Upload folder using huggingface_hub

9_128_e5_3e-5/checkpoint-1137/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

9_128_e5_3e-5/checkpoint-1137/adapter_config.json

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

9_128_e5_3e-5/checkpoint-1137/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a2d3626c095836987f0e04a0b946f403d8993ea72ed02deac03ac27f3194dd5d
+size 791751704

9_128_e5_3e-5/checkpoint-1137/latest

@@ -0,0 +1 @@
+global_step1137

9_128_e5_3e-5/checkpoint-1137/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f3b3af6ba3b74fe1f56150d6de471f7f324da246bfa2e44fbaad856df8127ced
+size 15920

9_128_e5_3e-5/checkpoint-1137/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:650335af2529d3e10cac812df7ba3da0972ae9eb4a7fe09f13a6cbb351148111
+size 15920

9_128_e5_3e-5/checkpoint-1137/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:428b47e285d2c083ac67c4e40b5a4b87a9abb24410595e3f9c7dc7816ca7a07e
+size 15920

9_128_e5_3e-5/checkpoint-1137/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fb9a2d8d061708c10362ca93cd18182afe1201b00ad444ee327910963b146da
+size 15920

9_128_e5_3e-5/checkpoint-1137/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:006f253d77168bb4183381590bd6461464fccebff5e58e37d61cfd8011ee3f8e
+size 15920

9_128_e5_3e-5/checkpoint-1137/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bf0902ceeb62ac37f06c5dd3272e7f0086f6d316f24b157c227708b325c24cbf
+size 15920

9_128_e5_3e-5/checkpoint-1137/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8fb2c480c59c062fe67aa341473a7a3c991b400599fdeec49190939f44bdbbb4
+size 15920

9_128_e5_3e-5/checkpoint-1137/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ae65767f1dd6aa5a261ff55da02bfc66dee54011c99881813ee9971d150c4f77
+size 15920

9_128_e5_3e-5/checkpoint-1137/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f0b2876e6a42a4ffc94a18691d8127eec02124ae368a715a9e06ba6c4b70567d
+size 1064

9_128_e5_3e-5/checkpoint-1137/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
+  }
+}

9_128_e5_3e-5/checkpoint-1137/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
+}

