "step": 880
+    },
+    {
+      "epoch": 2.234848484848485,
+      "grad_norm": 1.0003025531768799,
+      "learning_rate": 1.8852028578854532e-05,
+      "loss": 0.2441,
+      "step": 885
+    },
+    {
+      "epoch": 2.2474747474747474,
+      "grad_norm": 1.0413899421691895,
+      "learning_rate": 1.873083361539851e-05,
+      "loss": 0.2414,
+      "step": 890
+    },
+    {
+      "epoch": 2.26010101010101,
+      "grad_norm": 1.1812835931777954,
+      "learning_rate": 1.8609378476968232e-05,
+      "loss": 0.2434,
+      "step": 895
+    },
+    {
+      "epoch": 2.2727272727272725,
+      "grad_norm": 1.0742385387420654,
+      "learning_rate": 1.8487671633410052e-05,
+      "loss": 0.2539,
+      "step": 900
+    },
+    {
+      "epoch": 2.2853535353535355,
+      "grad_norm": 1.2803466320037842,
+      "learning_rate": 1.836572157212334e-05,
+      "loss": 0.2478,
+      "step": 905
+    },
+    {
+      "epoch": 2.297979797979798,
+      "grad_norm": 1.0669132471084595,
+      "learning_rate": 1.824353679746861e-05,
+      "loss": 0.2271,
+      "step": 910
+    },
+    {
+      "epoch": 2.3106060606060606,
+      "grad_norm": 1.1045911312103271,
+      "learning_rate": 1.8121125830174437e-05,
+      "loss": 0.2572,
+      "step": 915
+    },
+    {
+      "epoch": 2.323232323232323,
+      "grad_norm": 1.1256022453308105,
+      "learning_rate": 1.799849720674326e-05,
+      "loss": 0.2582,
+      "step": 920
+    },
+    {
+      "epoch": 2.3358585858585856,
+      "grad_norm": 1.091371774673462,
+      "learning_rate": 1.7875659478856077e-05,
+      "loss": 0.2495,
+      "step": 925
+    },
+    {
+      "epoch": 2.3484848484848486,
+      "grad_norm": 1.0141551494598389,
+      "learning_rate": 1.775262121277609e-05,
+      "loss": 0.2513,
+      "step": 930
+    },
+    {
+      "epoch": 2.361111111111111,
+      "grad_norm": 1.1098726987838745,
+      "learning_rate": 1.7629390988751307e-05,
+      "loss": 0.2375,
+      "step": 935
+    },
+    {
+      "epoch": 2.3737373737373737,
+      "grad_norm": 1.0376847982406616,
+      "learning_rate": 1.7505977400416207e-05,
+      "loss": 0.2161,
+      "step": 940
+    },
+    {
+      "epoch": 2.3863636363636362,
+      "grad_norm": 1.1401861906051636,
+      "learning_rate": 1.738238905419242e-05,
+      "loss": 0.2434,
+      "step": 945
+    },
+    {
+      "epoch": 2.398989898989899,
+      "grad_norm": 1.2411974668502808,
+      "learning_rate": 1.7258634568688577e-05,
+      "loss": 0.245,
+      "step": 950
+    },
+    {
+      "epoch": 2.4116161616161618,
+      "grad_norm": 1.1614488363265991,
+      "learning_rate": 1.713472257409928e-05,
+      "loss": 0.216,
+      "step": 955
+    },
+    {
+      "epoch": 2.4242424242424243,
+      "grad_norm": 1.2484320402145386,
+      "learning_rate": 1.7010661711603224e-05,
+      "loss": 0.218,
+      "step": 960
+    },
+    {
+      "epoch": 2.436868686868687,
+      "grad_norm": 1.0049769878387451,
+      "learning_rate": 1.688646063276064e-05,
+      "loss": 0.1978,
+      "step": 965
+    },
+    {
+      "epoch": 2.4494949494949494,
+      "grad_norm": 1.1801090240478516,
+      "learning_rate": 1.6762127998909933e-05,
+      "loss": 0.2106,
+      "step": 970
+    },
+    {
+      "epoch": 2.462121212121212,
+      "grad_norm": 1.0992910861968994,
+      "learning_rate": 1.6637672480563694e-05,
+      "loss": 0.2194,
+      "step": 975
+    },
+    {
+      "epoch": 2.474747474747475,
+      "grad_norm": 1.0028361082077026,
+      "learning_rate": 1.6513102756804025e-05,
+      "loss": 0.2475,
+      "step": 980
+    },
+    {
+      "epoch": 2.4873737373737375,
+      "grad_norm": 1.0803316831588745,
+      "learning_rate": 1.6388427514677315e-05,
+      "loss": 0.2189,
+      "step": 985
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.0498230457305908,
+      "learning_rate": 1.6263655448588417e-05,
+      "loss": 0.2219,
+      "step": 990
+    },
+    {
+      "epoch": 2.5126262626262625,
+      "grad_norm": 1.12675142288208,
+      "learning_rate": 1.613879525969435e-05,
+      "loss": 0.2049,
+      "step": 995
+    },
+    {
+      "epoch": 2.525252525252525,
+      "grad_norm": 1.0779552459716797,
+      "learning_rate": 1.6013855655297498e-05,
+      "loss": 0.2169,
+      "step": 1000
+    },
+    {
+      "epoch": 2.537878787878788,
+      "grad_norm": 1.267693281173706,
+      "learning_rate": 1.5888845348238388e-05,
+      "loss": 0.1932,
+      "step": 1005
+    },
+    {
+      "epoch": 2.5505050505050506,
+      "grad_norm": 1.1487102508544922,
+      "learning_rate": 1.5763773056288127e-05,
+      "loss": 0.1986,
+      "step": 1010
+    },
+    {
+      "epoch": 2.563131313131313,
+      "grad_norm": 1.0239760875701904,
+      "learning_rate": 1.5638647501540387e-05,
+      "loss": 0.2137,
+      "step": 1015
+    },
+    {
+      "epoch": 2.5757575757575757,
+      "grad_norm": 1.0446583032608032,
+      "learning_rate": 1.551347740980322e-05,
+      "loss": 0.2193,
+      "step": 1020
+    },
+    {
+      "epoch": 2.5883838383838382,
+      "grad_norm": 1.0630428791046143,
+      "learning_rate": 1.5388271509990532e-05,
+      "loss": 0.2113,
+      "step": 1025
+    },
+    {
+      "epoch": 2.601010101010101,
+      "grad_norm": 1.213524341583252,
+      "learning_rate": 1.5263038533513338e-05,
+      "loss": 0.2096,
+      "step": 1030
+    },
+    {
+      "epoch": 2.6136363636363638,
+      "grad_norm": 1.0546088218688965,
+      "learning_rate": 1.5137787213670899e-05,
+      "loss": 0.2017,
+      "step": 1035
+    },
+    {
+      "epoch": 2.6262626262626263,
+      "grad_norm": 1.0828937292099,
+      "learning_rate": 1.5012526285041662e-05,
+      "loss": 0.1966,
+      "step": 1040
+    },
+    {
+      "epoch": 2.638888888888889,
+      "grad_norm": 1.0778100490570068,
+      "learning_rate": 1.4887264482874173e-05,
+      "loss": 0.2025,
+      "step": 1045
+    },
+    {
+      "epoch": 2.6515151515151514,
+      "grad_norm": 1.1268481016159058,
+      "learning_rate": 1.4762010542477881e-05,
+      "loss": 0.2176,
+      "step": 1050
+    },
+    {
+      "epoch": 2.6641414141414144,
+      "grad_norm": 1.0860501527786255,
+      "learning_rate": 1.4636773198613994e-05,
+      "loss": 0.1815,
+      "step": 1055
+    },
+    {
+      "epoch": 2.676767676767677,
+      "grad_norm": 1.1054151058197021,
+      "learning_rate": 1.451156118488633e-05,
+      "loss": 0.2091,
+      "step": 1060
+    },
+    {
+      "epoch": 2.6893939393939394,
+      "grad_norm": 1.1649571657180786,
+      "learning_rate": 1.438638323313227e-05,
+      "loss": 0.2269,
+      "step": 1065
+    },
+    {
+      "epoch": 2.702020202020202,
+      "grad_norm": 1.3010584115982056,
+      "learning_rate": 1.4261248072813852e-05,
+      "loss": 0.1895,
+      "step": 1070
+    },
+    {
+      "epoch": 2.7146464646464645,
+      "grad_norm": 1.0134211778640747,
+      "learning_rate": 1.4136164430408979e-05,
+      "loss": 0.2275,
+      "step": 1075
+    },
+    {
+      "epoch": 2.7272727272727275,
+      "grad_norm": 1.1953684091567993,
+      "learning_rate": 1.4011141028802873e-05,
+      "loss": 0.1942,
+      "step": 1080
+    },
+    {
+      "epoch": 2.73989898989899,
+      "grad_norm": 1.1163067817687988,
+      "learning_rate": 1.3886186586679798e-05,
+      "loss": 0.1962,
+      "step": 1085
+    },
+    {
+      "epoch": 2.7525252525252526,
+      "grad_norm": 1.0360524654388428,
+      "learning_rate": 1.3761309817915017e-05,
+      "loss": 0.1854,
+      "step": 1090
+    },
+    {
+      "epoch": 2.765151515151515,
+      "grad_norm": 1.16696035861969,
+      "learning_rate": 1.3636519430967129e-05,
+      "loss": 0.1711,
+      "step": 1095
+    },
+    {
+      "epoch": 2.7777777777777777,
+      "grad_norm": 1.136378526687622,
+      "learning_rate": 1.351182412827079e-05,
+      "loss": 0.1741,
+      "step": 1100
+    },
+    {
+      "epoch": 2.7904040404040407,
+      "grad_norm": 1.100380539894104,
+      "learning_rate": 1.3387232605629804e-05,
+      "loss": 0.1977,
+      "step": 1105
+    },
+    {
+      "epoch": 2.8030303030303028,
+      "grad_norm": 1.1871534585952759,
+      "learning_rate": 1.326275355161073e-05,
+      "loss": 0.1906,
+      "step": 1110
+    },
+    {
+      "epoch": 2.8156565656565657,
+      "grad_norm": 1.1915061473846436,
+      "learning_rate": 1.3138395646936974e-05,
+      "loss": 0.1835,
+      "step": 1115
+    },
+    {
+      "epoch": 2.8282828282828283,
+      "grad_norm": 1.1094458103179932,
+      "learning_rate": 1.301416756388342e-05,
+      "loss": 0.1901,
+      "step": 1120
+    },
+    {
+      "epoch": 2.840909090909091,
+      "grad_norm": 1.1110031604766846,
+      "learning_rate": 1.289007796567165e-05,
+      "loss": 0.1731,
+      "step": 1125
+    },
+    {
+      "epoch": 2.8535353535353534,
+      "grad_norm": 1.219718098640442,
+      "learning_rate": 1.2766135505865808e-05,
+      "loss": 0.1704,
+      "step": 1130
+    },
+    {
+      "epoch": 2.866161616161616,
+      "grad_norm": 1.1409133672714233,
+      "learning_rate": 1.2642348827769152e-05,
+      "loss": 0.1875,
+      "step": 1135
+    },
+    {
+      "epoch": 2.878787878787879,
+      "grad_norm": 1.1201096773147583,
+      "learning_rate": 1.2518726563821253e-05,
+      "loss": 0.1799,
+      "step": 1140
+    },
+    {
+      "epoch": 2.8914141414141414,
+      "grad_norm": 1.0194494724273682,
+      "learning_rate": 1.2395277334996045e-05,
+      "loss": 0.1806,
+      "step": 1145
+    },
+    {
+      "epoch": 2.904040404040404,
+      "grad_norm": 1.1575123071670532,
+      "learning_rate": 1.2272009750200618e-05,
+      "loss": 0.1957,
+      "step": 1150
+    },
+    {
+      "epoch": 2.9166666666666665,
+      "grad_norm": 1.1770744323730469,
+      "learning_rate": 1.2148932405674843e-05,
+      "loss": 0.1747,
+      "step": 1155
+    },
+    {
+      "epoch": 2.929292929292929,
+      "grad_norm": 1.186741590499878,
+      "learning_rate": 1.2026053884391918e-05,
+      "loss": 0.1663,
+      "step": 1160
+    },
+    {
+      "epoch": 2.941919191919192,
+      "grad_norm": 1.1966912746429443,
+      "learning_rate": 1.1903382755459839e-05,
+      "loss": 0.1682,
+      "step": 1165
+    },
+    {
+      "epoch": 2.9545454545454546,
+      "grad_norm": 1.1320679187774658,
+      "learning_rate": 1.1780927573523776e-05,
+      "loss": 0.1939,
+      "step": 1170
+    },
+    {
+      "epoch": 2.967171717171717,
+      "grad_norm": 1.068954586982727,
+      "learning_rate": 1.1658696878169541e-05,
+      "loss": 0.1622,
+      "step": 1175
+    },
+    {
+      "epoch": 2.9797979797979797,
+      "grad_norm": 1.070233702659607,
+      "learning_rate": 1.1536699193328063e-05,
+      "loss": 0.1833,
+      "step": 1180
+    },
+    {
+      "epoch": 2.992424242424242,
+      "grad_norm": 1.156294345855713,
+      "learning_rate": 1.1414943026680939e-05,
+      "loss": 0.1666,
+      "step": 1185
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1980,
+  "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": 1.6863338603688755e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