9_128_e5_3e-5/checkpoint-1137/trainer_state.json

@@ -0,0 +1,1623 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 1137,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.013192612137203167,
+      "grad_norm": 1.0549921989440918,
+      "learning_rate": 1.2631578947368422e-06,
+      "loss": 1.3195,
+      "step": 5
+    },
+    {
+      "epoch": 0.026385224274406333,
+      "grad_norm": 0.9549628496170044,
+      "learning_rate": 2.842105263157895e-06,
+      "loss": 1.3227,
+      "step": 10
+    },
+    {
+      "epoch": 0.0395778364116095,
+      "grad_norm": 0.7358924746513367,
+      "learning_rate": 4.421052631578947e-06,
+      "loss": 1.2851,
+      "step": 15
+    },
+    {
+      "epoch": 0.052770448548812667,
+      "grad_norm": 0.7462388277053833,
+      "learning_rate": 6e-06,
+      "loss": 1.2932,
+      "step": 20
+    },
+    {
+      "epoch": 0.06596306068601583,
+      "grad_norm": 0.6171491742134094,
+      "learning_rate": 7.578947368421053e-06,
+      "loss": 1.2672,
+      "step": 25
+    },
+    {
+      "epoch": 0.079155672823219,
+      "grad_norm": 0.5139086246490479,
+      "learning_rate": 9.157894736842105e-06,
+      "loss": 1.2624,
+      "step": 30
+    },
+    {
+      "epoch": 0.09234828496042216,
+      "grad_norm": 0.5310834646224976,
+      "learning_rate": 1.0736842105263158e-05,
+      "loss": 1.2188,
+      "step": 35
+    },
+    {
+      "epoch": 0.10554089709762533,
+      "grad_norm": 0.43072062730789185,
+      "learning_rate": 1.231578947368421e-05,
+      "loss": 1.1514,
+      "step": 40
+    },
+    {
+      "epoch": 0.11873350923482849,
+      "grad_norm": 0.4742017686367035,
+      "learning_rate": 1.3894736842105263e-05,
+      "loss": 1.1738,
+      "step": 45
+    },
+    {
+      "epoch": 0.13192612137203166,
+      "grad_norm": 0.5168235301971436,
+      "learning_rate": 1.547368421052632e-05,
+      "loss": 1.2002,
+      "step": 50
+    },
+    {
+      "epoch": 0.14511873350923482,
+      "grad_norm": 0.5260159969329834,
+      "learning_rate": 1.705263157894737e-05,
+      "loss": 1.1859,
+      "step": 55
+    },
+    {
+      "epoch": 0.158311345646438,
+      "grad_norm": 0.4556325078010559,
+      "learning_rate": 1.8631578947368424e-05,
+      "loss": 1.1207,
+      "step": 60
+    },
+    {
+      "epoch": 0.17150395778364116,
+      "grad_norm": 0.45433366298675537,
+      "learning_rate": 2.0210526315789475e-05,
+      "loss": 1.1734,
+      "step": 65
+    },
+    {
+      "epoch": 0.18469656992084432,
+      "grad_norm": 0.4594454765319824,
+      "learning_rate": 2.178947368421053e-05,
+      "loss": 1.158,
+      "step": 70
+    },
+    {
+      "epoch": 0.19788918205804748,
+      "grad_norm": 0.4666861295700073,
+      "learning_rate": 2.336842105263158e-05,
+      "loss": 1.0862,
+      "step": 75
+    },
+    {
+      "epoch": 0.21108179419525067,
+      "grad_norm": 0.49248138070106506,
+      "learning_rate": 2.4947368421052635e-05,
+      "loss": 1.1507,
+      "step": 80
+    },
+    {
+      "epoch": 0.22427440633245382,
+      "grad_norm": 0.5129343867301941,
+      "learning_rate": 2.6526315789473685e-05,
+      "loss": 1.1259,
+      "step": 85
+    },
+    {
+      "epoch": 0.23746701846965698,
+      "grad_norm": 0.5006796717643738,
+      "learning_rate": 2.810526315789474e-05,
+      "loss": 1.0959,
+      "step": 90
+    },
+    {
+      "epoch": 0.25065963060686014,
+      "grad_norm": 0.473619282245636,
+      "learning_rate": 2.968421052631579e-05,
+      "loss": 1.1076,
+      "step": 95
+    },
+    {
+      "epoch": 0.2638522427440633,
+      "grad_norm": 0.4560292661190033,
+      "learning_rate": 2.999963446058092e-05,
+      "loss": 1.1277,
+      "step": 100
+    },
+    {
+      "epoch": 0.2770448548812665,
+      "grad_norm": 0.5987527370452881,
+      "learning_rate": 2.999814948722491e-05,
+      "loss": 1.077,
+      "step": 105
+    },
+    {
+      "epoch": 0.29023746701846964,
+      "grad_norm": 0.5044912695884705,
+      "learning_rate": 2.999552234671775e-05,
+      "loss": 1.0728,
+      "step": 110
+    },
+    {
+      "epoch": 0.3034300791556728,
+      "grad_norm": 0.6184911131858826,
+      "learning_rate": 2.999175323912636e-05,
+      "loss": 1.0044,
+      "step": 115
+    },
+    {
+      "epoch": 0.316622691292876,
+      "grad_norm": 0.5510838627815247,
+      "learning_rate": 2.9986842451482876e-05,
+      "loss": 1.0165,
+      "step": 120
+    },
+    {
+      "epoch": 0.32981530343007914,
+      "grad_norm": 0.559280514717102,
+      "learning_rate": 2.9980790357762792e-05,
+      "loss": 1.0037,
+      "step": 125
+    },
+    {
+      "epoch": 0.34300791556728233,
+      "grad_norm": 0.6674004197120667,
+      "learning_rate": 2.9973597418856484e-05,
+      "loss": 1.0493,
+      "step": 130
+    },
+    {
+      "epoch": 0.3562005277044855,
+      "grad_norm": 0.5931965112686157,
+      "learning_rate": 2.996526418253408e-05,
+      "loss": 0.9745,
+      "step": 135
+    },
+    {
+      "epoch": 0.36939313984168864,
+      "grad_norm": 0.8309524655342102,
+      "learning_rate": 2.9955791283403805e-05,
+      "loss": 0.9851,
+      "step": 140
+    },
+    {
+      "epoch": 0.38258575197889183,
+      "grad_norm": 0.60931795835495,
+      "learning_rate": 2.9945179442863596e-05,
+      "loss": 1.0394,
+      "step": 145
+    },
+    {
+      "epoch": 0.39577836411609496,
+      "grad_norm": 0.7102614045143127,
+      "learning_rate": 2.9933429469046202e-05,
+      "loss": 0.9659,
+      "step": 150
+    },
+    {
+      "epoch": 0.40897097625329815,
+      "grad_norm": 0.6455872058868408,
+      "learning_rate": 2.9920542256757612e-05,
+      "loss": 0.9623,
+      "step": 155
+    },
+    {
+      "epoch": 0.42216358839050133,
+      "grad_norm": 0.6636059880256653,
+      "learning_rate": 2.9906518787408948e-05,
+      "loss": 0.9242,
+      "step": 160
+    },
+    {
+      "epoch": 0.43535620052770446,
+      "grad_norm": 0.6934926509857178,
+      "learning_rate": 2.9891360128941685e-05,
+      "loss": 0.9438,
+      "step": 165
+    },
+    {
+      "epoch": 0.44854881266490765,
+      "grad_norm": 0.6457786560058594,
+      "learning_rate": 2.9875067435746357e-05,
+      "loss": 0.974,
+      "step": 170
+    },
+    {
+      "epoch": 0.46174142480211083,
+      "grad_norm": 0.6471094489097595,
+      "learning_rate": 2.9857641948574636e-05,
+      "loss": 0.9771,
+      "step": 175
+    },
+    {
+      "epoch": 0.47493403693931396,
+      "grad_norm": 0.7950399518013,
+      "learning_rate": 2.983908499444483e-05,
+      "loss": 0.97,
+      "step": 180
+    },
+    {
+      "epoch": 0.48812664907651715,
+      "grad_norm": 0.7481000423431396,
+      "learning_rate": 2.981939798654084e-05,
+      "loss": 0.8588,
+      "step": 185
+    },
+    {
+      "epoch": 0.5013192612137203,
+      "grad_norm": 0.7145957946777344,
+      "learning_rate": 2.9798582424104542e-05,
+      "loss": 0.9219,
+      "step": 190
+    },
+    {
+      "epoch": 0.5145118733509235,
+      "grad_norm": 0.7868131399154663,
+      "learning_rate": 2.977663989232161e-05,
+      "loss": 0.9295,
+      "step": 195
+    },
+    {
+      "epoch": 0.5277044854881267,
+      "grad_norm": 0.790393590927124,
+      "learning_rate": 2.975357206220079e-05,
+      "loss": 0.8696,
+      "step": 200
+    },
+    {
+      "epoch": 0.5408970976253298,
+      "grad_norm": 0.756610095500946,
+      "learning_rate": 2.9729380690446658e-05,
+      "loss": 0.8478,
+      "step": 205
+    },
+    {
+      "epoch": 0.554089709762533,
+      "grad_norm": 1.0109566450119019,
+      "learning_rate": 2.9704067619325828e-05,
+      "loss": 0.8446,
+      "step": 210
+    },
+    {
+      "epoch": 0.5672823218997362,
+      "grad_norm": 0.7256650924682617,
+      "learning_rate": 2.967763477652668e-05,
+      "loss": 0.8314,
+      "step": 215
+    },
+    {
+      "epoch": 0.5804749340369393,
+      "grad_norm": 0.7625542283058167,
+      "learning_rate": 2.965008417501252e-05,
+      "loss": 0.831,
+      "step": 220
+    },
+    {
+      "epoch": 0.5936675461741425,
+      "grad_norm": 0.7654553651809692,
+      "learning_rate": 2.9621417912868326e-05,
+      "loss": 0.8307,
+      "step": 225
+    },
+    {
+      "epoch": 0.6068601583113457,
+      "grad_norm": 0.8893048167228699,
+      "learning_rate": 2.959163817314095e-05,
+      "loss": 0.8333,
+      "step": 230
+    },
+    {
+      "epoch": 0.6200527704485488,
+      "grad_norm": 0.8360992670059204,
+      "learning_rate": 2.956074722367286e-05,
+      "loss": 0.8838,
+      "step": 235
+    },
+    {
+      "epoch": 0.633245382585752,
+      "grad_norm": 0.780232310295105,
+      "learning_rate": 2.9528747416929467e-05,
+      "loss": 0.8506,
+      "step": 240
+    },
+    {
+      "epoch": 0.6464379947229552,
+      "grad_norm": 0.7864925265312195,
+      "learning_rate": 2.9495641189819943e-05,
+      "loss": 0.8311,
+      "step": 245
+    },
+    {
+      "epoch": 0.6596306068601583,
+      "grad_norm": 0.9330180287361145,
+      "learning_rate": 2.9461431063511652e-05,
+      "loss": 0.8233,
+      "step": 250
+    },
+    {
+      "epoch": 0.6728232189973615,
+      "grad_norm": 0.8171327114105225,
+      "learning_rate": 2.942611964323817e-05,
+      "loss": 0.8322,
+      "step": 255
+    },
+    {
+      "epoch": 0.6860158311345647,
+      "grad_norm": 0.7330072522163391,
+      "learning_rate": 2.9389709618100862e-05,
+      "loss": 0.7972,
+      "step": 260
+    },
+    {
+      "epoch": 0.6992084432717678,
+      "grad_norm": 0.8041938543319702,
+      "learning_rate": 2.9352203760864114e-05,
+      "loss": 0.7522,
+      "step": 265
+    },
+    {
+      "epoch": 0.712401055408971,
+      "grad_norm": 1.0043526887893677,
+      "learning_rate": 2.9313604927744153e-05,
+      "loss": 0.7906,
+      "step": 270
+    },
+    {
+      "epoch": 0.7255936675461742,
+      "grad_norm": 0.8837683796882629,
+      "learning_rate": 2.927391605819157e-05,
+      "loss": 0.8128,
+      "step": 275
+    },
+    {
+      "epoch": 0.7387862796833773,
+      "grad_norm": 0.8561952114105225,
+      "learning_rate": 2.923314017466745e-05,
+      "loss": 0.7538,
+      "step": 280
+    },
+    {
+      "epoch": 0.7519788918205804,
+      "grad_norm": 0.8416121006011963,
+      "learning_rate": 2.919128038241318e-05,
+      "loss": 0.7784,
+      "step": 285
+    },
+    {
+      "epoch": 0.7651715039577837,
+      "grad_norm": 0.8697754144668579,
+      "learning_rate": 2.9148339869214013e-05,
+      "loss": 0.7543,
+      "step": 290
+    },
+    {
+      "epoch": 0.7783641160949868,
+      "grad_norm": 0.8075074553489685,
+      "learning_rate": 2.9104321905156283e-05,
+      "loss": 0.7487,
+      "step": 295
+    },
+    {
+      "epoch": 0.7915567282321899,
+      "grad_norm": 0.9222568869590759,
+      "learning_rate": 2.9059229842378373e-05,
+      "loss": 0.7145,
+      "step": 300
+    },
+    {
+      "epoch": 0.8047493403693932,
+      "grad_norm": 0.9191045761108398,
+      "learning_rate": 2.9013067114815443e-05,
+      "loss": 0.7203,
+      "step": 305
+    },
+    {
+      "epoch": 0.8179419525065963,
+      "grad_norm": 0.9708392024040222,
+      "learning_rate": 2.8965837237937926e-05,
+      "loss": 0.7371,
+      "step": 310
+    },
+    {
+      "epoch": 0.8311345646437994,
+      "grad_norm": 0.9243896007537842,
+      "learning_rate": 2.89175438084838e-05,
+      "loss": 0.7589,
+      "step": 315
+    },
+    {
+      "epoch": 0.8443271767810027,
+      "grad_norm": 0.9135913252830505,
+      "learning_rate": 2.88681905041847e-05,
+      "loss": 0.7005,
+      "step": 320
+    },
+    {
+      "epoch": 0.8575197889182058,
+      "grad_norm": 0.9154238104820251,
+      "learning_rate": 2.8817781083485822e-05,
+      "loss": 0.6913,
+      "step": 325
+    },
+    {
+      "epoch": 0.8707124010554089,
+      "grad_norm": 0.9932613968849182,
+      "learning_rate": 2.8766319385259717e-05,
+      "loss": 0.7173,
+      "step": 330
+    },
+    {
+      "epoch": 0.8839050131926122,
+      "grad_norm": 0.8960279822349548,
+      "learning_rate": 2.8713809328513957e-05,
+      "loss": 0.6725,
+      "step": 335
+    },
+    {
+      "epoch": 0.8970976253298153,
+      "grad_norm": 1.1202912330627441,
+      "learning_rate": 2.8660254912092655e-05,
+      "loss": 0.6522,
+      "step": 340
+    },
+    {
+      "epoch": 0.9102902374670184,
+      "grad_norm": 0.903082549571991,
+      "learning_rate": 2.8605660214371974e-05,
+      "loss": 0.6815,
+      "step": 345
+    },
+    {
+      "epoch": 0.9234828496042217,
+      "grad_norm": 0.9124132394790649,
+      "learning_rate": 2.8550029392949516e-05,
+      "loss": 0.7262,
+      "step": 350
+    },
+    {
+      "epoch": 0.9366754617414248,
+      "grad_norm": 0.9651408195495605,
+      "learning_rate": 2.8493366684327704e-05,
+      "loss": 0.6348,
+      "step": 355
+    },
+    {
+      "epoch": 0.9498680738786279,
+      "grad_norm": 0.9096381664276123,
+      "learning_rate": 2.8435676403591193e-05,
+      "loss": 0.6787,
+      "step": 360
+    },
+    {
+      "epoch": 0.9630606860158312,
+      "grad_norm": 0.8840961456298828,
+      "learning_rate": 2.8376962944078212e-05,
+      "loss": 0.6426,
+      "step": 365
+    },
+    {
+      "epoch": 0.9762532981530343,
+      "grad_norm": 0.9949501156806946,
+      "learning_rate": 2.831723077704602e-05,
+      "loss": 0.6543,
+      "step": 370
+    },
+    {
+      "epoch": 0.9894459102902374,
+      "grad_norm": 0.9188775420188904,
+      "learning_rate": 2.8256484451330406e-05,
+      "loss": 0.6966,
+      "step": 375
+    },
+    {
+      "epoch": 1.0026385224274406,
+      "grad_norm": 0.9968132376670837,
+      "learning_rate": 2.8194728592999248e-05,
+      "loss": 0.6264,
+      "step": 380
+    },
+    {
+      "epoch": 1.0158311345646438,
+      "grad_norm": 0.9197973608970642,
+      "learning_rate": 2.8131967905000268e-05,
+      "loss": 0.5856,
+      "step": 385
+    },
+    {
+      "epoch": 1.029023746701847,
+      "grad_norm": 1.139959692955017,
+      "learning_rate": 2.8068207166802843e-05,
+      "loss": 0.6237,
+      "step": 390
+    },
+    {
+      "epoch": 1.04221635883905,
+      "grad_norm": 3.2350668907165527,
+      "learning_rate": 2.800345123403404e-05,
+      "loss": 0.5456,
+      "step": 395
+    },
+    {
+      "epoch": 1.0554089709762533,
+      "grad_norm": 1.071649432182312,
+      "learning_rate": 2.793770503810886e-05,
+      "loss": 0.5137,
+      "step": 400
+    },
+    {
+      "epoch": 1.0686015831134565,
+      "grad_norm": 1.060426115989685,
+      "learning_rate": 2.7870973585854672e-05,
+      "loss": 0.5075,
+      "step": 405
+    },
+    {
+      "epoch": 1.0817941952506596,
+      "grad_norm": 1.042592167854309,
+      "learning_rate": 2.780326195912991e-05,
+      "loss": 0.6077,
+      "step": 410
+    },
+    {
+      "epoch": 1.0949868073878628,
+      "grad_norm": 1.0673960447311401,
+      "learning_rate": 2.7734575314437124e-05,
+      "loss": 0.5346,
+      "step": 415
+    },
+    {
+      "epoch": 1.108179419525066,
+      "grad_norm": 1.0204036235809326,
+      "learning_rate": 2.7664918882530227e-05,
+      "loss": 0.5143,
+      "step": 420
+    },
+    {
+      "epoch": 1.121372031662269,
+      "grad_norm": 0.956579864025116,
+      "learning_rate": 2.75942979680162e-05,
+      "loss": 0.5328,
+      "step": 425
+    },
+    {
+      "epoch": 1.1345646437994723,
+      "grad_norm": 1.1918470859527588,
+      "learning_rate": 2.75227179489511e-05,
+      "loss": 0.5444,
+      "step": 430
+    },
+    {
+      "epoch": 1.1477572559366755,
+      "grad_norm": 1.1074199676513672,
+      "learning_rate": 2.7450184276430513e-05,
+      "loss": 0.5744,
+      "step": 435
+    },
+    {
+      "epoch": 1.1609498680738786,
+      "grad_norm": 1.095699429512024,
+      "learning_rate": 2.7376702474174428e-05,
+      "loss": 0.5006,
+      "step": 440
+    },
+    {
+      "epoch": 1.1741424802110818,
+      "grad_norm": 1.0307425260543823,
+      "learning_rate": 2.7302278138106585e-05,
+      "loss": 0.5001,
+      "step": 445
+    },
+    {
+      "epoch": 1.187335092348285,
+      "grad_norm": 1.0099748373031616,
+      "learning_rate": 2.7226916935928312e-05,
+      "loss": 0.5473,
+      "step": 450
+    },
+    {
+      "epoch": 1.200527704485488,
+      "grad_norm": 1.1488866806030273,
+      "learning_rate": 2.715062460668694e-05,
+      "loss": 0.5467,
+      "step": 455
+    },
+    {
+      "epoch": 1.2137203166226913,
+      "grad_norm": 1.0268123149871826,
+      "learning_rate": 2.7073406960338712e-05,
+      "loss": 0.5332,
+      "step": 460
+    },
+    {
+      "epoch": 1.2269129287598945,
+      "grad_norm": 1.2799384593963623,
+      "learning_rate": 2.699526987730636e-05,
+      "loss": 0.5192,
+      "step": 465
+    },
+    {
+      "epoch": 1.2401055408970976,
+      "grad_norm": 0.9568777084350586,
+      "learning_rate": 2.6916219308031273e-05,
+      "loss": 0.4968,
+      "step": 470
+    },
+    {
+      "epoch": 1.2532981530343008,
+      "grad_norm": 0.9864116311073303,
+      "learning_rate": 2.683626127252036e-05,
+      "loss": 0.4219,
+      "step": 475
+    },
+    {
+      "epoch": 1.266490765171504,
+      "grad_norm": 1.3926217555999756,
+      "learning_rate": 2.6755401859887598e-05,
+      "loss": 0.5213,
+      "step": 480
+    },
+    {
+      "epoch": 1.279683377308707,
+      "grad_norm": 1.126845121383667,
+      "learning_rate": 2.6673647227890315e-05,
+      "loss": 0.5119,
+      "step": 485
+    },
+    {
+      "epoch": 1.2928759894459103,
+      "grad_norm": 1.0567692518234253,
+      "learning_rate": 2.6591003602460266e-05,
+      "loss": 0.5134,
+      "step": 490
+    },
+    {
+      "epoch": 1.3060686015831133,
+      "grad_norm": 1.1358412504196167,
+      "learning_rate": 2.6507477277229496e-05,
+      "loss": 0.5154,
+      "step": 495
+    },
+    {
+      "epoch": 1.3192612137203166,
+      "grad_norm": 1.0255495309829712,
+      "learning_rate": 2.6423074613051053e-05,
+      "loss": 0.4393,
+      "step": 500
+    },
+    {
+      "epoch": 1.3324538258575198,
+      "grad_norm": 0.9631456732749939,
+      "learning_rate": 2.6337802037514594e-05,
+      "loss": 0.4987,
+      "step": 505
+    },
+    {
+      "epoch": 1.345646437994723,
+      "grad_norm": 1.2038588523864746,
+      "learning_rate": 2.6251666044456895e-05,
+      "loss": 0.4976,
+      "step": 510
+    },
+    {
+      "epoch": 1.358839050131926,
+      "grad_norm": 1.1599222421646118,
+      "learning_rate": 2.6164673193467312e-05,
+      "loss": 0.4918,
+      "step": 515
+    },
+    {
+      "epoch": 1.3720316622691293,
+      "grad_norm": 1.13398015499115,
+      "learning_rate": 2.607683010938826e-05,
+      "loss": 0.4322,
+      "step": 520
+    },
+    {
+      "epoch": 1.3852242744063323,
+      "grad_norm": 1.08286714553833,
+      "learning_rate": 2.598814348181068e-05,
+      "loss": 0.4819,
+      "step": 525
+    },
+    {
+      "epoch": 1.3984168865435356,
+      "grad_norm": 1.1023141145706177,
+      "learning_rate": 2.589862006456464e-05,
+      "loss": 0.5217,
+      "step": 530
+    },
+    {
+      "epoch": 1.4116094986807388,
+      "grad_norm": 1.166322112083435,
+      "learning_rate": 2.5808266675204957e-05,
+      "loss": 0.4647,
+      "step": 535
+    },
+    {
+      "epoch": 1.424802110817942,
+      "grad_norm": 1.1918513774871826,
+      "learning_rate": 2.571709019449205e-05,
+      "loss": 0.4617,
+      "step": 540
+    },
+    {
+      "epoch": 1.437994722955145,
+      "grad_norm": 0.9963548183441162,
+      "learning_rate": 2.5625097565867934e-05,
+      "loss": 0.4547,
+      "step": 545
+    },
+    {
+      "epoch": 1.4511873350923483,
+      "grad_norm": 1.0815931558609009,
+      "learning_rate": 2.5532295794927437e-05,
+      "loss": 0.4493,
+      "step": 550
+    },
+    {
+      "epoch": 1.4643799472295513,
+      "grad_norm": 1.0886744260787964,
+      "learning_rate": 2.5438691948884715e-05,
+      "loss": 0.4585,
+      "step": 555
+    },
+    {
+      "epoch": 1.4775725593667546,
+      "grad_norm": 1.1902480125427246,
+      "learning_rate": 2.5344293156035048e-05,
+      "loss": 0.446,
+      "step": 560
+    },
+    {
+      "epoch": 1.4907651715039578,
+      "grad_norm": 1.0291579961776733,
+      "learning_rate": 2.5249106605211988e-05,
+      "loss": 0.4684,
+      "step": 565
+    },
+    {
+      "epoch": 1.503957783641161,
+      "grad_norm": 1.0797675848007202,
+      "learning_rate": 2.515313954523991e-05,
+      "loss": 0.4296,
+      "step": 570
+    },
+    {
+      "epoch": 1.517150395778364,
+      "grad_norm": 1.0742638111114502,
+      "learning_rate": 2.5056399284381988e-05,
+      "loss": 0.4349,
+      "step": 575
+    },
+    {
+      "epoch": 1.5303430079155673,
+      "grad_norm": 1.0267268419265747,
+      "learning_rate": 2.4958893189783624e-05,
+      "loss": 0.4364,
+      "step": 580
+    },
+    {
+      "epoch": 1.5435356200527703,
+      "grad_norm": 1.2104828357696533,
+      "learning_rate": 2.486062868691144e-05,
+      "loss": 0.4447,
+      "step": 585
+    },
+    {
+      "epoch": 1.5567282321899736,
+      "grad_norm": 1.0328969955444336,
+      "learning_rate": 2.4761613258987763e-05,
+      "loss": 0.3914,
+      "step": 590
+    },
+    {
+      "epoch": 1.5699208443271768,
+      "grad_norm": 1.0258134603500366,
+      "learning_rate": 2.46618544464208e-05,
+      "loss": 0.4381,
+      "step": 595
+    },
+    {
+      "epoch": 1.58311345646438,
+      "grad_norm": 1.0421680212020874,
+      "learning_rate": 2.4561359846230346e-05,
+      "loss": 0.4314,
+      "step": 600
+    },
+    {
+      "epoch": 1.596306068601583,
+      "grad_norm": 1.0862973928451538,
+      "learning_rate": 2.4460137111469296e-05,
+      "loss": 0.3924,
+      "step": 605
+    },
+    {
+      "epoch": 1.6094986807387863,
+      "grad_norm": 1.1535123586654663,
+      "learning_rate": 2.4358193950640795e-05,
+      "loss": 0.4301,
+      "step": 610
+    },
+    {
+      "epoch": 1.6226912928759893,
+      "grad_norm": 1.0382863283157349,
+      "learning_rate": 2.425553812711123e-05,
+      "loss": 0.4189,
+      "step": 615
+    },
+    {
+      "epoch": 1.6358839050131926,
+      "grad_norm": 1.0676803588867188,
+      "learning_rate": 2.415217745851902e-05,
+      "loss": 0.4298,
+      "step": 620
+    },
+    {
+      "epoch": 1.6490765171503958,
+      "grad_norm": 0.9531147480010986,
+      "learning_rate": 2.4048119816179236e-05,
+      "loss": 0.4137,
+      "step": 625
+    },
+    {
+      "epoch": 1.662269129287599,
+      "grad_norm": 1.1409662961959839,
+      "learning_rate": 2.394337312448424e-05,
+      "loss": 0.4151,
+      "step": 630
+    },
+    {
+      "epoch": 1.675461741424802,
+      "grad_norm": 1.1769294738769531,
+      "learning_rate": 2.3837945360300132e-05,
+      "loss": 0.4161,
+      "step": 635
+    },
+    {
+      "epoch": 1.6886543535620053,
+      "grad_norm": 1.134934663772583,
+      "learning_rate": 2.3731844552359342e-05,
+      "loss": 0.3873,
+      "step": 640
+    },
+    {
+      "epoch": 1.7018469656992083,
+      "grad_norm": 1.1829636096954346,
+      "learning_rate": 2.362507878064918e-05,
+      "loss": 0.3887,
+      "step": 645
+    },
+    {
+      "epoch": 1.7150395778364116,
+      "grad_norm": 1.275728702545166,
+      "learning_rate": 2.351765617579652e-05,
+      "loss": 0.3842,
+      "step": 650
+    },
+    {
+      "epoch": 1.7282321899736148,
+      "grad_norm": 1.1850225925445557,
+      "learning_rate": 2.340958491844863e-05,
+      "loss": 0.4037,
+      "step": 655
+    },
+    {
+      "epoch": 1.741424802110818,
+      "grad_norm": 1.0635439157485962,
+      "learning_rate": 2.3300873238650163e-05,
+      "loss": 0.3702,
+      "step": 660
+    },
+    {
+      "epoch": 1.754617414248021,
+      "grad_norm": 1.2718092203140259,
+      "learning_rate": 2.3191529415216438e-05,
+      "loss": 0.4084,
+      "step": 665
+    },
+    {
+      "epoch": 1.767810026385224,
+      "grad_norm": 0.9605919122695923,
+      "learning_rate": 2.3081561775102946e-05,
+      "loss": 0.4055,
+      "step": 670
+    },
+    {
+      "epoch": 1.7810026385224274,
+      "grad_norm": 1.1679564714431763,
+      "learning_rate": 2.2970978692771243e-05,
+      "loss": 0.4023,
+      "step": 675
+    },
+    {
+      "epoch": 1.7941952506596306,
+      "grad_norm": 1.2143523693084717,
+      "learning_rate": 2.285978858955119e-05,
+      "loss": 0.386,
+      "step": 680
+    },
+    {
+      "epoch": 1.8073878627968338,
+      "grad_norm": 1.1865150928497314,
+      "learning_rate": 2.274799993299963e-05,
+      "loss": 0.3744,
+      "step": 685
+    },
+    {
+      "epoch": 1.820580474934037,
+      "grad_norm": 1.044783115386963,
+      "learning_rate": 2.263562123625557e-05,
+      "loss": 0.3635,
+      "step": 690
+    },
+    {
+      "epoch": 1.83377308707124,
+      "grad_norm": 1.1943895816802979,
+      "learning_rate": 2.2522661057391863e-05,
+      "loss": 0.3549,
+      "step": 695
+    },
+    {
+      "epoch": 1.8469656992084431,
+      "grad_norm": 1.1102036237716675,
+      "learning_rate": 2.2409127998763466e-05,
+      "loss": 0.3716,
+      "step": 700
+    },
+    {
+      "epoch": 1.8601583113456464,
+      "grad_norm": 0.9347606301307678,
+      "learning_rate": 2.2295030706352357e-05,
+      "loss": 0.3684,
+      "step": 705
+    },
+    {
+      "epoch": 1.8733509234828496,
+      "grad_norm": 1.1853909492492676,
+      "learning_rate": 2.2180377869109105e-05,
+      "loss": 0.3621,
+      "step": 710
+    },
+    {
+      "epoch": 1.8865435356200528,
+      "grad_norm": 1.0905243158340454,
+      "learning_rate": 2.206517821829115e-05,
+      "loss": 0.3604,
+      "step": 715
+    },
+    {
+      "epoch": 1.899736147757256,
+      "grad_norm": 1.1180747747421265,
+      "learning_rate": 2.1949440526797928e-05,
+      "loss": 0.3268,
+      "step": 720
+    },
+    {
+      "epoch": 1.912928759894459,
+      "grad_norm": 1.1372334957122803,
+      "learning_rate": 2.1833173608502736e-05,
+      "loss": 0.359,
+      "step": 725
+    },
+    {
+      "epoch": 1.9261213720316621,
+      "grad_norm": 1.0345664024353027,
+      "learning_rate": 2.1716386317581545e-05,
+      "loss": 0.3436,
+      "step": 730
+    },
+    {
+      "epoch": 1.9393139841688654,
+      "grad_norm": 1.2815654277801514,
+      "learning_rate": 2.1599087547838727e-05,
+      "loss": 0.3409,
+      "step": 735
+    },
+    {
+      "epoch": 1.9525065963060686,
+      "grad_norm": 1.0196033716201782,
+      "learning_rate": 2.1481286232029736e-05,
+      "loss": 0.3203,
+      "step": 740
+    },
+    {
+      "epoch": 1.9656992084432718,
+      "grad_norm": 1.1076465845108032,
+      "learning_rate": 2.136299134118085e-05,
+      "loss": 0.3728,
+      "step": 745
+    },
+    {
+      "epoch": 1.978891820580475,
+      "grad_norm": 1.0937527418136597,
+      "learning_rate": 2.124421188390602e-05,
+      "loss": 0.3403,
+      "step": 750
+    },
+    {
+      "epoch": 1.992084432717678,
+      "grad_norm": 1.189061164855957,
+      "learning_rate": 2.1124956905720775e-05,
+      "loss": 0.3388,
+      "step": 755
+    },
+    {
+      "epoch": 2.005277044854881,
+      "grad_norm": 0.986851155757904,
+      "learning_rate": 2.100523548835343e-05,