31_128_e5_3e-5/checkpoint-1188/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1f7003266cf4b03470fdb9a31d0f745932e977fccbde9f17b310070d8bf4df35
+size 7736

31_128_e5_3e-5/checkpoint-1188/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)

31_128_e5_3e-5/checkpoint-1584/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

31_128_e5_3e-5/checkpoint-1584/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": [
+    "k_proj",
+    "q_proj",
+    "up_proj",
+    "o_proj",
+    "gate_proj",
+    "v_proj",
+    "down_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

31_128_e5_3e-5/checkpoint-1584/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:35edc6e9513146caade56e7f5cecc7f3acacb7a26e417eb4e62e5f33855e7728
+size 791751704

31_128_e5_3e-5/checkpoint-1584/latest

@@ -0,0 +1 @@
+global_step1584

31_128_e5_3e-5/checkpoint-1584/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:63edbfad5eb7544cf69f8721324b9fb2efeac4787fcaf5cd4cb1b9b73c3e0a90
+size 15920

31_128_e5_3e-5/checkpoint-1584/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:127c90a3353676d5bbc11b752a7a59014da8e8cf5e7f63bd45818094df1d503a
+size 15920

31_128_e5_3e-5/checkpoint-1584/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:53410d1f38dfb9ec82c17fe6045c06cde6654287836d4f4a1c2a4a1ebbb4f990
+size 15920

31_128_e5_3e-5/checkpoint-1584/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4e61a1f2206e9056a119924a8ea460631b34c5b49d6aed19761147b2c6129bc0
+size 15920

31_128_e5_3e-5/checkpoint-1584/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:647062990e876a45ac3df7a381303210c104ed30dcd4606405d2c3f2d519e3d5
+size 15920

31_128_e5_3e-5/checkpoint-1584/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:70491554e335e716f534bb1c70092cf9872e4a47ddae6011588c07f784d5e9bb
+size 15920

31_128_e5_3e-5/checkpoint-1584/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:68df504f485052af830ece1abbd2b3e8c58c0f957bd52097e344434743b394cb
+size 15920

31_128_e5_3e-5/checkpoint-1584/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aac1a7600318cdecfd72c740be339daadb10eae62f9882cf4f3bf439ee5f4eb3
+size 15920

31_128_e5_3e-5/checkpoint-1584/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:686ac34d593f7083664c7adca69627690fac1f1a96a72cfd6ed6935f552813ea
+size 1064

31_128_e5_3e-5/checkpoint-1584/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
+  }
+}

31_128_e5_3e-5/checkpoint-1584/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
+}