+      "loss": 0.3128,
+      "step": 760
+    },
+    {
+      "epoch": 2.0184696569920844,
+      "grad_norm": 1.0393791198730469,
+      "learning_rate": 2.088505674905342e-05,
+      "loss": 0.2541,
+      "step": 765
+    },
+    {
+      "epoch": 2.0316622691292876,
+      "grad_norm": 1.0699952840805054,
+      "learning_rate": 2.0764429839897054e-05,
+      "loss": 0.2468,
+      "step": 770
+    },
+    {
+      "epoch": 2.044854881266491,
+      "grad_norm": 1.1072733402252197,
+      "learning_rate": 2.0643363947090484e-05,
+      "loss": 0.2495,
+      "step": 775
+    },
+    {
+      "epoch": 2.058047493403694,
+      "grad_norm": 1.1739845275878906,
+      "learning_rate": 2.052186829027017e-05,
+      "loss": 0.2399,
+      "step": 780
+    },
+    {
+      "epoch": 2.0712401055408973,
+      "grad_norm": 1.26287043094635,
+      "learning_rate": 2.039995212180077e-05,
+      "loss": 0.2716,
+      "step": 785
+    },
+    {
+      "epoch": 2.0844327176781,
+      "grad_norm": 1.0228111743927002,
+      "learning_rate": 2.0277624726070526e-05,
+      "loss": 0.2364,
+      "step": 790
+    },
+    {
+      "epoch": 2.0976253298153034,
+      "grad_norm": 0.946446418762207,
+      "learning_rate": 2.0154895418784245e-05,
+      "loss": 0.2239,
+      "step": 795
+    },
+    {
+      "epoch": 2.1108179419525066,
+      "grad_norm": 1.168146014213562,
+      "learning_rate": 2.0031773546253828e-05,
+      "loss": 0.2387,
+      "step": 800
+    },
+    {
+      "epoch": 2.12401055408971,
+      "grad_norm": 0.9823072552680969,
+      "learning_rate": 1.990826848468656e-05,
+      "loss": 0.2641,
+      "step": 805
+    },
+    {
+      "epoch": 2.137203166226913,
+      "grad_norm": 1.0455418825149536,
+      "learning_rate": 1.978438963947105e-05,
+      "loss": 0.2553,
+      "step": 810
+    },
+    {
+      "epoch": 2.150395778364116,
+      "grad_norm": 1.1157386302947998,
+      "learning_rate": 1.9660146444460977e-05,
+      "loss": 0.2642,
+      "step": 815
+    },
+    {
+      "epoch": 2.163588390501319,
+      "grad_norm": 1.002679467201233,
+      "learning_rate": 1.953554836125667e-05,
+      "loss": 0.2428,
+      "step": 820
+    },
+    {
+      "epoch": 2.1767810026385224,
+      "grad_norm": 1.1035633087158203,
+      "learning_rate": 1.941060487848456e-05,
+      "loss": 0.268,
+      "step": 825
+    },
+    {
+      "epoch": 2.1899736147757256,
+      "grad_norm": 1.0666553974151611,
+      "learning_rate": 1.9285325511074603e-05,
+      "loss": 0.2449,
+      "step": 830
+    },
+    {
+      "epoch": 2.203166226912929,
+      "grad_norm": 1.1026744842529297,
+      "learning_rate": 1.915971979953567e-05,
+      "loss": 0.259,
+      "step": 835
+    },
+    {
+      "epoch": 2.216358839050132,
+      "grad_norm": 1.2465076446533203,
+      "learning_rate": 1.9033797309228984e-05,
+      "loss": 0.2419,
+      "step": 840
+    },
+    {
+      "epoch": 2.229551451187335,
+      "grad_norm": 1.2780468463897705,
+      "learning_rate": 1.8907567629639727e-05,
+      "loss": 0.2715,
+      "step": 845
+    },
+    {
+      "epoch": 2.242744063324538,
+      "grad_norm": 1.1911516189575195,
+      "learning_rate": 1.878104037364671e-05,
+      "loss": 0.242,
+      "step": 850
+    },
+    {
+      "epoch": 2.2559366754617414,
+      "grad_norm": 1.2382546663284302,
+      "learning_rate": 1.865422517679034e-05,
+      "loss": 0.254,
+      "step": 855
+    },
+    {
+      "epoch": 2.2691292875989446,
+      "grad_norm": 1.0796617269515991,
+      "learning_rate": 1.852713169653885e-05,
+      "loss": 0.2754,
+      "step": 860
+    },
+    {
+      "epoch": 2.282321899736148,
+      "grad_norm": 1.1549172401428223,
+      "learning_rate": 1.8399769611552826e-05,
+      "loss": 0.2182,
+      "step": 865
+    },
+    {
+      "epoch": 2.295514511873351,
+      "grad_norm": 1.3042552471160889,
+      "learning_rate": 1.8272148620948143e-05,
+      "loss": 0.224,
+      "step": 870
+    },
+    {
+      "epoch": 2.308707124010554,
+      "grad_norm": 1.0916672945022583,
+      "learning_rate": 1.814427844355733e-05,
+      "loss": 0.2595,
+      "step": 875
+    },
+    {
+      "epoch": 2.321899736147757,
+      "grad_norm": 1.3098633289337158,
+      "learning_rate": 1.8016168817189474e-05,
+      "loss": 0.254,
+      "step": 880
+    },
+    {
+      "epoch": 2.3350923482849604,
+      "grad_norm": 1.0905364751815796,
+      "learning_rate": 1.7887829497888614e-05,
+      "loss": 0.2563,
+      "step": 885
+    },
+    {
+      "epoch": 2.3482849604221636,
+      "grad_norm": 1.167807936668396,
+      "learning_rate": 1.7759270259190804e-05,
+      "loss": 0.2479,
+      "step": 890
+    },
+    {
+      "epoch": 2.361477572559367,
+      "grad_norm": 0.9638789296150208,
+      "learning_rate": 1.7630500891379808e-05,
+      "loss": 0.24,
+      "step": 895
+    },
+    {
+      "epoch": 2.37467018469657,
+      "grad_norm": 1.1495943069458008,
+      "learning_rate": 1.7501531200741536e-05,
+      "loss": 0.2517,
+      "step": 900
+    },
+    {
+      "epoch": 2.387862796833773,
+      "grad_norm": 1.1110082864761353,
+      "learning_rate": 1.7372371008817258e-05,
+      "loss": 0.2626,
+      "step": 905
+    },
+    {
+      "epoch": 2.401055408970976,
+      "grad_norm": 1.041497826576233,
+      "learning_rate": 1.7243030151655645e-05,
+      "loss": 0.2386,
+      "step": 910
+    },
+    {
+      "epoch": 2.4142480211081794,
+      "grad_norm": 1.18124258518219,
+      "learning_rate": 1.711351847906374e-05,
+      "loss": 0.2454,
+      "step": 915
+    },
+    {
+      "epoch": 2.4274406332453826,
+      "grad_norm": 1.1296191215515137,
+      "learning_rate": 1.698384585385684e-05,
+      "loss": 0.2559,
+      "step": 920
+    },
+    {
+      "epoch": 2.440633245382586,
+      "grad_norm": 1.4808402061462402,
+      "learning_rate": 1.685402215110739e-05,
+      "loss": 0.2431,
+      "step": 925
+    },
+    {
+      "epoch": 2.453825857519789,
+      "grad_norm": 1.17811119556427,
+      "learning_rate": 1.6724057257393e-05,
+      "loss": 0.229,
+      "step": 930
+    },
+    {
+      "epoch": 2.467018469656992,
+      "grad_norm": 1.0907628536224365,
+      "learning_rate": 1.65939610700435e-05,
+      "loss": 0.2103,
+      "step": 935
+    },
+    {
+      "epoch": 2.480211081794195,
+      "grad_norm": 1.1194047927856445,
+      "learning_rate": 1.6463743496387244e-05,
+      "loss": 0.2358,
+      "step": 940
+    },
+    {
+      "epoch": 2.4934036939313984,
+      "grad_norm": 1.0814589262008667,
+      "learning_rate": 1.6333414452996625e-05,
+      "loss": 0.2305,
+      "step": 945
+    },
+    {
+      "epoch": 2.5065963060686016,
+      "grad_norm": 1.0963690280914307,
+      "learning_rate": 1.6202983864932882e-05,
+      "loss": 0.2172,
+      "step": 950
+    },
+    {
+      "epoch": 2.519788918205805,
+      "grad_norm": 1.0521795749664307,
+      "learning_rate": 1.607246166499029e-05,
+      "loss": 0.2249,
+      "step": 955
+    },
+    {
+      "epoch": 2.532981530343008,
+      "grad_norm": 1.0296999216079712,
+      "learning_rate": 1.5941857792939702e-05,
+      "loss": 0.2218,
+      "step": 960
+    },
+    {
+      "epoch": 2.5461741424802113,
+      "grad_norm": 1.1036309003829956,
+      "learning_rate": 1.5811182194771634e-05,
+      "loss": 0.2198,
+      "step": 965
+    },
+    {
+      "epoch": 2.559366754617414,
+      "grad_norm": 1.0965986251831055,
+      "learning_rate": 1.5680444821938804e-05,
+      "loss": 0.2277,
+      "step": 970
+    },
+    {
+      "epoch": 2.5725593667546174,
+      "grad_norm": 1.2318143844604492,
+      "learning_rate": 1.5549655630598345e-05,
+      "loss": 0.2201,
+      "step": 975
+    },
+    {
+      "epoch": 2.5857519788918206,
+      "grad_norm": 1.1489981412887573,
+      "learning_rate": 1.5418824580853536e-05,
+      "loss": 0.1994,
+      "step": 980
+    },
+    {
+      "epoch": 2.598944591029024,
+      "grad_norm": 1.2300056219100952,
+      "learning_rate": 1.528796163599535e-05,
+      "loss": 0.2068,
+      "step": 985
+    },
+    {
+      "epoch": 2.6121372031662267,
+      "grad_norm": 1.00900399684906,
+      "learning_rate": 1.5157076761743687e-05,
+      "loss": 0.2159,
+      "step": 990
+    },
+    {
+      "epoch": 2.62532981530343,
+      "grad_norm": 1.1844401359558105,
+      "learning_rate": 1.5026179925488476e-05,
+      "loss": 0.1943,
+      "step": 995
+    },
+    {
+      "epoch": 2.638522427440633,
+      "grad_norm": 1.164178490638733,
+      "learning_rate": 1.4895281095530577e-05,
+      "loss": 0.1945,
+      "step": 1000
+    },
+    {
+      "epoch": 2.6517150395778364,
+      "grad_norm": 1.209371566772461,
+      "learning_rate": 1.4764390240322693e-05,
+      "loss": 0.2356,
+      "step": 1005
+    },
+    {
+      "epoch": 2.6649076517150396,
+      "grad_norm": 1.2932913303375244,
+      "learning_rate": 1.4633517327710205e-05,
+      "loss": 0.2101,
+      "step": 1010
+    },
+    {
+      "epoch": 2.678100263852243,
+      "grad_norm": 1.1144524812698364,
+      "learning_rate": 1.4502672324172109e-05,
+      "loss": 0.1993,
+      "step": 1015
+    },
+    {
+      "epoch": 2.691292875989446,
+      "grad_norm": 1.1121306419372559,
+      "learning_rate": 1.4371865194062009e-05,
+      "loss": 0.1895,
+      "step": 1020
+    },
+    {
+      "epoch": 2.7044854881266494,
+      "grad_norm": 1.119805097579956,
+      "learning_rate": 1.4241105898849302e-05,
+      "loss": 0.1794,
+      "step": 1025
+    },
+    {
+      "epoch": 2.717678100263852,
+      "grad_norm": 1.107885479927063,
+      "learning_rate": 1.4110404396360578e-05,
+      "loss": 0.2025,
+      "step": 1030
+    },
+    {
+      "epoch": 2.7308707124010554,
+      "grad_norm": 1.0021132230758667,
+      "learning_rate": 1.3979770640021284e-05,
+      "loss": 0.1958,
+      "step": 1035
+    },
+    {
+      "epoch": 2.7440633245382586,
+      "grad_norm": 1.1132313013076782,
+      "learning_rate": 1.3849214578097721e-05,
+      "loss": 0.2017,
+      "step": 1040
+    },
+    {
+      "epoch": 2.757255936675462,
+      "grad_norm": 1.1634668111801147,
+      "learning_rate": 1.371874615293949e-05,
+      "loss": 0.1921,
+      "step": 1045
+    },
+    {
+      "epoch": 2.7704485488126647,
+      "grad_norm": 0.9888617396354675,
+      "learning_rate": 1.3588375300222285e-05,
+      "loss": 0.1817,
+      "step": 1050
+    },
+    {
+      "epoch": 2.783641160949868,
+      "grad_norm": 1.1379151344299316,
+      "learning_rate": 1.34581119481913e-05,
+      "loss": 0.2046,
+      "step": 1055
+    },
+    {
+      "epoch": 2.796833773087071,
+      "grad_norm": 1.1285576820373535,
+      "learning_rate": 1.332796601690512e-05,
+      "loss": 0.1954,
+      "step": 1060
+    },
+    {
+      "epoch": 2.8100263852242744,
+      "grad_norm": 0.9891535639762878,
+      "learning_rate": 1.3197947417480295e-05,
+      "loss": 0.2002,
+      "step": 1065
+    },
+    {
+      "epoch": 2.8232189973614776,
+      "grad_norm": 1.0049059391021729,
+      "learning_rate": 1.3068066051336561e-05,
+      "loss": 0.1998,
+      "step": 1070
+    },
+    {
+      "epoch": 2.836411609498681,
+      "grad_norm": 1.0179656744003296,
+      "learning_rate": 1.2938331809442803e-05,
+      "loss": 0.2071,
+      "step": 1075
+    },
+    {
+      "epoch": 2.849604221635884,
+      "grad_norm": 1.2731091976165771,
+      "learning_rate": 1.2808754571563827e-05,
+      "loss": 0.1891,
+      "step": 1080
+    },
+    {
+      "epoch": 2.862796833773087,
+      "grad_norm": 1.1574561595916748,
+      "learning_rate": 1.2679344205507982e-05,
+      "loss": 0.1871,
+      "step": 1085
+    },
+    {
+      "epoch": 2.87598944591029,
+      "grad_norm": 0.9525778293609619,
+      "learning_rate": 1.2550110566375671e-05,
+      "loss": 0.1869,
+      "step": 1090
+    },
+    {
+      "epoch": 2.8891820580474934,
+      "grad_norm": 1.1481444835662842,
+      "learning_rate": 1.2421063495808856e-05,
+      "loss": 0.1925,
+      "step": 1095
+    },
+    {
+      "epoch": 2.9023746701846966,
+      "grad_norm": 1.1231403350830078,
+      "learning_rate": 1.2292212821241602e-05,
+      "loss": 0.1894,
+      "step": 1100
+    },
+    {
+      "epoch": 2.9155672823219,
+      "grad_norm": 1.1024245023727417,
+      "learning_rate": 1.2163568355151629e-05,
+      "loss": 0.1857,
+      "step": 1105
+    },
+    {
+      "epoch": 2.9287598944591027,
+      "grad_norm": 1.1011011600494385,
+      "learning_rate": 1.2035139894313108e-05,
+      "loss": 0.1598,
+      "step": 1110
+    },
+    {
+      "epoch": 2.941952506596306,
+      "grad_norm": 1.1969327926635742,
+      "learning_rate": 1.1906937219050558e-05,
+      "loss": 0.1654,
+      "step": 1115
+    },
+    {
+      "epoch": 2.955145118733509,
+      "grad_norm": 1.1062734127044678,
+      "learning_rate": 1.1778970092494051e-05,
+      "loss": 0.1543,
+      "step": 1120
+    },
+    {
+      "epoch": 2.9683377308707124,
+      "grad_norm": 1.1782795190811157,
+      "learning_rate": 1.1651248259835731e-05,
+      "loss": 0.1592,
+      "step": 1125
+    },
+    {
+      "epoch": 2.9815303430079156,
+      "grad_norm": 1.08965265750885,
+      "learning_rate": 1.1523781447587642e-05,
+      "loss": 0.1638,
+      "step": 1130
+    },
+    {
+      "epoch": 2.994722955145119,
+      "grad_norm": 0.9494940638542175,
+      "learning_rate": 1.1396579362841045e-05,
+      "loss": 0.1728,
+      "step": 1135
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1895,
+  "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": 1.6406420042012426e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