31_128_e5_3e-5/checkpoint-1584/trainer_state.json

@@ -0,0 +1,2246 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 1584,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.012626262626262626,
+      "grad_norm": 1.2214584350585938,
+      "learning_rate": 1.2121212121212122e-06,
+      "loss": 1.3078,
+      "step": 5
+    },
+    {
+      "epoch": 0.025252525252525252,
+      "grad_norm": 0.9447234272956848,
+      "learning_rate": 2.7272727272727272e-06,
+      "loss": 1.3299,
+      "step": 10
+    },
+    {
+      "epoch": 0.03787878787878788,
+      "grad_norm": 0.6590636968612671,
+      "learning_rate": 4.242424242424242e-06,
+      "loss": 1.2777,
+      "step": 15
+    },
+    {
+      "epoch": 0.050505050505050504,
+      "grad_norm": 0.6515881419181824,
+      "learning_rate": 5.757575757575758e-06,
+      "loss": 1.2685,
+      "step": 20
+    },
+    {
+      "epoch": 0.06313131313131314,
+      "grad_norm": 0.5434719324111938,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 1.2521,
+      "step": 25
+    },
+    {
+      "epoch": 0.07575757575757576,
+      "grad_norm": 0.5539897084236145,
+      "learning_rate": 8.787878787878788e-06,
+      "loss": 1.1989,
+      "step": 30
+    },
+    {
+      "epoch": 0.08838383838383838,
+      "grad_norm": 0.5708086490631104,
+      "learning_rate": 1.0303030303030302e-05,
+      "loss": 1.1534,
+      "step": 35
+    },
+    {
+      "epoch": 0.10101010101010101,
+      "grad_norm": 0.5267711281776428,
+      "learning_rate": 1.1818181818181819e-05,
+      "loss": 1.2464,
+      "step": 40
+    },
+    {
+      "epoch": 0.11363636363636363,
+      "grad_norm": 0.5095251798629761,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.1572,
+      "step": 45
+    },
+    {
+      "epoch": 0.12626262626262627,
+      "grad_norm": 0.4897612929344177,
+      "learning_rate": 1.484848484848485e-05,
+      "loss": 1.1698,
+      "step": 50
+    },
+    {
+      "epoch": 0.1388888888888889,
+      "grad_norm": 0.4447384476661682,
+      "learning_rate": 1.6363636363636363e-05,
+      "loss": 1.0932,
+      "step": 55
+    },
+    {
+      "epoch": 0.15151515151515152,
+      "grad_norm": 0.5444252490997314,
+      "learning_rate": 1.7878787878787877e-05,
+      "loss": 1.1421,
+      "step": 60
+    },
+    {
+      "epoch": 0.16414141414141414,
+      "grad_norm": 0.4492619037628174,
+      "learning_rate": 1.9393939393939395e-05,
+      "loss": 1.1182,
+      "step": 65
+    },
+    {
+      "epoch": 0.17676767676767677,
+      "grad_norm": 0.541714072227478,
+      "learning_rate": 2.090909090909091e-05,
+      "loss": 1.1767,
+      "step": 70
+    },
+    {
+      "epoch": 0.1893939393939394,
+      "grad_norm": 0.43421369791030884,
+      "learning_rate": 2.2424242424242424e-05,
+      "loss": 1.0917,
+      "step": 75
+    },
+    {
+      "epoch": 0.20202020202020202,
+      "grad_norm": 0.4997319281101227,
+      "learning_rate": 2.3939393939393942e-05,
+      "loss": 1.1036,
+      "step": 80
+    },
+    {
+      "epoch": 0.21464646464646464,
+      "grad_norm": 0.4616036117076874,
+      "learning_rate": 2.5454545454545457e-05,
+      "loss": 1.1136,
+      "step": 85
+    },
+    {
+      "epoch": 0.22727272727272727,
+      "grad_norm": 0.5620744228363037,
+      "learning_rate": 2.696969696969697e-05,
+      "loss": 1.1174,
+      "step": 90
+    },
+    {
+      "epoch": 0.2398989898989899,
+      "grad_norm": 0.48610466718673706,
+      "learning_rate": 2.8484848484848486e-05,
+      "loss": 1.1173,
+      "step": 95
+    },
+    {
+      "epoch": 0.25252525252525254,
+      "grad_norm": 0.8980438113212585,
+      "learning_rate": 3e-05,
+      "loss": 1.0533,
+      "step": 100
+    },
+    {
+      "epoch": 0.26515151515151514,
+      "grad_norm": 0.5355147123336792,
+      "learning_rate": 2.9999476976861445e-05,
+      "loss": 1.0703,
+      "step": 105
+    },
+    {
+      "epoch": 0.2777777777777778,
+      "grad_norm": 0.6903505325317383,
+      "learning_rate": 2.999790794391954e-05,
+      "loss": 1.0783,
+      "step": 110
+    },
+    {
+      "epoch": 0.2904040404040404,
+      "grad_norm": 0.5622875094413757,
+      "learning_rate": 2.999529301059302e-05,
+      "loss": 1.0516,
+      "step": 115
+    },
+    {
+      "epoch": 0.30303030303030304,
+      "grad_norm": 0.5711376070976257,
+      "learning_rate": 2.9991632359237976e-05,
+      "loss": 1.016,
+      "step": 120
+    },
+    {
+      "epoch": 0.31565656565656564,
+      "grad_norm": 0.6692824959754944,
+      "learning_rate": 2.9986926245135113e-05,
+      "loss": 1.0827,
+      "step": 125
+    },
+    {
+      "epoch": 0.3282828282828283,
+      "grad_norm": 0.5953907370567322,
+      "learning_rate": 2.9981174996471984e-05,
+      "loss": 1.0046,
+      "step": 130
+    },
+    {
+      "epoch": 0.3409090909090909,
+      "grad_norm": 0.5587031841278076,
+      "learning_rate": 2.9974379014320066e-05,
+      "loss": 1.0265,
+      "step": 135
+    },
+    {
+      "epoch": 0.35353535353535354,
+      "grad_norm": 0.5270025730133057,
+      "learning_rate": 2.996653877260682e-05,
+      "loss": 1.0009,
+      "step": 140
+    },
+    {
+      "epoch": 0.3661616161616162,
+      "grad_norm": 0.6027882695198059,
+      "learning_rate": 2.9957654818082615e-05,
+      "loss": 1.0222,
+      "step": 145
+    },
+    {
+      "epoch": 0.3787878787878788,
+      "grad_norm": 0.5629182457923889,
+      "learning_rate": 2.9947727770282626e-05,
+      "loss": 0.9617,
+      "step": 150
+    },
+    {
+      "epoch": 0.39141414141414144,
+      "grad_norm": 0.6864057183265686,
+      "learning_rate": 2.9936758321483616e-05,
+      "loss": 0.9682,
+      "step": 155
+    },
+    {
+      "epoch": 0.40404040404040403,
+      "grad_norm": 0.623359739780426,
+      "learning_rate": 2.992474723665565e-05,
+      "loss": 0.9456,
+      "step": 160
+    },
+    {
+      "epoch": 0.4166666666666667,
+      "grad_norm": 0.6435149312019348,
+      "learning_rate": 2.9911695353408766e-05,
+      "loss": 0.965,
+      "step": 165
+    },
+    {
+      "epoch": 0.4292929292929293,
+      "grad_norm": 0.6187605857849121,
+      "learning_rate": 2.989760358193456e-05,
+      "loss": 0.9599,
+      "step": 170
+    },
+    {
+      "epoch": 0.44191919191919193,
+      "grad_norm": 0.7720528244972229,
+      "learning_rate": 2.9882472904942707e-05,
+      "loss": 0.9487,
+      "step": 175
+    },
+    {
+      "epoch": 0.45454545454545453,
+      "grad_norm": 0.7527753710746765,
+      "learning_rate": 2.9866304377592427e-05,
+      "loss": 0.9445,
+      "step": 180
+    },
+    {
+      "epoch": 0.4671717171717172,
+      "grad_norm": 0.7059020400047302,
+      "learning_rate": 2.9849099127418905e-05,
+      "loss": 0.8939,
+      "step": 185
+    },
+    {
+      "epoch": 0.4797979797979798,
+      "grad_norm": 0.6808345913887024,
+      "learning_rate": 2.9830858354254672e-05,
+      "loss": 0.8791,
+      "step": 190
+    },
+    {
+      "epoch": 0.49242424242424243,
+      "grad_norm": 0.7607003450393677,
+      "learning_rate": 2.9811583330145915e-05,
+      "loss": 0.9394,
+      "step": 195
+    },
+    {
+      "epoch": 0.5050505050505051,
+      "grad_norm": 0.6947716474533081,
+      "learning_rate": 2.9791275399263785e-05,
+      "loss": 0.8957,
+      "step": 200
+    },
+    {
+      "epoch": 0.5176767676767676,
+      "grad_norm": 0.7062451839447021,
+      "learning_rate": 2.976993597781065e-05,
+      "loss": 0.9051,
+      "step": 205
+    },
+    {
+      "epoch": 0.5303030303030303,
+      "grad_norm": 0.7581499814987183,
+      "learning_rate": 2.9747566553921326e-05,
+      "loss": 0.8459,
+      "step": 210
+    },
+    {
+      "epoch": 0.5429292929292929,
+      "grad_norm": 1.0111594200134277,
+      "learning_rate": 2.9724168687559325e-05,
+      "loss": 0.8905,
+      "step": 215
+    },
+    {
+      "epoch": 0.5555555555555556,
+      "grad_norm": 0.915945291519165,
+      "learning_rate": 2.969974401040805e-05,
+      "loss": 0.8879,
+      "step": 220
+    },
+    {
+      "epoch": 0.5681818181818182,
+      "grad_norm": 0.7350049018859863,
+      "learning_rate": 2.9674294225757e-05,
+      "loss": 0.8922,
+      "step": 225
+    },
+    {
+      "epoch": 0.5808080808080808,
+      "grad_norm": 0.7845703363418579,
+      "learning_rate": 2.964782110838301e-05,
+      "loss": 0.8542,
+      "step": 230
+    },
+    {
+      "epoch": 0.5934343434343434,
+      "grad_norm": 0.8170814514160156,
+      "learning_rate": 2.9620326504426476e-05,
+      "loss": 0.8349,
+      "step": 235
+    },
+    {
+      "epoch": 0.6060606060606061,
+      "grad_norm": 0.7793628573417664,
+      "learning_rate": 2.9591812331262598e-05,
+      "loss": 0.8179,
+      "step": 240
+    },
+    {
+      "epoch": 0.6186868686868687,
+      "grad_norm": 0.7981735467910767,
+      "learning_rate": 2.9562280577367694e-05,
+      "loss": 0.8412,
+      "step": 245
+    },
+    {
+      "epoch": 0.6313131313131313,
+      "grad_norm": 0.7543655633926392,
+      "learning_rate": 2.953173330218051e-05,
+      "loss": 0.8018,
+      "step": 250
+    },
+    {
+      "epoch": 0.6439393939393939,
+      "grad_norm": 0.8550294041633606,
+      "learning_rate": 2.950017263595861e-05,
+      "loss": 0.8431,
+      "step": 255
+    },
+    {
+      "epoch": 0.6565656565656566,
+      "grad_norm": 0.8755800724029541,
+      "learning_rate": 2.9467600779629825e-05,
+      "loss": 0.8049,
+      "step": 260
+    },
+    {
+      "epoch": 0.6691919191919192,
+      "grad_norm": 0.8669186234474182,
+      "learning_rate": 2.9434020004638757e-05,
+      "loss": 0.8233,
+      "step": 265
+    },
+    {
+      "epoch": 0.6818181818181818,
+      "grad_norm": 0.7797918915748596,
+      "learning_rate": 2.9399432652788383e-05,
+      "loss": 0.8118,
+      "step": 270
+    },
+    {
+      "epoch": 0.6944444444444444,
+      "grad_norm": 0.9602000117301941,
+      "learning_rate": 2.936384113607674e-05,
+      "loss": 0.809,
+      "step": 275
+    },
+    {
+      "epoch": 0.7070707070707071,
+      "grad_norm": 0.7648482918739319,
+      "learning_rate": 2.9327247936528742e-05,
+      "loss": 0.7763,
+      "step": 280
+    },
+    {
+      "epoch": 0.7196969696969697,
+      "grad_norm": 0.8220114707946777,
+      "learning_rate": 2.9289655606023056e-05,
+      "loss": 0.8279,
+      "step": 285
+    },
+    {
+      "epoch": 0.7323232323232324,
+      "grad_norm": 0.8677099347114563,
+      "learning_rate": 2.925106676611418e-05,
+      "loss": 0.7877,
+      "step": 290
+    },
+    {
+      "epoch": 0.7449494949494949,
+      "grad_norm": 0.912298858165741,
+      "learning_rate": 2.92114841078496e-05,
+      "loss": 0.7746,
+      "step": 295
+    },
+    {
+      "epoch": 0.7575757575757576,
+      "grad_norm": 0.9250658750534058,
+      "learning_rate": 2.917091039158214e-05,
+      "loss": 0.7766,
+      "step": 300
+    },
+    {
+      "epoch": 0.7702020202020202,
+      "grad_norm": 0.9757301807403564,
+      "learning_rate": 2.9129348446777456e-05,
+      "loss": 0.7492,
+      "step": 305
+    },
+    {
+      "epoch": 0.7828282828282829,
+      "grad_norm": 0.9662628769874573,
+      "learning_rate": 2.9086801171816717e-05,
+      "loss": 0.764,
+      "step": 310
+    },
+    {
+      "epoch": 0.7954545454545454,
+      "grad_norm": 0.8361748456954956,
+      "learning_rate": 2.90432715337945e-05,
+      "loss": 0.6907,
+      "step": 315
+    },
+    {
+      "epoch": 0.8080808080808081,
+      "grad_norm": 0.8114555478096008,
+      "learning_rate": 2.8998762568311857e-05,
+      "loss": 0.6749,
+      "step": 320
+    },
+    {
+      "epoch": 0.8207070707070707,
+      "grad_norm": 0.9168850779533386,
+      "learning_rate": 2.8953277379264633e-05,
+      "loss": 0.7378,
+      "step": 325
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 0.9979681968688965,
+      "learning_rate": 2.8906819138627002e-05,
+      "loss": 0.7289,
+      "step": 330
+    },
+    {
+      "epoch": 0.8459595959595959,
+      "grad_norm": 0.9025216698646545,
+      "learning_rate": 2.885939108623028e-05,
+      "loss": 0.7326,
+      "step": 335
+    },
+    {
+      "epoch": 0.8585858585858586,
+      "grad_norm": 0.8616014122962952,
+      "learning_rate": 2.8810996529536967e-05,
+      "loss": 0.6868,
+      "step": 340
+    },
+    {
+      "epoch": 0.8712121212121212,
+      "grad_norm": 0.9671257734298706,
+      "learning_rate": 2.8761638843410133e-05,
+      "loss": 0.7142,
+      "step": 345
+    },
+    {
+      "epoch": 0.8838383838383839,
+      "grad_norm": 1.1255998611450195,
+      "learning_rate": 2.8711321469878024e-05,
+      "loss": 0.7196,
+      "step": 350
+    },
+    {
+      "epoch": 0.8964646464646465,
+      "grad_norm": 0.9464638829231262,
+      "learning_rate": 2.866004791789406e-05,
+      "loss": 0.6761,
+      "step": 355
+    },
+    {
+      "epoch": 0.9090909090909091,
+      "grad_norm": 1.016979455947876,
+      "learning_rate": 2.8607821763092116e-05,
+      "loss": 0.7013,
+      "step": 360
+    },
+    {
+      "epoch": 0.9217171717171717,
+      "grad_norm": 1.0915886163711548,
+      "learning_rate": 2.8554646647537178e-05,
+      "loss": 0.6671,
+      "step": 365
+    },
+    {
+      "epoch": 0.9343434343434344,
+      "grad_norm": 1.0145480632781982,
+      "learning_rate": 2.8500526279471362e-05,
+      "loss": 0.6824,
+      "step": 370
+    },
+    {
+      "epoch": 0.946969696969697,
+      "grad_norm": 0.9449098706245422,
+      "learning_rate": 2.8445464433055303e-05,
+      "loss": 0.7025,
+      "step": 375
+    },
+    {
+      "epoch": 0.9595959595959596,
+      "grad_norm": 1.0181083679199219,
+      "learning_rate": 2.8389464948104962e-05,
+      "loss": 0.6374,
+      "step": 380
+    },
+    {
+      "epoch": 0.9722222222222222,
+      "grad_norm": 1.1749608516693115,
+      "learning_rate": 2.8332531729823853e-05,
+      "loss": 0.6567,
+      "step": 385
+    },
+    {
+      "epoch": 0.9848484848484849,
+      "grad_norm": 1.0283434391021729,
+      "learning_rate": 2.8274668748530718e-05,
+      "loss": 0.7205,
+      "step": 390
+    },
+    {
+      "epoch": 0.9974747474747475,
+      "grad_norm": 0.9649919867515564,
+      "learning_rate": 2.8215880039382625e-05,
+      "loss": 0.671,
+      "step": 395
+    },
+    {
+      "epoch": 1.0101010101010102,
+      "grad_norm": 0.9317898154258728,
+      "learning_rate": 2.8156169702093608e-05,
+      "loss": 0.592,
+      "step": 400
+    },
+    {
+      "epoch": 1.0227272727272727,
+      "grad_norm": 1.047213077545166,
+      "learning_rate": 2.809554190064873e-05,
+      "loss": 0.5692,
+      "step": 405
+    },
+    {
+      "epoch": 1.0353535353535352,
+      "grad_norm": 0.9808226823806763,
+      "learning_rate": 2.803400086301372e-05,
+      "loss": 0.549,
+      "step": 410
+    },
+    {
+      "epoch": 1.047979797979798,
+      "grad_norm": 1.1340376138687134,
+      "learning_rate": 2.7971550880840138e-05,
+      "loss": 0.5457,
+      "step": 415
+    },
+    {
+      "epoch": 1.0606060606060606,
+      "grad_norm": 0.9728055000305176,
+      "learning_rate": 2.7908196309166074e-05,
+      "loss": 0.5686,
+      "step": 420
+    },
+    {
+      "epoch": 1.0732323232323233,
+      "grad_norm": 0.9779664278030396,
+      "learning_rate": 2.7843941566112442e-05,
+      "loss": 0.5344,
+      "step": 425
+    },
+    {
+      "epoch": 1.0858585858585859,
+      "grad_norm": 0.9450783133506775,
+      "learning_rate": 2.7778791132574908e-05,
+      "loss": 0.5322,
+      "step": 430
+    },
+    {
+      "epoch": 1.0984848484848484,
+      "grad_norm": 0.994297981262207,
+      "learning_rate": 2.7712749551911355e-05,
+      "loss": 0.5255,
+      "step": 435
+    },
+    {
+      "epoch": 1.1111111111111112,
+      "grad_norm": 0.9942148327827454,
+      "learning_rate": 2.7645821429625092e-05,
+      "loss": 0.5137,
+      "step": 440
+    },
+    {
+      "epoch": 1.1237373737373737,
+      "grad_norm": 1.0680190324783325,
+      "learning_rate": 2.757801143304367e-05,
+      "loss": 0.5647,
+      "step": 445
+    },
+    {
+      "epoch": 1.1363636363636362,
+      "grad_norm": 1.0500057935714722,
+      "learning_rate": 2.750932429099338e-05,
+      "loss": 0.5349,
+      "step": 450
+    },
+    {
+      "epoch": 1.148989898989899,
+      "grad_norm": 1.1989120244979858,
+      "learning_rate": 2.7439764793469504e-05,
+      "loss": 0.5376,
+      "step": 455
+    },
+    {
+      "epoch": 1.1616161616161615,
+      "grad_norm": 1.0777108669281006,
+      "learning_rate": 2.7369337791302272e-05,
+      "loss": 0.5937,
+      "step": 460
+    },
+    {
+      "epoch": 1.1742424242424243,
+      "grad_norm": 1.0498294830322266,
+      "learning_rate": 2.729804819581858e-05,
+      "loss": 0.5608,
+      "step": 465
+    },
+    {
+      "epoch": 1.1868686868686869,
+      "grad_norm": 0.9939706325531006,
+      "learning_rate": 2.7225900978499487e-05,
+      "loss": 0.4959,
+      "step": 470
+    },
+    {
+      "epoch": 1.1994949494949494,
+      "grad_norm": 1.0636131763458252,
+      "learning_rate": 2.715290117063354e-05,
+      "loss": 0.5269,
+      "step": 475
+    },
+    {
+      "epoch": 1.2121212121212122,
+      "grad_norm": 1.0121461153030396,
+      "learning_rate": 2.707905386296588e-05,
+      "loss": 0.4775,
+      "step": 480
+    },
+    {
+      "epoch": 1.2247474747474747,
+      "grad_norm": 1.0765471458435059,
+      "learning_rate": 2.700436420534326e-05,
+      "loss": 0.4939,
+      "step": 485
+    },
+    {
+      "epoch": 1.2373737373737375,
+      "grad_norm": 1.1116187572479248,
+      "learning_rate": 2.6928837406354905e-05,
+      "loss": 0.5349,
+      "step": 490
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.9618929624557495,
+      "learning_rate": 2.6852478732969272e-05,
+      "loss": 0.5085,
+      "step": 495
+    },
+    {
+      "epoch": 1.2626262626262625,
+      "grad_norm": 1.0578511953353882,
+      "learning_rate": 2.677529351016676e-05,
+      "loss": 0.4741,
+      "step": 500
+    },
+    {
+      "epoch": 1.2752525252525253,
+      "grad_norm": 0.9444258809089661,
+      "learning_rate": 2.6697287120568364e-05,
+      "loss": 0.5153,
+      "step": 505
+    },
+    {
+      "epoch": 1.2878787878787878,
+      "grad_norm": 1.1680361032485962,
+      "learning_rate": 2.6618465004060324e-05,
+      "loss": 0.5033,
+      "step": 510
+    },
+    {
+      "epoch": 1.3005050505050506,
+      "grad_norm": 1.0985764265060425,
+      "learning_rate": 2.653883265741473e-05,
+      "loss": 0.4917,
+      "step": 515
+    },
+    {
+      "epoch": 1.3131313131313131,
+      "grad_norm": 1.1677299737930298,
+      "learning_rate": 2.645839563390624e-05,
+      "loss": 0.5526,
+      "step": 520
+    },
+    {
+      "epoch": 1.3257575757575757,
+      "grad_norm": 1.122022271156311,
+      "learning_rate": 2.637715954292478e-05,
+      "loss": 0.5197,
+      "step": 525
+    },
+    {
+      "epoch": 1.3383838383838385,
+      "grad_norm": 1.02968430519104,
+      "learning_rate": 2.6295130049584388e-05,
+      "loss": 0.4619,
+      "step": 530
+    },
+    {
+      "epoch": 1.351010101010101,
+      "grad_norm": 0.9885948300361633,
+      "learning_rate": 2.6212312874328136e-05,
+      "loss": 0.4599,
+      "step": 535
+    },
+    {
+      "epoch": 1.3636363636363638,
+      "grad_norm": 1.1250592470169067,
+      "learning_rate": 2.6128713792529225e-05,
+      "loss": 0.4805,
+      "step": 540
+    },
+    {
+      "epoch": 1.3762626262626263,
+      "grad_norm": 1.128479242324829,
+      "learning_rate": 2.6044338634088198e-05,
+      "loss": 0.4858,
+      "step": 545
+    },
+    {
+      "epoch": 1.3888888888888888,
+      "grad_norm": 1.0660187005996704,
+      "learning_rate": 2.595919328302641e-05,
+      "loss": 0.4568,
+      "step": 550
+    },
+    {
+      "epoch": 1.4015151515151514,
+      "grad_norm": 1.115667700767517,
+      "learning_rate": 2.5873283677075704e-05,
+      "loss": 0.4843,
+      "step": 555
+    },
+    {
+      "epoch": 1.4141414141414141,
+      "grad_norm": 1.0488578081130981,
+      "learning_rate": 2.5786615807264306e-05,
+      "loss": 0.4969,
+      "step": 560
+    },
+    {
+      "epoch": 1.4267676767676767,
+      "grad_norm": 0.9793870449066162,
+      "learning_rate": 2.569919571749905e-05,
+      "loss": 0.486,
+      "step": 565
+    },
+    {
+      "epoch": 1.4393939393939394,
+      "grad_norm": 1.2411023378372192,
+      "learning_rate": 2.5611029504143905e-05,
+      "loss": 0.415,
+      "step": 570
+    },
+    {
+      "epoch": 1.452020202020202,
+      "grad_norm": 1.0087097883224487,
+      "learning_rate": 2.552212331559482e-05,
+      "loss": 0.4396,
+      "step": 575
+    },
+    {
+      "epoch": 1.4646464646464645,
+      "grad_norm": 1.0943728685379028,
+      "learning_rate": 2.5432483351850962e-05,
+      "loss": 0.4658,
+      "step": 580
+    },
+    {
+      "epoch": 1.4772727272727273,
+      "grad_norm": 1.065333604812622,
+      "learning_rate": 2.5342115864082355e-05,
+      "loss": 0.4314,
+      "step": 585
+    },
+    {
+      "epoch": 1.4898989898989898,
+      "grad_norm": 1.1207003593444824,
+      "learning_rate": 2.5251027154193945e-05,
+      "loss": 0.4763,
+      "step": 590
+    },
+    {
+      "epoch": 1.5025252525252526,
+      "grad_norm": 1.0626108646392822,
+      "learning_rate": 2.5159223574386117e-05,
+      "loss": 0.408,
+      "step": 595
+    },
+    {
+      "epoch": 1.5151515151515151,
+      "grad_norm": 1.0062793493270874,
+      "learning_rate": 2.5066711526711736e-05,
+      "loss": 0.4337,
+      "step": 600
+    },
+    {
+      "epoch": 1.5277777777777777,
+      "grad_norm": 1.0512022972106934,
+      "learning_rate": 2.497349746262967e-05,
+      "loss": 0.4004,
+      "step": 605
+    },
+    {
+      "epoch": 1.5404040404040404,
+      "grad_norm": 1.0447345972061157,
+      "learning_rate": 2.4879587882554903e-05,
+      "loss": 0.4434,
+      "step": 610
+    },
+    {
+      "epoch": 1.553030303030303,
+      "grad_norm": 1.0264374017715454,
+      "learning_rate": 2.4784989335405203e-05,
+      "loss": 0.4561,
+      "step": 615
+    },
+    {
+      "epoch": 1.5656565656565657,
+      "grad_norm": 1.2251687049865723,
+      "learning_rate": 2.468970841814445e-05,
+      "loss": 0.4333,
+      "step": 620
+    },
+    {
+      "epoch": 1.5782828282828283,
+      "grad_norm": 1.082527995109558,
+      "learning_rate": 2.4593751775322553e-05,
+      "loss": 0.4615,
+      "step": 625
+    },
+    {
+      "epoch": 1.5909090909090908,
+      "grad_norm": 1.155479073524475,
+      "learning_rate": 2.4497126098612115e-05,
+      "loss": 0.4304,
+      "step": 630
+    },
+    {
+      "epoch": 1.6035353535353534,