9_128_e5_3e-5/checkpoint-1137/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f0c9334f6fbff83eadc1fc0cb18b7b9f0220f80a8560293b3502a56c8ccec92
+size 7736

9_128_e5_3e-5/checkpoint-1137/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)

9_128_e5_3e-5/checkpoint-1516/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

9_128_e5_3e-5/checkpoint-1516/adapter_config.json

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

9_128_e5_3e-5/checkpoint-1516/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ba1699c3f0c85cddb0417cc9c4b1733bbe7136cecb954b2678400d1fa3739687
+size 791751704

9_128_e5_3e-5/checkpoint-1516/latest

@@ -0,0 +1 @@
+global_step1516

9_128_e5_3e-5/checkpoint-1516/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c28e6d760f5d928bca73cf3632977a35be81d6ba167649dfef99b696b3f2d818
+size 15920

9_128_e5_3e-5/checkpoint-1516/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eb1bd82e434a2d199477b388bb2a9e9543d25b34a807e0806a0cdb0e0d2f97e8
+size 15920

9_128_e5_3e-5/checkpoint-1516/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d550cd77dd2e99d1744dbb4e50579d1d09f6a0392fc2016132473b9d48eb2349
+size 15920

9_128_e5_3e-5/checkpoint-1516/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f78ce6dca1f6e82a4f254662038894ce7532cc38827da2b8328f71c7dd3cba4
+size 15920

9_128_e5_3e-5/checkpoint-1516/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:954f8d0d7fd96bf98b8ab275fc78dfdbae36557dd4bbe04942c691aec4d9c976
+size 15920

9_128_e5_3e-5/checkpoint-1516/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:39cbbe7adb84059c8756db7fac9fc11f70b4b1c27c58fb031ee4d20eba175d4f
+size 15920

9_128_e5_3e-5/checkpoint-1516/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:68083d60c0dcdeb316fc703fe7c8645657f7478876c9017b17f999b741596539
+size 15920

9_128_e5_3e-5/checkpoint-1516/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:31e197e3e08d55cd7a23337b0d15231dc57c231e23ccaab06bebf34dfc80f183
+size 15920

9_128_e5_3e-5/checkpoint-1516/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1c41dfc2877ea264ab34fc5e3393c9a872e0a4db231d3d32f51b8b30ed97fe3b
+size 1064

9_128_e5_3e-5/checkpoint-1516/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
+  }
+}

9_128_e5_3e-5/checkpoint-1516/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
+}