+      "grad_norm": 1.0841268301010132,
+      "learning_rate": 2.4399838126341768e-05,
+      "loss": 0.4345,
+      "step": 635
+    },
+    {
+      "epoch": 1.6161616161616161,
+      "grad_norm": 1.0636403560638428,
+      "learning_rate": 2.430189464302625e-05,
+      "loss": 0.3878,
+      "step": 640
+    },
+    {
+      "epoch": 1.628787878787879,
+      "grad_norm": 1.0887062549591064,
+      "learning_rate": 2.4203302478893307e-05,
+      "loss": 0.4154,
+      "step": 645
+    },
+    {
+      "epoch": 1.6414141414141414,
+      "grad_norm": 1.071465015411377,
+      "learning_rate": 2.410406850940735e-05,
+      "loss": 0.4158,
+      "step": 650
+    },
+    {
+      "epoch": 1.654040404040404,
+      "grad_norm": 1.0511163473129272,
+      "learning_rate": 2.4004199654790008e-05,
+      "loss": 0.3878,
+      "step": 655
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 1.1062862873077393,
+      "learning_rate": 2.3903702879537513e-05,
+      "loss": 0.4166,
+      "step": 660
+    },
+    {
+      "epoch": 1.6792929292929293,
+      "grad_norm": 1.101647138595581,
+      "learning_rate": 2.3802585191935044e-05,
+      "loss": 0.3954,
+      "step": 665
+    },
+    {
+      "epoch": 1.691919191919192,
+      "grad_norm": 1.1077150106430054,
+      "learning_rate": 2.3700853643567973e-05,
+      "loss": 0.413,
+      "step": 670
+    },
+    {
+      "epoch": 1.7045454545454546,
+      "grad_norm": 1.1404215097427368,
+      "learning_rate": 2.3598515328830135e-05,
+      "loss": 0.4299,
+      "step": 675
+    },
+    {
+      "epoch": 1.7171717171717171,
+      "grad_norm": 0.9413719773292542,
+      "learning_rate": 2.349557738442907e-05,
+      "loss": 0.361,
+      "step": 680
+    },
+    {
+      "epoch": 1.7297979797979797,
+      "grad_norm": 1.1357874870300293,
+      "learning_rate": 2.3392046988888345e-05,
+      "loss": 0.3782,
+      "step": 685
+    },
+    {
+      "epoch": 1.7424242424242424,
+      "grad_norm": 1.0469523668289185,
+      "learning_rate": 2.328793136204695e-05,
+      "loss": 0.3949,
+      "step": 690
+    },
+    {
+      "epoch": 1.7550505050505052,
+      "grad_norm": 1.0579326152801514,
+      "learning_rate": 2.3183237764555807e-05,
+      "loss": 0.4216,
+      "step": 695
+    },
+    {
+      "epoch": 1.7676767676767677,
+      "grad_norm": 1.1634066104888916,
+      "learning_rate": 2.307797349737144e-05,
+      "loss": 0.4091,
+      "step": 700
+    },
+    {
+      "epoch": 1.7803030303030303,
+      "grad_norm": 1.1639355421066284,
+      "learning_rate": 2.297214590124684e-05,
+      "loss": 0.4212,
+      "step": 705
+    },
+    {
+      "epoch": 1.7929292929292928,
+      "grad_norm": 0.9826981425285339,
+      "learning_rate": 2.2865762356219533e-05,
+      "loss": 0.3518,
+      "step": 710
+    },
+    {
+      "epoch": 1.8055555555555556,
+      "grad_norm": 1.0579651594161987,
+      "learning_rate": 2.2758830281096935e-05,
+      "loss": 0.3899,
+      "step": 715
+    },
+    {
+      "epoch": 1.8181818181818183,
+      "grad_norm": 1.061562418937683,
+      "learning_rate": 2.265135713293899e-05,
+      "loss": 0.3874,
+      "step": 720
+    },
+    {
+      "epoch": 1.8308080808080809,
+      "grad_norm": 1.1707763671875,
+      "learning_rate": 2.254335040653812e-05,
+      "loss": 0.376,
+      "step": 725
+    },
+    {
+      "epoch": 1.8434343434343434,
+      "grad_norm": 1.1716967821121216,
+      "learning_rate": 2.243481763389661e-05,
+      "loss": 0.3391,
+      "step": 730
+    },
+    {
+      "epoch": 1.856060606060606,
+      "grad_norm": 1.1718958616256714,
+      "learning_rate": 2.23257663837013e-05,
+      "loss": 0.361,
+      "step": 735
+    },
+    {
+      "epoch": 1.8686868686868687,
+      "grad_norm": 1.1531727313995361,
+      "learning_rate": 2.2216204260795813e-05,
+      "loss": 0.3678,
+      "step": 740
+    },
+    {
+      "epoch": 1.8813131313131313,
+      "grad_norm": 1.0929404497146606,
+      "learning_rate": 2.21061389056502e-05,
+      "loss": 0.3785,
+      "step": 745
+    },
+    {
+      "epoch": 1.893939393939394,
+      "grad_norm": 1.059478521347046,
+      "learning_rate": 2.199557799382813e-05,
+      "loss": 0.3319,
+      "step": 750
+    },
+    {
+      "epoch": 1.9065656565656566,
+      "grad_norm": 1.037405014038086,
+      "learning_rate": 2.1884529235451618e-05,
+      "loss": 0.3493,
+      "step": 755
+    },
+    {
+      "epoch": 1.9191919191919191,
+      "grad_norm": 1.3604897260665894,
+      "learning_rate": 2.177300037466334e-05,
+      "loss": 0.3749,
+      "step": 760
+    },
+    {
+      "epoch": 1.9318181818181817,
+      "grad_norm": 0.9598345160484314,
+      "learning_rate": 2.1660999189086613e-05,
+      "loss": 0.3459,
+      "step": 765
+    },
+    {
+      "epoch": 1.9444444444444444,
+      "grad_norm": 1.1897592544555664,
+      "learning_rate": 2.1548533489282977e-05,
+      "loss": 0.361,
+      "step": 770
+    },
+    {
+      "epoch": 1.9570707070707072,
+      "grad_norm": 1.1726038455963135,
+      "learning_rate": 2.1435611118207546e-05,
+      "loss": 0.354,
+      "step": 775
+    },
+    {
+      "epoch": 1.9696969696969697,
+      "grad_norm": 1.1047810316085815,
+      "learning_rate": 2.1322239950662035e-05,
+      "loss": 0.3034,
+      "step": 780
+    },
+    {
+      "epoch": 1.9823232323232323,
+      "grad_norm": 1.1414453983306885,
+      "learning_rate": 2.120842789274563e-05,
+      "loss": 0.3174,
+      "step": 785
+    },
+    {
+      "epoch": 1.9949494949494948,
+      "grad_norm": 1.0688472986221313,
+      "learning_rate": 2.1094182881303636e-05,
+      "loss": 0.3343,
+      "step": 790
+    },
+    {
+      "epoch": 2.007575757575758,
+      "grad_norm": 1.1202057600021362,
+      "learning_rate": 2.0979512883373972e-05,
+      "loss": 0.2793,
+      "step": 795
+    },
+    {
+      "epoch": 2.0202020202020203,
+      "grad_norm": 1.4000380039215088,
+      "learning_rate": 2.08644258956316e-05,
+      "loss": 0.2709,
+      "step": 800
+    },
+    {
+      "epoch": 2.032828282828283,
+      "grad_norm": 0.9448375105857849,
+      "learning_rate": 2.0748929943830863e-05,
+      "loss": 0.2726,
+      "step": 805
+    },
+    {
+      "epoch": 2.0454545454545454,
+      "grad_norm": 1.2739609479904175,
+      "learning_rate": 2.0633033082245782e-05,
+      "loss": 0.2872,
+      "step": 810
+    },
+    {
+      "epoch": 2.058080808080808,
+      "grad_norm": 1.1996492147445679,
+      "learning_rate": 2.05167433931084e-05,
+      "loss": 0.311,
+      "step": 815
+    },
+    {
+      "epoch": 2.0707070707070705,
+      "grad_norm": 1.1399296522140503,
+      "learning_rate": 2.0400068986045142e-05,
+      "loss": 0.2853,
+      "step": 820
+    },
+    {
+      "epoch": 2.0833333333333335,
+      "grad_norm": 1.0565412044525146,
+      "learning_rate": 2.0283017997511283e-05,
+      "loss": 0.2629,
+      "step": 825
+    },
+    {
+      "epoch": 2.095959595959596,
+      "grad_norm": 1.0858699083328247,
+      "learning_rate": 2.016559859022355e-05,
+      "loss": 0.2907,
+      "step": 830
+    },
+    {
+      "epoch": 2.1085858585858586,
+      "grad_norm": 1.0091416835784912,
+      "learning_rate": 2.0047818952590854e-05,
+      "loss": 0.2576,
+      "step": 835
+    },
+    {
+      "epoch": 2.121212121212121,
+      "grad_norm": 1.1173107624053955,
+      "learning_rate": 1.99296872981433e-05,
+      "loss": 0.2398,
+      "step": 840
+    },
+    {
+      "epoch": 2.1338383838383836,
+      "grad_norm": 1.1509712934494019,
+      "learning_rate": 1.9811211864959374e-05,
+      "loss": 0.2555,
+      "step": 845
+    },
+    {
+      "epoch": 2.1464646464646466,
+      "grad_norm": 1.1491074562072754,
+      "learning_rate": 1.969240091509147e-05,
+      "loss": 0.2791,
+      "step": 850
+    },
+    {
+      "epoch": 2.159090909090909,
+      "grad_norm": 1.2488752603530884,
+      "learning_rate": 1.95732627339897e-05,
+      "loss": 0.236,
+      "step": 855
+    },
+    {
+      "epoch": 2.1717171717171717,
+      "grad_norm": 1.0945607423782349,
+      "learning_rate": 1.9453805629924126e-05,
+      "loss": 0.2854,
+      "step": 860
+    },
+    {
+      "epoch": 2.1843434343434343,
+      "grad_norm": 1.2910445928573608,
+      "learning_rate": 1.9334037933405337e-05,
+      "loss": 0.2543,
+      "step": 865
+    },
+    {
+      "epoch": 2.196969696969697,
+      "grad_norm": 1.092241883277893,
+      "learning_rate": 1.9213967996603542e-05,
+      "loss": 0.2535,
+      "step": 870
+    },
+    {
+      "epoch": 2.20959595959596,
+      "grad_norm": 1.2114317417144775,
+      "learning_rate": 1.9093604192766102e-05,
+      "loss": 0.2615,
+      "step": 875
+    },
+    {
+      "epoch": 2.2222222222222223,
+      "grad_norm": 0.9923803806304932,
+      "learning_rate": 1.8972954915633604e-05,
+      "loss": 0.2231,
+      "step": 880
+    },
+    {
+      "epoch": 2.234848484848485,
+      "grad_norm": 1.0003025531768799,
+      "learning_rate": 1.8852028578854532e-05,
+      "loss": 0.2441,
+      "step": 885
+    },
+    {
+      "epoch": 2.2474747474747474,
+      "grad_norm": 1.0413899421691895,
+      "learning_rate": 1.873083361539851e-05,
+      "loss": 0.2414,
+      "step": 890
+    },
+    {
+      "epoch": 2.26010101010101,
+      "grad_norm": 1.1812835931777954,
+      "learning_rate": 1.8609378476968232e-05,
+      "loss": 0.2434,
+      "step": 895
+    },
+    {
+      "epoch": 2.2727272727272725,
+      "grad_norm": 1.0742385387420654,
+      "learning_rate": 1.8487671633410052e-05,
+      "loss": 0.2539,
+      "step": 900
+    },
+    {
+      "epoch": 2.2853535353535355,
+      "grad_norm": 1.2803466320037842,
+      "learning_rate": 1.836572157212334e-05,
+      "loss": 0.2478,
+      "step": 905
+    },
+    {
+      "epoch": 2.297979797979798,
+      "grad_norm": 1.0669132471084595,
+      "learning_rate": 1.824353679746861e-05,
+      "loss": 0.2271,
+      "step": 910
+    },
+    {
+      "epoch": 2.3106060606060606,
+      "grad_norm": 1.1045911312103271,
+      "learning_rate": 1.8121125830174437e-05,
+      "loss": 0.2572,
+      "step": 915
+    },
+    {
+      "epoch": 2.323232323232323,
+      "grad_norm": 1.1256022453308105,
+      "learning_rate": 1.799849720674326e-05,
+      "loss": 0.2582,
+      "step": 920
+    },
+    {
+      "epoch": 2.3358585858585856,
+      "grad_norm": 1.091371774673462,
+      "learning_rate": 1.7875659478856077e-05,
+      "loss": 0.2495,
+      "step": 925
+    },
+    {
+      "epoch": 2.3484848484848486,
+      "grad_norm": 1.0141551494598389,
+      "learning_rate": 1.775262121277609e-05,
+      "loss": 0.2513,
+      "step": 930
+    },
+    {
+      "epoch": 2.361111111111111,
+      "grad_norm": 1.1098726987838745,
+      "learning_rate": 1.7629390988751307e-05,
+      "loss": 0.2375,
+      "step": 935
+    },
+    {
+      "epoch": 2.3737373737373737,
+      "grad_norm": 1.0376847982406616,
+      "learning_rate": 1.7505977400416207e-05,
+      "loss": 0.2161,
+      "step": 940
+    },
+    {
+      "epoch": 2.3863636363636362,
+      "grad_norm": 1.1401861906051636,
+      "learning_rate": 1.738238905419242e-05,
+      "loss": 0.2434,
+      "step": 945
+    },
+    {
+      "epoch": 2.398989898989899,
+      "grad_norm": 1.2411974668502808,
+      "learning_rate": 1.7258634568688577e-05,
+      "loss": 0.245,
+      "step": 950
+    },