9_128_e5_3e-5/checkpoint-1516/trainer_state.json

@@ -0,0 +1,2155 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 1516,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.013192612137203167,
+      "grad_norm": 1.0549921989440918,
+      "learning_rate": 1.2631578947368422e-06,
+      "loss": 1.3195,
+      "step": 5
+    },
+    {
+      "epoch": 0.026385224274406333,
+      "grad_norm": 0.9549628496170044,
+      "learning_rate": 2.842105263157895e-06,
+      "loss": 1.3227,
+      "step": 10
+    },
+    {
+      "epoch": 0.0395778364116095,
+      "grad_norm": 0.7358924746513367,
+      "learning_rate": 4.421052631578947e-06,
+      "loss": 1.2851,
+      "step": 15
+    },
+    {
+      "epoch": 0.052770448548812667,
+      "grad_norm": 0.7462388277053833,
+      "learning_rate": 6e-06,
+      "loss": 1.2932,
+      "step": 20
+    },
+    {
+      "epoch": 0.06596306068601583,
+      "grad_norm": 0.6171491742134094,
+      "learning_rate": 7.578947368421053e-06,
+      "loss": 1.2672,
+      "step": 25
+    },
+    {
+      "epoch": 0.079155672823219,
+      "grad_norm": 0.5139086246490479,
+      "learning_rate": 9.157894736842105e-06,
+      "loss": 1.2624,
+      "step": 30
+    },
+    {
+      "epoch": 0.09234828496042216,
+      "grad_norm": 0.5310834646224976,
+      "learning_rate": 1.0736842105263158e-05,
+      "loss": 1.2188,
+      "step": 35
+    },
+    {
+      "epoch": 0.10554089709762533,
+      "grad_norm": 0.43072062730789185,
+      "learning_rate": 1.231578947368421e-05,
+      "loss": 1.1514,
+      "step": 40
+    },
+    {
+      "epoch": 0.11873350923482849,
+      "grad_norm": 0.4742017686367035,
+      "learning_rate": 1.3894736842105263e-05,
+      "loss": 1.1738,
+      "step": 45
+    },
+    {
+      "epoch": 0.13192612137203166,
+      "grad_norm": 0.5168235301971436,
+      "learning_rate": 1.547368421052632e-05,
+      "loss": 1.2002,
+      "step": 50
+    },
+    {
+      "epoch": 0.14511873350923482,
+      "grad_norm": 0.5260159969329834,
+      "learning_rate": 1.705263157894737e-05,
+      "loss": 1.1859,
+      "step": 55
+    },
+    {
+      "epoch": 0.158311345646438,
+      "grad_norm": 0.4556325078010559,
+      "learning_rate": 1.8631578947368424e-05,
+      "loss": 1.1207,
+      "step": 60
+    },
+    {
+      "epoch": 0.17150395778364116,
+      "grad_norm": 0.45433366298675537,
+      "learning_rate": 2.0210526315789475e-05,
+      "loss": 1.1734,
+      "step": 65
+    },
+    {
+      "epoch": 0.18469656992084432,
+      "grad_norm": 0.4594454765319824,
+      "learning_rate": 2.178947368421053e-05,
+      "loss": 1.158,
+      "step": 70
+    },
+    {
+      "epoch": 0.19788918205804748,
+      "grad_norm": 0.4666861295700073,
+      "learning_rate": 2.336842105263158e-05,
+      "loss": 1.0862,
+      "step": 75
+    },
+    {
+      "epoch": 0.21108179419525067,
+      "grad_norm": 0.49248138070106506,
+      "learning_rate": 2.4947368421052635e-05,
+      "loss": 1.1507,
+      "step": 80
+    },
+    {
+      "epoch": 0.22427440633245382,
+      "grad_norm": 0.5129343867301941,
+      "learning_rate": 2.6526315789473685e-05,
+      "loss": 1.1259,
+      "step": 85
+    },
+    {
+      "epoch": 0.23746701846965698,
+      "grad_norm": 0.5006796717643738,
+      "learning_rate": 2.810526315789474e-05,
+      "loss": 1.0959,
+      "step": 90
+    },
+    {
+      "epoch": 0.25065963060686014,
+      "grad_norm": 0.473619282245636,
+      "learning_rate": 2.968421052631579e-05,
+      "loss": 1.1076,
+      "step": 95
+    },
+    {
+      "epoch": 0.2638522427440633,
+      "grad_norm": 0.4560292661190033,
+      "learning_rate": 2.999963446058092e-05,
+      "loss": 1.1277,
+      "step": 100
+    },
+    {
+      "epoch": 0.2770448548812665,
+      "grad_norm": 0.5987527370452881,
+      "learning_rate": 2.999814948722491e-05,
+      "loss": 1.077,
+      "step": 105
+    },
+    {
+      "epoch": 0.29023746701846964,
+      "grad_norm": 0.5044912695884705,
+      "learning_rate": 2.999552234671775e-05,
+      "loss": 1.0728,
+      "step": 110
+    },
+    {
+      "epoch": 0.3034300791556728,
+      "grad_norm": 0.6184911131858826,
+      "learning_rate": 2.999175323912636e-05,
+      "loss": 1.0044,
+      "step": 115
+    },
+    {
+      "epoch": 0.316622691292876,
+      "grad_norm": 0.5510838627815247,
+      "learning_rate": 2.9986842451482876e-05,
+      "loss": 1.0165,
+      "step": 120
+    },
+    {
+      "epoch": 0.32981530343007914,
+      "grad_norm": 0.559280514717102,
+      "learning_rate": 2.9980790357762792e-05,
+      "loss": 1.0037,
+      "step": 125
+    },
+    {
+      "epoch": 0.34300791556728233,
+      "grad_norm": 0.6674004197120667,
+      "learning_rate": 2.9973597418856484e-05,
+      "loss": 1.0493,
+      "step": 130
+    },
+    {
+      "epoch": 0.3562005277044855,
+      "grad_norm": 0.5931965112686157,
+      "learning_rate": 2.996526418253408e-05,
+      "loss": 0.9745,
+      "step": 135
+    },
+    {
+      "epoch": 0.36939313984168864,
+      "grad_norm": 0.8309524655342102,
+      "learning_rate": 2.9955791283403805e-05,
+      "loss": 0.9851,
+      "step": 140
+    },
+    {
+      "epoch": 0.38258575197889183,
+      "grad_norm": 0.60931795835495,
+      "learning_rate": 2.9945179442863596e-05,
+      "loss": 1.0394,
+      "step": 145
+    },
+    {
+      "epoch": 0.39577836411609496,
+      "grad_norm": 0.7102614045143127,
+      "learning_rate": 2.9933429469046202e-05,
+      "loss": 0.9659,
+      "step": 150
+    },
+    {
+      "epoch": 0.40897097625329815,
+      "grad_norm": 0.6455872058868408,
+      "learning_rate": 2.9920542256757612e-05,
+      "loss": 0.9623,
+      "step": 155
+    },
+    {
+      "epoch": 0.42216358839050133,
+      "grad_norm": 0.6636059880256653,
+      "learning_rate": 2.9906518787408948e-05,
+      "loss": 0.9242,
+      "step": 160
+    },
+    {
+      "epoch": 0.43535620052770446,
+      "grad_norm": 0.6934926509857178,
+      "learning_rate": 2.9891360128941685e-05,
+      "loss": 0.9438,
+      "step": 165
+    },
+    {
+      "epoch": 0.44854881266490765,
+      "grad_norm": 0.6457786560058594,
+      "learning_rate": 2.9875067435746357e-05,
+      "loss": 0.974,
+      "step": 170
+    },
+    {
+      "epoch": 0.46174142480211083,
+      "grad_norm": 0.6471094489097595,
+      "learning_rate": 2.9857641948574636e-05,
+      "loss": 0.9771,
+      "step": 175
+    },
+    {
+      "epoch": 0.47493403693931396,
+      "grad_norm": 0.7950399518013,
+      "learning_rate": 2.983908499444483e-05,
+      "loss": 0.97,
+      "step": 180
+    },
+    {
+      "epoch": 0.48812664907651715,
+      "grad_norm": 0.7481000423431396,
+      "learning_rate": 2.981939798654084e-05,
+      "loss": 0.8588,
+      "step": 185
+    },
+    {
+      "epoch": 0.5013192612137203,
+      "grad_norm": 0.7145957946777344,
+      "learning_rate": 2.9798582424104542e-05,
+      "loss": 0.9219,
+      "step": 190
+    },
+    {
+      "epoch": 0.5145118733509235,
+      "grad_norm": 0.7868131399154663,
+      "learning_rate": 2.977663989232161e-05,
+      "loss": 0.9295,
+      "step": 195
+    },
+    {
+      "epoch": 0.5277044854881267,
+      "grad_norm": 0.790393590927124,
+      "learning_rate": 2.975357206220079e-05,
+      "loss": 0.8696,
+      "step": 200
+    },
+    {
+      "epoch": 0.5408970976253298,
+      "grad_norm": 0.756610095500946,
+      "learning_rate": 2.9729380690446658e-05,
+      "loss": 0.8478,
+      "step": 205
+    },
+    {
+      "epoch": 0.554089709762533,
+      "grad_norm": 1.0109566450119019,
+      "learning_rate": 2.9704067619325828e-05,
+      "loss": 0.8446,
+      "step": 210
+    },
+    {
+      "epoch": 0.5672823218997362,
+      "grad_norm": 0.7256650924682617,
+      "learning_rate": 2.967763477652668e-05,
+      "loss": 0.8314,
+      "step": 215
+    },
+    {
+      "epoch": 0.5804749340369393,
+      "grad_norm": 0.7625542283058167,
+      "learning_rate": 2.965008417501252e-05,
+      "loss": 0.831,
+      "step": 220
+    },
+    {
+      "epoch": 0.5936675461741425,
+      "grad_norm": 0.7654553651809692,
+      "learning_rate": 2.9621417912868326e-05,
+      "loss": 0.8307,
+      "step": 225
+    },
+    {
+      "epoch": 0.6068601583113457,
+      "grad_norm": 0.8893048167228699,
+      "learning_rate": 2.959163817314095e-05,
+      "loss": 0.8333,
+      "step": 230
+    },
+    {
+      "epoch": 0.6200527704485488,
+      "grad_norm": 0.8360992670059204,
+      "learning_rate": 2.956074722367286e-05,
+      "loss": 0.8838,
+      "step": 235
+    },
+    {
+      "epoch": 0.633245382585752,
+      "grad_norm": 0.780232310295105,
+      "learning_rate": 2.9528747416929467e-05,
+      "loss": 0.8506,
+      "step": 240
+    },
+    {
+      "epoch": 0.6464379947229552,
+      "grad_norm": 0.7864925265312195,
+      "learning_rate": 2.9495641189819943e-05,
+      "loss": 0.8311,
+      "step": 245
+    },
+    {
+      "epoch": 0.6596306068601583,
+      "grad_norm": 0.9330180287361145,
+      "learning_rate": 2.9461431063511652e-05,
+      "loss": 0.8233,
+      "step": 250
+    },
+    {
+      "epoch": 0.6728232189973615,
+      "grad_norm": 0.8171327114105225,
+      "learning_rate": 2.942611964323817e-05,
+      "loss": 0.8322,
+      "step": 255
+    },
+    {
+      "epoch": 0.6860158311345647,
+      "grad_norm": 0.7330072522163391,
+      "learning_rate": 2.9389709618100862e-05,
+      "loss": 0.7972,
+      "step": 260
+    },
+    {
+      "epoch": 0.6992084432717678,
+      "grad_norm": 0.8041938543319702,
+      "learning_rate": 2.9352203760864114e-05,
+      "loss": 0.7522,
+      "step": 265
+    },
+    {
+      "epoch": 0.712401055408971,
+      "grad_norm": 1.0043526887893677,
+      "learning_rate": 2.9313604927744153e-05,
+      "loss": 0.7906,
+      "step": 270
+    },
+    {
+      "epoch": 0.7255936675461742,
+      "grad_norm": 0.8837683796882629,
+      "learning_rate": 2.927391605819157e-05,
+      "loss": 0.8128,
+      "step": 275
+    },
+    {
+      "epoch": 0.7387862796833773,
+      "grad_norm": 0.8561952114105225,
+      "learning_rate": 2.923314017466745e-05,
+      "loss": 0.7538,
+      "step": 280
+    },
+    {
+      "epoch": 0.7519788918205804,
+      "grad_norm": 0.8416121006011963,
+      "learning_rate": 2.919128038241318e-05,
+      "loss": 0.7784,
+      "step": 285
+    },
+    {
+      "epoch": 0.7651715039577837,
+      "grad_norm": 0.8697754144668579,
+      "learning_rate": 2.9148339869214013e-05,
+      "loss": 0.7543,
+      "step": 290
+    },
+    {
+      "epoch": 0.7783641160949868,
+      "grad_norm": 0.8075074553489685,
+      "learning_rate": 2.9104321905156283e-05,
+      "loss": 0.7487,
+      "step": 295
+    },
+    {
+      "epoch": 0.7915567282321899,
+      "grad_norm": 0.9222568869590759,
+      "learning_rate": 2.9059229842378373e-05,
+      "loss": 0.7145,
+      "step": 300
+    },
+    {
+      "epoch": 0.8047493403693932,
+      "grad_norm": 0.9191045761108398,
+      "learning_rate": 2.9013067114815443e-05,
+      "loss": 0.7203,
+      "step": 305
+    },
+    {
+      "epoch": 0.8179419525065963,
+      "grad_norm": 0.9708392024040222,
+      "learning_rate": 2.8965837237937926e-05,
+      "loss": 0.7371,
+      "step": 310
+    },
+    {
+      "epoch": 0.8311345646437994,
+      "grad_norm": 0.9243896007537842,
+      "learning_rate": 2.89175438084838e-05,
+      "loss": 0.7589,
+      "step": 315
+    },
+    {
+      "epoch": 0.8443271767810027,
+      "grad_norm": 0.9135913252830505,
+      "learning_rate": 2.88681905041847e-05,
+      "loss": 0.7005,
+      "step": 320
+    },
+    {
+      "epoch": 0.8575197889182058,
+      "grad_norm": 0.9154238104820251,
+      "learning_rate": 2.8817781083485822e-05,
+      "loss": 0.6913,
+      "step": 325
+    },
+    {
+      "epoch": 0.8707124010554089,
+      "grad_norm": 0.9932613968849182,
+      "learning_rate": 2.8766319385259717e-05,
+      "loss": 0.7173,
+      "step": 330
+    },
+    {
+      "epoch": 0.8839050131926122,
+      "grad_norm": 0.8960279822349548,
+      "learning_rate": 2.8713809328513957e-05,
+      "loss": 0.6725,
+      "step": 335
+    },
+    {
+      "epoch": 0.8970976253298153,
+      "grad_norm": 1.1202912330627441,
+      "learning_rate": 2.8660254912092655e-05,
+      "loss": 0.6522,
+      "step": 340
+    },
+    {
+      "epoch": 0.9102902374670184,
+      "grad_norm": 0.903082549571991,
+      "learning_rate": 2.8605660214371974e-05,
+      "loss": 0.6815,
+      "step": 345
+    },
+    {
+      "epoch": 0.9234828496042217,
+      "grad_norm": 0.9124132394790649,
+      "learning_rate": 2.8550029392949516e-05,
+      "loss": 0.7262,
+      "step": 350
+    },
+    {
+      "epoch": 0.9366754617414248,
+      "grad_norm": 0.9651408195495605,
+      "learning_rate": 2.8493366684327704e-05,
+      "loss": 0.6348,
+      "step": 355
+    },
+    {
+      "epoch": 0.9498680738786279,
+      "grad_norm": 0.9096381664276123,
+      "learning_rate": 2.8435676403591193e-05,
+      "loss": 0.6787,
+      "step": 360
+    },
+    {
+      "epoch": 0.9630606860158312,
+      "grad_norm": 0.8840961456298828,
+      "learning_rate": 2.8376962944078212e-05,
+      "loss": 0.6426,
+      "step": 365
+    },
+    {
+      "epoch": 0.9762532981530343,
+      "grad_norm": 0.9949501156806946,
+      "learning_rate": 2.831723077704602e-05,
+      "loss": 0.6543,
+      "step": 370
+    },
+    {
+      "epoch": 0.9894459102902374,
+      "grad_norm": 0.9188775420188904,
+      "learning_rate": 2.8256484451330406e-05,
+      "loss": 0.6966,
+      "step": 375
+    },
+    {
+      "epoch": 1.0026385224274406,
+      "grad_norm": 0.9968132376670837,
+      "learning_rate": 2.8194728592999248e-05,
+      "loss": 0.6264,
+      "step": 380
+    },
+    {
+      "epoch": 1.0158311345646438,
+      "grad_norm": 0.9197973608970642,
+      "learning_rate": 2.8131967905000268e-05,
+      "loss": 0.5856,
+      "step": 385
+    },
+    {
+      "epoch": 1.029023746701847,
+      "grad_norm": 1.139959692955017,
+      "learning_rate": 2.8068207166802843e-05,
+      "loss": 0.6237,
+      "step": 390
+    },
+    {
+      "epoch": 1.04221635883905,
+      "grad_norm": 3.2350668907165527,
+      "learning_rate": 2.800345123403404e-05,
+      "loss": 0.5456,
+      "step": 395
+    },
+    {
+      "epoch": 1.0554089709762533,
+      "grad_norm": 1.071649432182312,
+      "learning_rate": 2.793770503810886e-05,
+      "loss": 0.5137,
+      "step": 400
+    },
+    {
+      "epoch": 1.0686015831134565,
+      "grad_norm": 1.060426115989685,
+      "learning_rate": 2.7870973585854672e-05,
+      "loss": 0.5075,
+      "step": 405
+    },
+    {
+      "epoch": 1.0817941952506596,
+      "grad_norm": 1.042592167854309,
+      "learning_rate": 2.780326195912991e-05,
+      "loss": 0.6077,
+      "step": 410
+    },
+    {
+      "epoch": 1.0949868073878628,
+      "grad_norm": 1.0673960447311401,
+      "learning_rate": 2.7734575314437124e-05,
+      "loss": 0.5346,
+      "step": 415
+    },
+    {
+      "epoch": 1.108179419525066,
+      "grad_norm": 1.0204036235809326,
+      "learning_rate": 2.7664918882530227e-05,
+      "loss": 0.5143,
+      "step": 420
+    },
+    {
+      "epoch": 1.121372031662269,
+      "grad_norm": 0.956579864025116,
+      "learning_rate": 2.75942979680162e-05,
+      "loss": 0.5328,
+      "step": 425
+    },
+    {
+      "epoch": 1.1345646437994723,
+      "grad_norm": 1.1918470859527588,
+      "learning_rate": 2.75227179489511e-05,
+      "loss": 0.5444,
+      "step": 430
+    },
+    {
+      "epoch": 1.1477572559366755,
+      "grad_norm": 1.1074199676513672,
+      "learning_rate": 2.7450184276430513e-05,
+      "loss": 0.5744,
+      "step": 435
+    },
+    {
+      "epoch": 1.1609498680738786,
+      "grad_norm": 1.095699429512024,
+      "learning_rate": 2.7376702474174428e-05,
+      "loss": 0.5006,
+      "step": 440
+    },
+    {
+      "epoch": 1.1741424802110818,
+      "grad_norm": 1.0307425260543823,
+      "learning_rate": 2.7302278138106585e-05,
+      "loss": 0.5001,
+      "step": 445
+    },
+    {
+      "epoch": 1.187335092348285,
+      "grad_norm": 1.0099748373031616,
+      "learning_rate": 2.7226916935928312e-05,
+      "loss": 0.5473,
+      "step": 450
+    },
+    {
+      "epoch": 1.200527704485488,
+      "grad_norm": 1.1488866806030273,
+      "learning_rate": 2.715062460668694e-05,
+      "loss": 0.5467,
+      "step": 455
+    },
+    {
+      "epoch": 1.2137203166226913,
+      "grad_norm": 1.0268123149871826,
+      "learning_rate": 2.7073406960338712e-05,
+      "loss": 0.5332,
+      "step": 460
+    },
+    {
+      "epoch": 1.2269129287598945,
+      "grad_norm": 1.2799384593963623,
+      "learning_rate": 2.699526987730636e-05,
+      "loss": 0.5192,
+      "step": 465
+    },
+    {
+      "epoch": 1.2401055408970976,
+      "grad_norm": 0.9568777084350586,
+      "learning_rate": 2.6916219308031273e-05,
+      "loss": 0.4968,
+      "step": 470
+    },
+    {
+      "epoch": 1.2532981530343008,
+      "grad_norm": 0.9864116311073303,
+      "learning_rate": 2.683626127252036e-05,
+      "loss": 0.4219,
+      "step": 475
+    },
+    {
+      "epoch": 1.266490765171504,
+      "grad_norm": 1.3926217555999756,
+      "learning_rate": 2.6755401859887598e-05,
+      "loss": 0.5213,
+      "step": 480
+    },
+    {
+      "epoch": 1.279683377308707,
+      "grad_norm": 1.126845121383667,
+      "learning_rate": 2.6673647227890315e-05,
+      "loss": 0.5119,
+      "step": 485
+    },
+    {
+      "epoch": 1.2928759894459103,
+      "grad_norm": 1.0567692518234253,
+      "learning_rate": 2.6591003602460266e-05,
+      "loss": 0.5134,
+      "step": 490
+    },
+    {
+      "epoch": 1.3060686015831133,
+      "grad_norm": 1.1358412504196167,
+      "learning_rate": 2.6507477277229496e-05,
+      "loss": 0.5154,
+      "step": 495
+    },
+    {
+      "epoch": 1.3192612137203166,
+      "grad_norm": 1.0255495309829712,
+      "learning_rate": 2.6423074613051053e-05,
+      "loss": 0.4393,
+      "step": 500
+    },
+    {
+      "epoch": 1.3324538258575198,
+      "grad_norm": 0.9631456732749939,
+      "learning_rate": 2.6337802037514594e-05,
+      "loss": 0.4987,
+      "step": 505
+    },
+    {
+      "epoch": 1.345646437994723,
+      "grad_norm": 1.2038588523864746,
+      "learning_rate": 2.6251666044456895e-05,
+      "loss": 0.4976,
+      "step": 510
+    },
+    {
+      "epoch": 1.358839050131926,
+      "grad_norm": 1.1599222421646118,
+      "learning_rate": 2.6164673193467312e-05,
+      "loss": 0.4918,
+      "step": 515
+    },
+    {
+      "epoch": 1.3720316622691293,
+      "grad_norm": 1.13398015499115,
+      "learning_rate": 2.607683010938826e-05,
+      "loss": 0.4322,
+      "step": 520
+    },
+    {
+      "epoch": 1.3852242744063323,
+      "grad_norm": 1.08286714553833,
+      "learning_rate": 2.598814348181068e-05,
+      "loss": 0.4819,
+      "step": 525
+    },
+    {
+      "epoch": 1.3984168865435356,
+      "grad_norm": 1.1023141145706177,
+      "learning_rate": 2.589862006456464e-05,
+      "loss": 0.5217,
+      "step": 530
+    },
+    {
+      "epoch": 1.4116094986807388,
+      "grad_norm": 1.166322112083435,
+      "learning_rate": 2.5808266675204957e-05,
+      "loss": 0.4647,
+      "step": 535
+    },
+    {
+      "epoch": 1.424802110817942,
+      "grad_norm": 1.1918513774871826,
+      "learning_rate": 2.571709019449205e-05,
+      "loss": 0.4617,
+      "step": 540
+    },
+    {
+      "epoch": 1.437994722955145,
+      "grad_norm": 0.9963548183441162,
+      "learning_rate": 2.5625097565867934e-05,
+      "loss": 0.4547,
+      "step": 545
+    },
+    {
+      "epoch": 1.4511873350923483,
+      "grad_norm": 1.0815931558609009,
+      "learning_rate": 2.5532295794927437e-05,
+      "loss": 0.4493,
+      "step": 550
+    },
+    {
+      "epoch": 1.4643799472295513,
+      "grad_norm": 1.0886744260787964,
+      "learning_rate": 2.5438691948884715e-05,
+      "loss": 0.4585,
+      "step": 555
+    },
+    {
+      "epoch": 1.4775725593667546,
+      "grad_norm": 1.1902480125427246,
+      "learning_rate": 2.5344293156035048e-05,
+      "loss": 0.446,
+      "step": 560
+    },
+    {
+      "epoch": 1.4907651715039578,
+      "grad_norm": 1.0291579961776733,
+      "learning_rate": 2.5249106605211988e-05,
+      "loss": 0.4684,
+      "step": 565
+    },
+    {
+      "epoch": 1.503957783641161,
+      "grad_norm": 1.0797675848007202,
+      "learning_rate": 2.515313954523991e-05,
+      "loss": 0.4296,
+      "step": 570
+    },
+    {
+      "epoch": 1.517150395778364,
+      "grad_norm": 1.0742638111114502,
+      "learning_rate": 2.5056399284381988e-05,
+      "loss": 0.4349,
+      "step": 575
+    },
+    {
+      "epoch": 1.5303430079155673,
+      "grad_norm": 1.0267268419265747,
+      "learning_rate": 2.4958893189783624e-05,
+      "loss": 0.4364,
+      "step": 580
+    },
+    {
+      "epoch": 1.5435356200527703,
+      "grad_norm": 1.2104828357696533,
+      "learning_rate": 2.486062868691144e-05,
+      "loss": 0.4447,
+      "step": 585
+    },
+    {
+      "epoch": 1.5567282321899736,
+      "grad_norm": 1.0328969955444336,
+      "learning_rate": 2.4761613258987763e-05,
+      "loss": 0.3914,
+      "step": 590
+    },
+    {
+      "epoch": 1.5699208443271768,
+      "grad_norm": 1.0258134603500366,
+      "learning_rate": 2.46618544464208e-05,
+      "loss": 0.4381,
+      "step": 595
+    },
+    {
+      "epoch": 1.58311345646438,
+      "grad_norm": 1.0421680212020874,
+      "learning_rate": 2.4561359846230346e-05,
+      "loss": 0.4314,
+      "step": 600
+    },
+    {
+      "epoch": 1.596306068601583,
+      "grad_norm": 1.0862973928451538,
+      "learning_rate": 2.4460137111469296e-05,
+      "loss": 0.3924,
+      "step": 605
+    },
+    {
+      "epoch": 1.6094986807387863,
+      "grad_norm": 1.1535123586654663,
+      "learning_rate": 2.4358193950640795e-05,
+      "loss": 0.4301,
+      "step": 610
+    },
+    {
+      "epoch": 1.6226912928759893,
+      "grad_norm": 1.0382863283157349,
+      "learning_rate": 2.425553812711123e-05,
+      "loss": 0.4189,
+      "step": 615
+    },
+    {
+      "epoch": 1.6358839050131926,
+      "grad_norm": 1.0676803588867188,
+      "learning_rate": 2.415217745851902e-05,
+      "loss": 0.4298,
+      "step": 620
+    },
+    {
+      "epoch": 1.6490765171503958,
+      "grad_norm": 0.9531147480010986,
+      "learning_rate": 2.4048119816179236e-05,
+      "loss": 0.4137,
+      "step": 625
+    },
+    {
+      "epoch": 1.662269129287599,
+      "grad_norm": 1.1409662961959839,
+      "learning_rate": 2.394337312448424e-05,
+      "loss": 0.4151,
+      "step": 630
+    },
+    {
+      "epoch": 1.675461741424802,
+      "grad_norm": 1.1769294738769531,
+      "learning_rate": 2.3837945360300132e-05,
+      "loss": 0.4161,
+      "step": 635
+    },
+    {
+      "epoch": 1.6886543535620053,
+      "grad_norm": 1.134934663772583,
+      "learning_rate": 2.3731844552359342e-05,
+      "loss": 0.3873,
+      "step": 640
+    },
+    {
+      "epoch": 1.7018469656992083,
+      "grad_norm": 1.1829636096954346,
+      "learning_rate": 2.362507878064918e-05,
+      "loss": 0.3887,
+      "step": 645
+    },
+    {
+      "epoch": 1.7150395778364116,
+      "grad_norm": 1.275728702545166,
+      "learning_rate": 2.351765617579652e-05,
+      "loss": 0.3842,
+      "step": 650
+    },
+    {
+      "epoch": 1.7282321899736148,
+      "grad_norm": 1.1850225925445557,
+      "learning_rate": 2.340958491844863e-05,
+      "loss": 0.4037,
+      "step": 655
+    },
+    {
+      "epoch": 1.741424802110818,
+      "grad_norm": 1.0635439157485962,
+      "learning_rate": 2.3300873238650163e-05,
+      "loss": 0.3702,
+      "step": 660
+    },
+    {
+      "epoch": 1.754617414248021,
+      "grad_norm": 1.2718092203140259,
+      "learning_rate": 2.3191529415216438e-05,
+      "loss": 0.4084,
+      "step": 665
+    },
+    {
+      "epoch": 1.767810026385224,
+      "grad_norm": 0.9605919122695923,
+      "learning_rate": 2.3081561775102946e-05,
+      "loss": 0.4055,
+      "step": 670
+    },
+    {
+      "epoch": 1.7810026385224274,
+      "grad_norm": 1.1679564714431763,
+      "learning_rate": 2.2970978692771243e-05,
+      "loss": 0.4023,
+      "step": 675
+    },
+    {
+      "epoch": 1.7941952506596306,
+      "grad_norm": 1.2143523693084717,
+      "learning_rate": 2.285978858955119e-05,
+      "loss": 0.386,
+      "step": 680
+    },
+    {
+      "epoch": 1.8073878627968338,
+      "grad_norm": 1.1865150928497314,
+      "learning_rate": 2.274799993299963e-05,
+      "loss": 0.3744,
+      "step": 685
+    },
+    {
+      "epoch": 1.820580474934037,
+      "grad_norm": 1.044783115386963,
+      "learning_rate": 2.263562123625557e-05,
+      "loss": 0.3635,
+      "step": 690
+    },
+    {
+      "epoch": 1.83377308707124,
+      "grad_norm": 1.1943895816802979,
+      "learning_rate": 2.2522661057391863e-05,
+      "loss": 0.3549,
+      "step": 695
+    },
+    {
+      "epoch": 1.8469656992084431,
+      "grad_norm": 1.1102036237716675,
+      "learning_rate": 2.2409127998763466e-05,
+      "loss": 0.3716,
+      "step": 700
+    },
+    {
+      "epoch": 1.8601583113456464,
+      "grad_norm": 0.9347606301307678,
+      "learning_rate": 2.2295030706352357e-05,
+      "loss": 0.3684,
+      "step": 705
+    },
+    {
+      "epoch": 1.8733509234828496,
+      "grad_norm": 1.1853909492492676,
+      "learning_rate": 2.2180377869109105e-05,
+      "loss": 0.3621,
+      "step": 710
+    },
+    {
+      "epoch": 1.8865435356200528,
+      "grad_norm": 1.0905243158340454,
+      "learning_rate": 2.206517821829115e-05,
+      "loss": 0.3604,
+      "step": 715
+    },
+    {
+      "epoch": 1.899736147757256,
+      "grad_norm": 1.1180747747421265,
+      "learning_rate": 2.1949440526797928e-05,
+      "loss": 0.3268,
+      "step": 720
+    },
+    {
+      "epoch": 1.912928759894459,
+      "grad_norm": 1.1372334957122803,
+      "learning_rate": 2.1833173608502736e-05,
+      "loss": 0.359,
+      "step": 725
+    },
+    {
+      "epoch": 1.9261213720316621,
+      "grad_norm": 1.0345664024353027,
+      "learning_rate": 2.1716386317581545e-05,
+      "loss": 0.3436,
+      "step": 730
+    },
+    {
+      "epoch": 1.9393139841688654,
+      "grad_norm": 1.2815654277801514,
+      "learning_rate": 2.1599087547838727e-05,
+      "loss": 0.3409,
+      "step": 735
+    },
+    {
+      "epoch": 1.9525065963060686,
+      "grad_norm": 1.0196033716201782,
+      "learning_rate": 2.1481286232029736e-05,
+      "loss": 0.3203,
+      "step": 740
+    },
+    {
+      "epoch": 1.9656992084432718,
+      "grad_norm": 1.1076465845108032,
+      "learning_rate": 2.136299134118085e-05,
+      "loss": 0.3728,
+      "step": 745
+    },
+    {
+      "epoch": 1.978891820580475,
+      "grad_norm": 1.0937527418136597,
+      "learning_rate": 2.124421188390602e-05,
+      "loss": 0.3403,
+      "step": 750
+    },
+    {
+      "epoch": 1.992084432717678,
+      "grad_norm": 1.189061164855957,
+      "learning_rate": 2.1124956905720775e-05,
+      "loss": 0.3388,
+      "step": 755
+    },
+    {
+      "epoch": 2.005277044854881,
+      "grad_norm": 0.986851155757904,
+      "learning_rate": 2.100523548835343e-05,
+      "loss": 0.3128,
+      "step": 760
+    },
+    {
+      "epoch": 2.0184696569920844,
+      "grad_norm": 1.0393791198730469,
+      "learning_rate": 2.088505674905342e-05,
+      "loss": 0.2541,
+      "step": 765
+    },
+    {
+      "epoch": 2.0316622691292876,
+      "grad_norm": 1.0699952840805054,
+      "learning_rate": 2.0764429839897054e-05,
+      "loss": 0.2468,
+      "step": 770
+    },
+    {
+      "epoch": 2.044854881266491,
+      "grad_norm": 1.1072733402252197,
+      "learning_rate": 2.0643363947090484e-05,
+      "loss": 0.2495,
+      "step": 775
+    },
+    {
+      "epoch": 2.058047493403694,
+      "grad_norm": 1.1739845275878906,
+      "learning_rate": 2.052186829027017e-05,
+      "loss": 0.2399,
+      "step": 780
+    },
+    {
+      "epoch": 2.0712401055408973,
+      "grad_norm": 1.26287043094635,
+      "learning_rate": 2.039995212180077e-05,
+      "loss": 0.2716,
+      "step": 785
+    },
+    {
+      "epoch": 2.0844327176781,
+      "grad_norm": 1.0228111743927002,
+      "learning_rate": 2.0277624726070526e-05,
+      "loss": 0.2364,
+      "step": 790
+    },
+    {
+      "epoch": 2.0976253298153034,
+      "grad_norm": 0.946446418762207,
+      "learning_rate": 2.0154895418784245e-05,
+      "loss": 0.2239,
+      "step": 795
+    },
+    {
+      "epoch": 2.1108179419525066,
+      "grad_norm": 1.168146014213562,
+      "learning_rate": 2.0031773546253828e-05,
+      "loss": 0.2387,
+      "step": 800
+    },
+    {
+      "epoch": 2.12401055408971,
+      "grad_norm": 0.9823072552680969,
+      "learning_rate": 1.990826848468656e-05,
+      "loss": 0.2641,
+      "step": 805
+    },
+    {
+      "epoch": 2.137203166226913,
+      "grad_norm": 1.0455418825149536,
+      "learning_rate": 1.978438963947105e-05,
+      "loss": 0.2553,
+      "step": 810
+    },
+    {
+      "epoch": 2.150395778364116,
+      "grad_norm": 1.1157386302947998,
+      "learning_rate": 1.9660146444460977e-05,
+      "loss": 0.2642,
+      "step": 815
+    },
+    {
+      "epoch": 2.163588390501319,
+      "grad_norm": 1.002679467201233,
+      "learning_rate": 1.953554836125667e-05,
+      "loss": 0.2428,
+      "step": 820
+    },
+    {
+      "epoch": 2.1767810026385224,
+      "grad_norm": 1.1035633087158203,
+      "learning_rate": 1.941060487848456e-05,
+      "loss": 0.268,
+      "step": 825
+    },
+    {
+      "epoch": 2.1899736147757256,
+      "grad_norm": 1.0666553974151611,
+      "learning_rate": 1.9285325511074603e-05,
+      "loss": 0.2449,
+      "step": 830
+    },
+    {
+      "epoch": 2.203166226912929,
+      "grad_norm": 1.1026744842529297,
+      "learning_rate": 1.915971979953567e-05,