+    {
+      "epoch": 2.4116161616161618,
+      "grad_norm": 1.1614488363265991,
+      "learning_rate": 1.713472257409928e-05,
+      "loss": 0.216,
+      "step": 955
+    },
+    {
+      "epoch": 2.4242424242424243,
+      "grad_norm": 1.2484320402145386,
+      "learning_rate": 1.7010661711603224e-05,
+      "loss": 0.218,
+      "step": 960
+    },
+    {
+      "epoch": 2.436868686868687,
+      "grad_norm": 1.0049769878387451,
+      "learning_rate": 1.688646063276064e-05,
+      "loss": 0.1978,
+      "step": 965
+    },
+    {
+      "epoch": 2.4494949494949494,
+      "grad_norm": 1.1801090240478516,
+      "learning_rate": 1.6762127998909933e-05,
+      "loss": 0.2106,
+      "step": 970
+    },
+    {
+      "epoch": 2.462121212121212,
+      "grad_norm": 1.0992910861968994,
+      "learning_rate": 1.6637672480563694e-05,
+      "loss": 0.2194,
+      "step": 975
+    },
+    {
+      "epoch": 2.474747474747475,
+      "grad_norm": 1.0028361082077026,
+      "learning_rate": 1.6513102756804025e-05,
+      "loss": 0.2475,
+      "step": 980
+    },
+    {
+      "epoch": 2.4873737373737375,
+      "grad_norm": 1.0803316831588745,
+      "learning_rate": 1.6388427514677315e-05,
+      "loss": 0.2189,
+      "step": 985
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.0498230457305908,
+      "learning_rate": 1.6263655448588417e-05,
+      "loss": 0.2219,
+      "step": 990
+    },
+    {
+      "epoch": 2.5126262626262625,
+      "grad_norm": 1.12675142288208,
+      "learning_rate": 1.613879525969435e-05,
+      "loss": 0.2049,
+      "step": 995
+    },
+    {
+      "epoch": 2.525252525252525,
+      "grad_norm": 1.0779552459716797,
+      "learning_rate": 1.6013855655297498e-05,
+      "loss": 0.2169,
+      "step": 1000
+    },
+    {
+      "epoch": 2.537878787878788,
+      "grad_norm": 1.267693281173706,
+      "learning_rate": 1.5888845348238388e-05,
+      "loss": 0.1932,
+      "step": 1005
+    },
+    {
+      "epoch": 2.5505050505050506,
+      "grad_norm": 1.1487102508544922,
+      "learning_rate": 1.5763773056288127e-05,
+      "loss": 0.1986,
+      "step": 1010
+    },
+    {
+      "epoch": 2.563131313131313,
+      "grad_norm": 1.0239760875701904,
+      "learning_rate": 1.5638647501540387e-05,
+      "loss": 0.2137,
+      "step": 1015
+    },
+    {
+      "epoch": 2.5757575757575757,
+      "grad_norm": 1.0446583032608032,
+      "learning_rate": 1.551347740980322e-05,
+      "loss": 0.2193,
+      "step": 1020
+    },
+    {
+      "epoch": 2.5883838383838382,
+      "grad_norm": 1.0630428791046143,
+      "learning_rate": 1.5388271509990532e-05,
+      "loss": 0.2113,
+      "step": 1025
+    },
+    {
+      "epoch": 2.601010101010101,
+      "grad_norm": 1.213524341583252,
+      "learning_rate": 1.5263038533513338e-05,
+      "loss": 0.2096,
+      "step": 1030
+    },
+    {
+      "epoch": 2.6136363636363638,
+      "grad_norm": 1.0546088218688965,
+      "learning_rate": 1.5137787213670899e-05,
+      "loss": 0.2017,
+      "step": 1035
+    },
+    {
+      "epoch": 2.6262626262626263,
+      "grad_norm": 1.0828937292099,
+      "learning_rate": 1.5012526285041662e-05,
+      "loss": 0.1966,
+      "step": 1040
+    },
+    {
+      "epoch": 2.638888888888889,
+      "grad_norm": 1.0778100490570068,
+      "learning_rate": 1.4887264482874173e-05,
+      "loss": 0.2025,
+      "step": 1045
+    },
+    {
+      "epoch": 2.6515151515151514,
+      "grad_norm": 1.1268481016159058,
+      "learning_rate": 1.4762010542477881e-05,
+      "loss": 0.2176,
+      "step": 1050
+    },
+    {
+      "epoch": 2.6641414141414144,
+      "grad_norm": 1.0860501527786255,
+      "learning_rate": 1.4636773198613994e-05,
+      "loss": 0.1815,
+      "step": 1055
+    },
+    {
+      "epoch": 2.676767676767677,
+      "grad_norm": 1.1054151058197021,
+      "learning_rate": 1.451156118488633e-05,
+      "loss": 0.2091,
+      "step": 1060
+    },
+    {
+      "epoch": 2.6893939393939394,
+      "grad_norm": 1.1649571657180786,
+      "learning_rate": 1.438638323313227e-05,
+      "loss": 0.2269,
+      "step": 1065
+    },
+    {
+      "epoch": 2.702020202020202,
+      "grad_norm": 1.3010584115982056,
+      "learning_rate": 1.4261248072813852e-05,
+      "loss": 0.1895,
+      "step": 1070
+    },
+    {
+      "epoch": 2.7146464646464645,
+      "grad_norm": 1.0134211778640747,
+      "learning_rate": 1.4136164430408979e-05,
+      "loss": 0.2275,
+      "step": 1075
+    },
+    {
+      "epoch": 2.7272727272727275,
+      "grad_norm": 1.1953684091567993,
+      "learning_rate": 1.4011141028802873e-05,
+      "loss": 0.1942,
+      "step": 1080
+    },
+    {
+      "epoch": 2.73989898989899,
+      "grad_norm": 1.1163067817687988,
+      "learning_rate": 1.3886186586679798e-05,
+      "loss": 0.1962,
+      "step": 1085
+    },
+    {
+      "epoch": 2.7525252525252526,
+      "grad_norm": 1.0360524654388428,
+      "learning_rate": 1.3761309817915017e-05,
+      "loss": 0.1854,
+      "step": 1090
+    },
+    {
+      "epoch": 2.765151515151515,
+      "grad_norm": 1.16696035861969,
+      "learning_rate": 1.3636519430967129e-05,
+      "loss": 0.1711,
+      "step": 1095
+    },
+    {
+      "epoch": 2.7777777777777777,
+      "grad_norm": 1.136378526687622,
+      "learning_rate": 1.351182412827079e-05,
+      "loss": 0.1741,
+      "step": 1100
+    },
+    {
+      "epoch": 2.7904040404040407,
+      "grad_norm": 1.100380539894104,
+      "learning_rate": 1.3387232605629804e-05,
+      "loss": 0.1977,
+      "step": 1105
+    },
+    {
+      "epoch": 2.8030303030303028,
+      "grad_norm": 1.1871534585952759,
+      "learning_rate": 1.326275355161073e-05,
+      "loss": 0.1906,
+      "step": 1110
+    },
+    {
+      "epoch": 2.8156565656565657,
+      "grad_norm": 1.1915061473846436,
+      "learning_rate": 1.3138395646936974e-05,
+      "loss": 0.1835,
+      "step": 1115
+    },
+    {
+      "epoch": 2.8282828282828283,
+      "grad_norm": 1.1094458103179932,
+      "learning_rate": 1.301416756388342e-05,
+      "loss": 0.1901,
+      "step": 1120
+    },
+    {
+      "epoch": 2.840909090909091,
+      "grad_norm": 1.1110031604766846,
+      "learning_rate": 1.289007796567165e-05,
+      "loss": 0.1731,
+      "step": 1125
+    },
+    {
+      "epoch": 2.8535353535353534,
+      "grad_norm": 1.219718098640442,
+      "learning_rate": 1.2766135505865808e-05,
+      "loss": 0.1704,
+      "step": 1130
+    },
+    {
+      "epoch": 2.866161616161616,
+      "grad_norm": 1.1409133672714233,
+      "learning_rate": 1.2642348827769152e-05,
+      "loss": 0.1875,
+      "step": 1135
+    },
+    {
+      "epoch": 2.878787878787879,
+      "grad_norm": 1.1201096773147583,
+      "learning_rate": 1.2518726563821253e-05,
+      "loss": 0.1799,
+      "step": 1140
+    },
+    {
+      "epoch": 2.8914141414141414,
+      "grad_norm": 1.0194494724273682,
+      "learning_rate": 1.2395277334996045e-05,
+      "loss": 0.1806,
+      "step": 1145
+    },
+    {
+      "epoch": 2.904040404040404,
+      "grad_norm": 1.1575123071670532,
+      "learning_rate": 1.2272009750200618e-05,
+      "loss": 0.1957,
+      "step": 1150
+    },
+    {
+      "epoch": 2.9166666666666665,
+      "grad_norm": 1.1770744323730469,
+      "learning_rate": 1.2148932405674843e-05,
+      "loss": 0.1747,
+      "step": 1155
+    },
+    {
+      "epoch": 2.929292929292929,
+      "grad_norm": 1.186741590499878,
+      "learning_rate": 1.2026053884391918e-05,
+      "loss": 0.1663,
+      "step": 1160
+    },
+    {
+      "epoch": 2.941919191919192,
+      "grad_norm": 1.1966912746429443,
+      "learning_rate": 1.1903382755459839e-05,
+      "loss": 0.1682,
+      "step": 1165
+    },
+    {
+      "epoch": 2.9545454545454546,
+      "grad_norm": 1.1320679187774658,
+      "learning_rate": 1.1780927573523776e-05,
+      "loss": 0.1939,
+      "step": 1170
+    },
+    {
+      "epoch": 2.967171717171717,
+      "grad_norm": 1.068954586982727,
+      "learning_rate": 1.1658696878169541e-05,
+      "loss": 0.1622,
+      "step": 1175
+    },
+    {
+      "epoch": 2.9797979797979797,
+      "grad_norm": 1.070233702659607,
+      "learning_rate": 1.1536699193328063e-05,
+      "loss": 0.1833,
+      "step": 1180
+    },
+    {
+      "epoch": 2.992424242424242,
+      "grad_norm": 1.156294345855713,
+      "learning_rate": 1.1414943026680939e-05,
+      "loss": 0.1666,
+      "step": 1185
+    },
+    {
+      "epoch": 3.005050505050505,
+      "grad_norm": 0.9665149450302124,
+      "learning_rate": 1.1293436869067157e-05,
+      "loss": 0.1346,
+      "step": 1190
+    },
+    {
+      "epoch": 3.0176767676767677,
+      "grad_norm": 1.2928872108459473,
+      "learning_rate": 1.1172189193890977e-05,
+      "loss": 0.1303,
+      "step": 1195
+    },
+    {
+      "epoch": 3.0303030303030303,
+      "grad_norm": 1.0999152660369873,
+      "learning_rate": 1.1051208456531016e-05,
+      "loss": 0.1241,
+      "step": 1200
+    },
+    {
+      "epoch": 3.042929292929293,
+      "grad_norm": 1.0285547971725464,
+      "learning_rate": 1.0930503093750584e-05,
+      "loss": 0.1332,
+      "step": 1205
+    },
+    {
+      "epoch": 3.0555555555555554,
+      "grad_norm": 1.0350072383880615,
+      "learning_rate": 1.0810081523109398e-05,
+      "loss": 0.1266,
+      "step": 1210
+    },
+    {
+      "epoch": 3.0681818181818183,
+      "grad_norm": 1.0541491508483887,
+      "learning_rate": 1.0689952142376485e-05,
+      "loss": 0.1171,
+      "step": 1215
+    },
+    {
+      "epoch": 3.080808080808081,
+      "grad_norm": 1.0109362602233887,
+      "learning_rate": 1.057012332894461e-05,
+      "loss": 0.1248,
+      "step": 1220
+    },
+    {
+      "epoch": 3.0934343434343434,
+      "grad_norm": 1.0019010305404663,
+      "learning_rate": 1.0450603439246063e-05,
+      "loss": 0.1219,
+      "step": 1225
+    },
+    {
+      "epoch": 3.106060606060606,
+      "grad_norm": 2.9002645015716553,
+      "learning_rate": 1.0331400808169879e-05,
+      "loss": 0.1256,
+      "step": 1230
+    },
+    {
+      "epoch": 3.1186868686868685,
+      "grad_norm": 1.0671870708465576,
+      "learning_rate": 1.021252374848062e-05,
+      "loss": 0.1325,
+      "step": 1235
+    },
+    {
+      "epoch": 3.1313131313131315,
+      "grad_norm": 1.0755785703659058,
+      "learning_rate": 1.0093980550238676e-05,
+      "loss": 0.1376,
+      "step": 1240
+    },
+    {
+      "epoch": 3.143939393939394,
+      "grad_norm": 1.0027779340744019,
+      "learning_rate": 9.975779480222124e-06,
+      "loss": 0.1297,
+      "step": 1245
+    },
+    {
+      "epoch": 3.1565656565656566,
+      "grad_norm": 0.9741926193237305,
+      "learning_rate": 9.857928781350242e-06,
+      "loss": 0.122,
+      "step": 1250
+    },
+    {
+      "epoch": 3.169191919191919,
+      "grad_norm": 0.9931155443191528,
+      "learning_rate": 9.740436672108686e-06,
+      "loss": 0.1398,
+      "step": 1255
+    },
+    {
+      "epoch": 3.1818181818181817,
+      "grad_norm": 1.0674378871917725,
+      "learning_rate": 9.623311345976355e-06,
+      "loss": 0.1217,
+      "step": 1260
+    },
+    {
+      "epoch": 3.1944444444444446,
+      "grad_norm": 0.9715051651000977,
+      "learning_rate": 9.506560970854011e-06,
+      "loss": 0.1155,
+      "step": 1265
+    },
+    {
+      "epoch": 3.207070707070707,
+      "grad_norm": 0.981411337852478,
+      "learning_rate": 9.390193688494657e-06,
+      "loss": 0.1289,
+      "step": 1270
+    },
+    {
+      "epoch": 3.2196969696969697,
+      "grad_norm": 1.0176677703857422,
+      "learning_rate": 9.274217613935806e-06,
+      "loss": 0.1288,
+      "step": 1275
+    },
+    {
+      "epoch": 3.2323232323232323,
+      "grad_norm": 1.0676851272583008,
+      "learning_rate": 9.158640834933519e-06,
+      "loss": 0.125,
+      "step": 1280
+    },
+    {
+      "epoch": 3.244949494949495,
+      "grad_norm": 0.998315155506134,
+      "learning_rate": 9.043471411398424e-06,
+      "loss": 0.1247,
+      "step": 1285
+    },
+    {
+      "epoch": 3.257575757575758,
+      "grad_norm": 1.1293725967407227,
+      "learning_rate": 8.928717374833639e-06,
+      "loss": 0.1334,
+      "step": 1290
+    },
+    {
+      "epoch": 3.2702020202020203,
+      "grad_norm": 0.914707601070404,
+      "learning_rate": 8.814386727774676e-06,
+      "loss": 0.1077,
+      "step": 1295
+    },
+    {
+      "epoch": 3.282828282828283,
+      "grad_norm": 1.154018521308899,
+      "learning_rate": 8.700487443231377e-06,
+      "loss": 0.1248,
+      "step": 1300
+    },
+    {
+      "epoch": 3.2954545454545454,
+      "grad_norm": 1.0889337062835693,
+      "learning_rate": 8.587027464131929e-06,
+      "loss": 0.1206,
+      "step": 1305
+    },
+    {
+      "epoch": 3.308080808080808,
+      "grad_norm": 0.9505720734596252,
+      "learning_rate": 8.474014702768905e-06,
+      "loss": 0.1109,
+      "step": 1310
+    },
+    {
+      "epoch": 3.320707070707071,
+      "grad_norm": 0.9936743378639221,
+      "learning_rate": 8.361457040247518e-06,
+      "loss": 0.118,
+      "step": 1315
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 1.0036802291870117,
+      "learning_rate": 8.249362325936036e-06,
+      "loss": 0.1095,
+      "step": 1320
+    },
+    {
+      "epoch": 3.345959595959596,
+      "grad_norm": 1.0567233562469482,
+      "learning_rate": 8.137738376918354e-06,
+      "loss": 0.1139,
+      "step": 1325
+    },
+    {
+      "epoch": 3.3585858585858586,
+      "grad_norm": 1.06975519657135,
+      "learning_rate": 8.026592977448898e-06,
+      "loss": 0.1173,
+      "step": 1330
+    },
+    {
+      "epoch": 3.371212121212121,
+      "grad_norm": 0.9553337097167969,
+      "learning_rate": 7.915933878409762e-06,
+      "loss": 0.1292,
+      "step": 1335
+    },
+    {
+      "epoch": 3.3838383838383836,
+      "grad_norm": 0.9045354127883911,
+      "learning_rate": 7.805768796770178e-06,
+      "loss": 0.1298,
+      "step": 1340
+    },
+    {
+      "epoch": 3.3964646464646466,
+      "grad_norm": 1.1572637557983398,
+      "learning_rate": 7.696105415048371e-06,
+      "loss": 0.1407,
+      "step": 1345
+    },
+    {
+      "epoch": 3.409090909090909,
+      "grad_norm": 0.9831897020339966,
+      "learning_rate": 7.5869513807758404e-06,
+      "loss": 0.1289,
+      "step": 1350
+    },
+    {
+      "epoch": 3.4217171717171717,
+      "grad_norm": 0.9966874718666077,
+      "learning_rate": 7.478314305963991e-06,
+      "loss": 0.108,
+      "step": 1355
+    },
+    {
+      "epoch": 3.4343434343434343,
+      "grad_norm": 0.921075165271759,
+      "learning_rate": 7.370201766573325e-06,
+      "loss": 0.1125,
+      "step": 1360
+    },
+    {
+      "epoch": 3.446969696969697,
+      "grad_norm": 0.8113449811935425,
+      "learning_rate": 7.262621301985145e-06,
+      "loss": 0.1034,
+      "step": 1365
+    },
+    {
+      "epoch": 3.45959595959596,
+      "grad_norm": 1.0315526723861694,
+      "learning_rate": 7.155580414475733e-06,
+      "loss": 0.1252,
+      "step": 1370
+    },
+    {
+      "epoch": 3.4722222222222223,
+      "grad_norm": 1.057673454284668,
+      "learning_rate": 7.049086568693231e-06,
+      "loss": 0.1193,
+      "step": 1375
+    },
+    {
+      "epoch": 3.484848484848485,
+      "grad_norm": 0.9585325121879578,
+      "learning_rate": 6.9431471911370205e-06,
+      "loss": 0.1092,
+      "step": 1380
+    },
+    {
+      "epoch": 3.4974747474747474,
+      "grad_norm": 0.9570455551147461,
+      "learning_rate": 6.837769669639877e-06,
+      "loss": 0.1216,
+      "step": 1385
+    },
+    {
+      "epoch": 3.51010101010101,
+      "grad_norm": 0.994683563709259,
+      "learning_rate": 6.732961352852728e-06,
+      "loss": 0.1286,
+      "step": 1390
+    },
+    {
+      "epoch": 3.5227272727272725,
+      "grad_norm": 0.8816766142845154,
+      "learning_rate": 6.628729549732217e-06,
+      "loss": 0.1143,
+      "step": 1395
+    },
+    {
+      "epoch": 3.5353535353535355,
+      "grad_norm": 0.836712121963501,
+      "learning_rate": 6.525081529030992e-06,
+      "loss": 0.11,
+      "step": 1400
+    },
+    {
+      "epoch": 3.547979797979798,
+      "grad_norm": 0.9519129395484924,
+      "learning_rate": 6.4220245187907915e-06,
+      "loss": 0.1165,
+      "step": 1405
+    },
+    {
+      "epoch": 3.5606060606060606,
+      "grad_norm": 1.030177116394043,
+      "learning_rate": 6.319565705838403e-06,
+      "loss": 0.11,
+      "step": 1410
+    },
+    {
+      "epoch": 3.573232323232323,
+      "grad_norm": 1.0223993062973022,
+      "learning_rate": 6.21771223528449e-06,
+      "loss": 0.1186,
+      "step": 1415
+    },
+    {
+      "epoch": 3.5858585858585856,
+      "grad_norm": 0.9433081746101379,
+      "learning_rate": 6.116471210025302e-06,
+      "loss": 0.1076,
+      "step": 1420
+    },
+    {
+      "epoch": 3.5984848484848486,
+      "grad_norm": 1.0501322746276855,
+      "learning_rate": 6.0158496902473325e-06,
+      "loss": 0.1099,
+      "step": 1425
+    },
+    {
+      "epoch": 3.611111111111111,
+      "grad_norm": 0.8067134618759155,
+      "learning_rate": 5.915854692935002e-06,
+      "loss": 0.1073,
+      "step": 1430
+    },
+    {
+      "epoch": 3.6237373737373737,
+      "grad_norm": 0.9593122005462646,
+      "learning_rate": 5.816493191381282e-06,
+      "loss": 0.117,
+      "step": 1435
+    },
+    {
+      "epoch": 3.6363636363636362,
+      "grad_norm": 1.2024868726730347,
+      "learning_rate": 5.717772114701424e-06,
+      "loss": 0.1049,
+      "step": 1440
+    },
+    {
+      "epoch": 3.648989898989899,
+      "grad_norm": 0.9438163042068481,
+      "learning_rate": 5.619698347349754e-06,
+      "loss": 0.1068,
+      "step": 1445
+    },
+    {
+      "epoch": 3.6616161616161618,
+      "grad_norm": 0.8761866688728333,
+      "learning_rate": 5.522278728639544e-06,
+      "loss": 0.1016,
+      "step": 1450
+    },
+    {
+      "epoch": 3.6742424242424243,
+      "grad_norm": 0.854404628276825,
+      "learning_rate": 5.4255200522660915e-06,
+      "loss": 0.1046,
+      "step": 1455
+    },
+    {
+      "epoch": 3.686868686868687,
+      "grad_norm": 1.1476513147354126,
+      "learning_rate": 5.329429065832944e-06,
+      "loss": 0.0954,
+      "step": 1460
+    },
+    {
+      "epoch": 3.6994949494949494,
+      "grad_norm": 0.8863788843154907,
+      "learning_rate": 5.234012470381349e-06,
+      "loss": 0.1242,
+      "step": 1465
+    },
+    {
+      "epoch": 3.712121212121212,
+      "grad_norm": 0.9078251719474792,
+      "learning_rate": 5.139276919922928e-06,
+      "loss": 0.0916,
+      "step": 1470
+    },
+    {
+      "epoch": 3.724747474747475,
+      "grad_norm": 0.9697856307029724,
+      "learning_rate": 5.045229020975681e-06,
+      "loss": 0.1047,
+      "step": 1475
+    },
+    {
+      "epoch": 3.7373737373737375,
+      "grad_norm": 0.8403934240341187,
+      "learning_rate": 4.951875332103237e-06,
+      "loss": 0.1018,
+      "step": 1480
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 1.1092280149459839,
+      "learning_rate": 4.859222363457512e-06,
+      "loss": 0.1281,
+      "step": 1485
+    },
+    {
+      "epoch": 3.7626262626262625,
+      "grad_norm": 0.8465602993965149,
+      "learning_rate": 4.767276576324706e-06,
+      "loss": 0.1037,
+      "step": 1490
+    },
+    {
+      "epoch": 3.775252525252525,
+      "grad_norm": 0.7505573630332947,
+      "learning_rate": 4.676044382674702e-06,
+      "loss": 0.0929,
+      "step": 1495
+    },
+    {
+      "epoch": 3.787878787878788,
+      "grad_norm": 1.048561930656433,
+      "learning_rate": 4.585532144713932e-06,
+      "loss": 0.1046,
+      "step": 1500
+    },
+    {
+      "epoch": 3.8005050505050506,
+      "grad_norm": 0.877079427242279,
+      "learning_rate": 4.495746174441701e-06,
+      "loss": 0.1032,
+      "step": 1505
+    },
+    {
+      "epoch": 3.813131313131313,
+      "grad_norm": 1.0151900053024292,
+      "learning_rate": 4.4066927332100145e-06,
+      "loss": 0.0987,
+      "step": 1510
+    },
+    {
+      "epoch": 3.8257575757575757,
+      "grad_norm": 0.8242208957672119,
+      "learning_rate": 4.318378031286907e-06,
+      "loss": 0.0955,
+      "step": 1515
+    },
+    {
+      "epoch": 3.8383838383838382,
+      "grad_norm": 1.2608996629714966,
+      "learning_rate": 4.2308082274233866e-06,
+      "loss": 0.0987,
+      "step": 1520
+    },
+    {
+      "epoch": 3.851010101010101,
+      "grad_norm": 0.9320261478424072,
+      "learning_rate": 4.1439894284239474e-06,
+      "loss": 0.0933,
+      "step": 1525
+    },
+    {
+      "epoch": 3.8636363636363638,
+      "grad_norm": 0.8348729610443115,
+      "learning_rate": 4.05792768872069e-06,
+      "loss": 0.0995,
+      "step": 1530
+    },
+    {
+      "epoch": 3.8762626262626263,
+      "grad_norm": 0.90358966588974,
+      "learning_rate": 3.972629009951101e-06,
+      "loss": 0.0942,
+      "step": 1535
+    },
+    {
+      "epoch": 3.888888888888889,
+      "grad_norm": 0.9079200029373169,
+      "learning_rate": 3.888099340539548e-06,
+      "loss": 0.1131,
+      "step": 1540
+    },
+    {
+      "epoch": 3.9015151515151514,
+      "grad_norm": 0.8452830910682678,
+      "learning_rate": 3.8043445752824213e-06,
+      "loss": 0.105,
+      "step": 1545
+    },
+    {
+      "epoch": 3.9141414141414144,
+      "grad_norm": 0.9456219673156738,
+      "learning_rate": 3.721370554937083e-06,
+      "loss": 0.1227,
+      "step": 1550
+    },
+    {
+      "epoch": 3.926767676767677,
+      "grad_norm": 0.8377255201339722,
+      "learning_rate": 3.639183065814543e-06,
+      "loss": 0.0905,
+      "step": 1555
+    },
+    {
+      "epoch": 3.9393939393939394,
+      "grad_norm": 0.8621814250946045,
+      "learning_rate": 3.557787839375934e-06,
+      "loss": 0.0973,
+      "step": 1560
+    },
+    {
+      "epoch": 3.952020202020202,
+      "grad_norm": 0.847291111946106,
+      "learning_rate": 3.4771905518328228e-06,
+      "loss": 0.0993,
+      "step": 1565
+    },
+    {
+      "epoch": 3.9646464646464645,
+      "grad_norm": 0.8391691446304321,
+      "learning_rate": 3.397396823751386e-06,
+      "loss": 0.0989,
+      "step": 1570
+    },
+    {
+      "epoch": 3.9772727272727275,
+      "grad_norm": 0.9375694394111633,
+      "learning_rate": 3.3184122196604403e-06,
+      "loss": 0.0881,
+      "step": 1575
+    },
+    {
+      "epoch": 3.98989898989899,
+      "grad_norm": 0.8996318578720093,
+      "learning_rate": 3.2402422476633818e-06,
+      "loss": 0.0773,
+      "step": 1580
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1980,
+  "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": 2.2473859611576238e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

31_128_e5_3e-5/checkpoint-1584/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1f7003266cf4b03470fdb9a31d0f745932e977fccbde9f17b310070d8bf4df35
+size 7736

31_128_e5_3e-5/checkpoint-1584/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)

31_128_e5_3e-5/checkpoint-1980/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

31_128_e5_3e-5/checkpoint-1980/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": [
+    "k_proj",
+    "q_proj",
+    "up_proj",
+    "o_proj",
+    "gate_proj",
+    "v_proj",
+    "down_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

31_128_e5_3e-5/checkpoint-1980/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1367a7a6698a88b3a8cabb91f1b710f4b208de5ce6ec098540342d760d0db468
+size 791751704

31_128_e5_3e-5/checkpoint-1980/latest

@@ -0,0 +1 @@
+global_step1980

31_128_e5_3e-5/checkpoint-1980/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:038812fe061e70ea30228e9e3ad7cbf881183bd8209300cfac502fd145f19cc4
+size 15920

31_128_e5_3e-5/checkpoint-1980/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c4ee0bb6ce203d0068eebbe2c0340ef63f66b6b8152ede680232186c0f99597a
+size 15920

31_128_e5_3e-5/checkpoint-1980/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:67db528b779518ff836784b5ef8a1e488e34809c6534c71316c9b6a7d110c7a1
+size 15920