+      "loss": 0.259,
+      "step": 835
+    },
+    {
+      "epoch": 2.216358839050132,
+      "grad_norm": 1.2465076446533203,
+      "learning_rate": 1.9033797309228984e-05,
+      "loss": 0.2419,
+      "step": 840
+    },
+    {
+      "epoch": 2.229551451187335,
+      "grad_norm": 1.2780468463897705,
+      "learning_rate": 1.8907567629639727e-05,
+      "loss": 0.2715,
+      "step": 845
+    },
+    {
+      "epoch": 2.242744063324538,
+      "grad_norm": 1.1911516189575195,
+      "learning_rate": 1.878104037364671e-05,
+      "loss": 0.242,
+      "step": 850
+    },
+    {
+      "epoch": 2.2559366754617414,
+      "grad_norm": 1.2382546663284302,
+      "learning_rate": 1.865422517679034e-05,
+      "loss": 0.254,
+      "step": 855
+    },
+    {
+      "epoch": 2.2691292875989446,
+      "grad_norm": 1.0796617269515991,
+      "learning_rate": 1.852713169653885e-05,
+      "loss": 0.2754,
+      "step": 860
+    },
+    {
+      "epoch": 2.282321899736148,
+      "grad_norm": 1.1549172401428223,
+      "learning_rate": 1.8399769611552826e-05,
+      "loss": 0.2182,
+      "step": 865
+    },
+    {
+      "epoch": 2.295514511873351,
+      "grad_norm": 1.3042552471160889,
+      "learning_rate": 1.8272148620948143e-05,
+      "loss": 0.224,
+      "step": 870
+    },
+    {
+      "epoch": 2.308707124010554,
+      "grad_norm": 1.0916672945022583,
+      "learning_rate": 1.814427844355733e-05,
+      "loss": 0.2595,
+      "step": 875
+    },
+    {
+      "epoch": 2.321899736147757,
+      "grad_norm": 1.3098633289337158,
+      "learning_rate": 1.8016168817189474e-05,
+      "loss": 0.254,
+      "step": 880
+    },
+    {
+      "epoch": 2.3350923482849604,
+      "grad_norm": 1.0905364751815796,
+      "learning_rate": 1.7887829497888614e-05,
+      "loss": 0.2563,
+      "step": 885
+    },
+    {
+      "epoch": 2.3482849604221636,
+      "grad_norm": 1.167807936668396,
+      "learning_rate": 1.7759270259190804e-05,
+      "loss": 0.2479,
+      "step": 890
+    },
+    {
+      "epoch": 2.361477572559367,
+      "grad_norm": 0.9638789296150208,
+      "learning_rate": 1.7630500891379808e-05,
+      "loss": 0.24,
+      "step": 895
+    },
+    {
+      "epoch": 2.37467018469657,
+      "grad_norm": 1.1495943069458008,
+      "learning_rate": 1.7501531200741536e-05,
+      "loss": 0.2517,
+      "step": 900
+    },
+    {
+      "epoch": 2.387862796833773,
+      "grad_norm": 1.1110082864761353,
+      "learning_rate": 1.7372371008817258e-05,
+      "loss": 0.2626,
+      "step": 905
+    },
+    {
+      "epoch": 2.401055408970976,
+      "grad_norm": 1.041497826576233,
+      "learning_rate": 1.7243030151655645e-05,
+      "loss": 0.2386,
+      "step": 910
+    },
+    {
+      "epoch": 2.4142480211081794,
+      "grad_norm": 1.18124258518219,
+      "learning_rate": 1.711351847906374e-05,
+      "loss": 0.2454,
+      "step": 915
+    },
+    {
+      "epoch": 2.4274406332453826,
+      "grad_norm": 1.1296191215515137,
+      "learning_rate": 1.698384585385684e-05,
+      "loss": 0.2559,
+      "step": 920
+    },
+    {
+      "epoch": 2.440633245382586,
+      "grad_norm": 1.4808402061462402,
+      "learning_rate": 1.685402215110739e-05,
+      "loss": 0.2431,
+      "step": 925
+    },
+    {
+      "epoch": 2.453825857519789,
+      "grad_norm": 1.17811119556427,
+      "learning_rate": 1.6724057257393e-05,
+      "loss": 0.229,
+      "step": 930
+    },
+    {
+      "epoch": 2.467018469656992,
+      "grad_norm": 1.0907628536224365,
+      "learning_rate": 1.65939610700435e-05,
+      "loss": 0.2103,
+      "step": 935
+    },
+    {
+      "epoch": 2.480211081794195,
+      "grad_norm": 1.1194047927856445,
+      "learning_rate": 1.6463743496387244e-05,
+      "loss": 0.2358,
+      "step": 940
+    },
+    {
+      "epoch": 2.4934036939313984,
+      "grad_norm": 1.0814589262008667,
+      "learning_rate": 1.6333414452996625e-05,
+      "loss": 0.2305,
+      "step": 945
+    },
+    {
+      "epoch": 2.5065963060686016,
+      "grad_norm": 1.0963690280914307,
+      "learning_rate": 1.6202983864932882e-05,
+      "loss": 0.2172,
+      "step": 950
+    },
+    {
+      "epoch": 2.519788918205805,
+      "grad_norm": 1.0521795749664307,
+      "learning_rate": 1.607246166499029e-05,
+      "loss": 0.2249,
+      "step": 955
+    },
+    {
+      "epoch": 2.532981530343008,
+      "grad_norm": 1.0296999216079712,
+      "learning_rate": 1.5941857792939702e-05,
+      "loss": 0.2218,
+      "step": 960
+    },
+    {
+      "epoch": 2.5461741424802113,
+      "grad_norm": 1.1036309003829956,
+      "learning_rate": 1.5811182194771634e-05,
+      "loss": 0.2198,
+      "step": 965
+    },
+    {
+      "epoch": 2.559366754617414,
+      "grad_norm": 1.0965986251831055,
+      "learning_rate": 1.5680444821938804e-05,
+      "loss": 0.2277,
+      "step": 970
+    },
+    {
+      "epoch": 2.5725593667546174,
+      "grad_norm": 1.2318143844604492,
+      "learning_rate": 1.5549655630598345e-05,
+      "loss": 0.2201,
+      "step": 975
+    },
+    {
+      "epoch": 2.5857519788918206,
+      "grad_norm": 1.1489981412887573,
+      "learning_rate": 1.5418824580853536e-05,
+      "loss": 0.1994,
+      "step": 980
+    },
+    {
+      "epoch": 2.598944591029024,
+      "grad_norm": 1.2300056219100952,
+      "learning_rate": 1.528796163599535e-05,
+      "loss": 0.2068,
+      "step": 985
+    },
+    {
+      "epoch": 2.6121372031662267,
+      "grad_norm": 1.00900399684906,
+      "learning_rate": 1.5157076761743687e-05,
+      "loss": 0.2159,
+      "step": 990
+    },
+    {
+      "epoch": 2.62532981530343,
+      "grad_norm": 1.1844401359558105,
+      "learning_rate": 1.5026179925488476e-05,
+      "loss": 0.1943,
+      "step": 995
+    },
+    {
+      "epoch": 2.638522427440633,
+      "grad_norm": 1.164178490638733,
+      "learning_rate": 1.4895281095530577e-05,
+      "loss": 0.1945,
+      "step": 1000
+    },
+    {
+      "epoch": 2.6517150395778364,
+      "grad_norm": 1.209371566772461,
+      "learning_rate": 1.4764390240322693e-05,
+      "loss": 0.2356,
+      "step": 1005
+    },
+    {
+      "epoch": 2.6649076517150396,
+      "grad_norm": 1.2932913303375244,
+      "learning_rate": 1.4633517327710205e-05,
+      "loss": 0.2101,
+      "step": 1010
+    },
+    {
+      "epoch": 2.678100263852243,
+      "grad_norm": 1.1144524812698364,
+      "learning_rate": 1.4502672324172109e-05,
+      "loss": 0.1993,
+      "step": 1015
+    },
+    {
+      "epoch": 2.691292875989446,
+      "grad_norm": 1.1121306419372559,
+      "learning_rate": 1.4371865194062009e-05,
+      "loss": 0.1895,
+      "step": 1020
+    },
+    {
+      "epoch": 2.7044854881266494,
+      "grad_norm": 1.119805097579956,
+      "learning_rate": 1.4241105898849302e-05,
+      "loss": 0.1794,
+      "step": 1025
+    },
+    {
+      "epoch": 2.717678100263852,
+      "grad_norm": 1.107885479927063,
+      "learning_rate": 1.4110404396360578e-05,
+      "loss": 0.2025,
+      "step": 1030
+    },
+    {
+      "epoch": 2.7308707124010554,
+      "grad_norm": 1.0021132230758667,
+      "learning_rate": 1.3979770640021284e-05,
+      "loss": 0.1958,
+      "step": 1035
+    },
+    {
+      "epoch": 2.7440633245382586,
+      "grad_norm": 1.1132313013076782,
+      "learning_rate": 1.3849214578097721e-05,
+      "loss": 0.2017,
+      "step": 1040
+    },
+    {
+      "epoch": 2.757255936675462,
+      "grad_norm": 1.1634668111801147,
+      "learning_rate": 1.371874615293949e-05,
+      "loss": 0.1921,
+      "step": 1045
+    },
+    {
+      "epoch": 2.7704485488126647,
+      "grad_norm": 0.9888617396354675,
+      "learning_rate": 1.3588375300222285e-05,
+      "loss": 0.1817,
+      "step": 1050
+    },
+    {
+      "epoch": 2.783641160949868,
+      "grad_norm": 1.1379151344299316,
+      "learning_rate": 1.34581119481913e-05,
+      "loss": 0.2046,
+      "step": 1055
+    },
+    {
+      "epoch": 2.796833773087071,
+      "grad_norm": 1.1285576820373535,
+      "learning_rate": 1.332796601690512e-05,
+      "loss": 0.1954,
+      "step": 1060
+    },
+    {
+      "epoch": 2.8100263852242744,
+      "grad_norm": 0.9891535639762878,
+      "learning_rate": 1.3197947417480295e-05,
+      "loss": 0.2002,
+      "step": 1065
+    },
+    {
+      "epoch": 2.8232189973614776,
+      "grad_norm": 1.0049059391021729,
+      "learning_rate": 1.3068066051336561e-05,
+      "loss": 0.1998,
+      "step": 1070
+    },
+    {
+      "epoch": 2.836411609498681,
+      "grad_norm": 1.0179656744003296,
+      "learning_rate": 1.2938331809442803e-05,
+      "loss": 0.2071,
+      "step": 1075
+    },
+    {
+      "epoch": 2.849604221635884,
+      "grad_norm": 1.2731091976165771,
+      "learning_rate": 1.2808754571563827e-05,
+      "loss": 0.1891,
+      "step": 1080
+    },
+    {
+      "epoch": 2.862796833773087,
+      "grad_norm": 1.1574561595916748,
+      "learning_rate": 1.2679344205507982e-05,
+      "loss": 0.1871,
+      "step": 1085
+    },
+    {
+      "epoch": 2.87598944591029,
+      "grad_norm": 0.9525778293609619,
+      "learning_rate": 1.2550110566375671e-05,
+      "loss": 0.1869,
+      "step": 1090
+    },
+    {
+      "epoch": 2.8891820580474934,
+      "grad_norm": 1.1481444835662842,
+      "learning_rate": 1.2421063495808856e-05,
+      "loss": 0.1925,
+      "step": 1095
+    },
+    {
+      "epoch": 2.9023746701846966,
+      "grad_norm": 1.1231403350830078,
+      "learning_rate": 1.2292212821241602e-05,
+      "loss": 0.1894,
+      "step": 1100
+    },
+    {
+      "epoch": 2.9155672823219,
+      "grad_norm": 1.1024245023727417,
+      "learning_rate": 1.2163568355151629e-05,
+      "loss": 0.1857,
+      "step": 1105
+    },
+    {
+      "epoch": 2.9287598944591027,
+      "grad_norm": 1.1011011600494385,
+      "learning_rate": 1.2035139894313108e-05,
+      "loss": 0.1598,
+      "step": 1110
+    },
+    {
+      "epoch": 2.941952506596306,
+      "grad_norm": 1.1969327926635742,
+      "learning_rate": 1.1906937219050558e-05,
+      "loss": 0.1654,
+      "step": 1115
+    },
+    {
+      "epoch": 2.955145118733509,
+      "grad_norm": 1.1062734127044678,
+      "learning_rate": 1.1778970092494051e-05,
+      "loss": 0.1543,
+      "step": 1120
+    },
+    {
+      "epoch": 2.9683377308707124,
+      "grad_norm": 1.1782795190811157,
+      "learning_rate": 1.1651248259835731e-05,
+      "loss": 0.1592,
+      "step": 1125
+    },
+    {
+      "epoch": 2.9815303430079156,
+      "grad_norm": 1.08965265750885,
+      "learning_rate": 1.1523781447587642e-05,
+      "loss": 0.1638,
+      "step": 1130
+    },
+    {
+      "epoch": 2.994722955145119,
+      "grad_norm": 0.9494940638542175,
+      "learning_rate": 1.1396579362841045e-05,
+      "loss": 0.1728,
+      "step": 1135
+    },
+    {
+      "epoch": 3.007915567282322,
+      "grad_norm": 1.0252009630203247,
+      "learning_rate": 1.1269651692527181e-05,
+      "loss": 0.1552,
+      "step": 1140
+    },
+    {
+      "epoch": 3.021108179419525,
+      "grad_norm": 0.9815875291824341,
+      "learning_rate": 1.1143008102679562e-05,
+      "loss": 0.1496,
+      "step": 1145
+    },
+    {
+      "epoch": 3.034300791556728,
+      "grad_norm": 0.9611314535140991,
+      "learning_rate": 1.1016658237697868e-05,
+      "loss": 0.1331,
+      "step": 1150
+    },
+    {
+      "epoch": 3.0474934036939314,
+      "grad_norm": 0.9483466744422913,
+      "learning_rate": 1.0890611719613514e-05,
+      "loss": 0.1148,
+      "step": 1155
+    },
+    {
+      "epoch": 3.0606860158311346,
+      "grad_norm": 0.9694867730140686,
+      "learning_rate": 1.0764878147356852e-05,
+      "loss": 0.1306,
+      "step": 1160
+    },
+    {
+      "epoch": 3.073878627968338,
+      "grad_norm": 1.0120265483856201,
+      "learning_rate": 1.0639467096026213e-05,
+      "loss": 0.1304,
+      "step": 1165
+    },
+    {
+      "epoch": 3.0870712401055407,
+      "grad_norm": 0.9505563378334045,
+      "learning_rate": 1.0514388116158701e-05,
+      "loss": 0.1338,
+      "step": 1170
+    },
+    {
+      "epoch": 3.100263852242744,
+      "grad_norm": 0.9862446784973145,
+      "learning_rate": 1.0389650733002895e-05,
+      "loss": 0.1343,
+      "step": 1175
+    },
+    {
+      "epoch": 3.113456464379947,
+      "grad_norm": 1.0759116411209106,
+      "learning_rate": 1.0265264445793465e-05,
+      "loss": 0.1291,
+      "step": 1180
+    },
+    {
+      "epoch": 3.1266490765171504,
+      "grad_norm": 1.0146280527114868,
+      "learning_rate": 1.0141238727027761e-05,
+      "loss": 0.1226,
+      "step": 1185
+    },
+    {
+      "epoch": 3.1398416886543536,
+      "grad_norm": 0.9516650438308716,
+      "learning_rate": 1.0017583021744454e-05,
+      "loss": 0.125,
+      "step": 1190
+    },
+    {
+      "epoch": 3.153034300791557,
+      "grad_norm": 1.0282009840011597,
+      "learning_rate": 9.894306746804251e-06,
+      "loss": 0.1364,
+      "step": 1195
+    },
+    {
+      "epoch": 3.16622691292876,
+      "grad_norm": 1.0702203512191772,
+      "learning_rate": 9.771419290172776e-06,
+      "loss": 0.1431,
+      "step": 1200
+    },
+    {
+      "epoch": 3.179419525065963,
+      "grad_norm": 1.1179991960525513,
+      "learning_rate": 9.64893001020562e-06,
+      "loss": 0.1388,
+      "step": 1205
+    },
+    {
+      "epoch": 3.192612137203166,
+      "grad_norm": 0.8874260187149048,
+      "learning_rate": 9.526848234935704e-06,
+      "loss": 0.1201,
+      "step": 1210
+    },
+    {
+      "epoch": 3.2058047493403694,
+      "grad_norm": 1.0987796783447266,
+      "learning_rate": 9.405183261362866e-06,
+      "loss": 0.1305,
+      "step": 1215
+    },
+    {
+      "epoch": 3.2189973614775726,
+      "grad_norm": 0.9366940259933472,
+      "learning_rate": 9.283944354745889e-06,
+      "loss": 0.1337,
+      "step": 1220
+    },
+    {
+      "epoch": 3.232189973614776,
+      "grad_norm": 1.0771100521087646,
+      "learning_rate": 9.163140747896907e-06,
+      "loss": 0.1308,
+      "step": 1225
+    },
+    {
+      "epoch": 3.2453825857519787,
+      "grad_norm": 0.9078076481819153,
+      "learning_rate": 9.042781640478292e-06,
+      "loss": 0.141,
+      "step": 1230
+    },
+    {
+      "epoch": 3.258575197889182,
+      "grad_norm": 1.0133605003356934,
+      "learning_rate": 8.922876198302063e-06,
+      "loss": 0.1225,
+      "step": 1235
+    },
+    {
+      "epoch": 3.271767810026385,
+      "grad_norm": 0.9746168255805969,
+      "learning_rate": 8.803433552631876e-06,
+      "loss": 0.1212,
+      "step": 1240
+    },
+    {
+      "epoch": 3.2849604221635884,
+      "grad_norm": 0.8782943487167358,
+      "learning_rate": 8.684462799487635e-06,
+      "loss": 0.1376,
+      "step": 1245
+    },
+    {
+      "epoch": 3.2981530343007917,
+      "grad_norm": 0.975928008556366,
+      "learning_rate": 8.565972998952815e-06,
+      "loss": 0.1106,
+      "step": 1250
+    },
+    {
+      "epoch": 3.311345646437995,
+      "grad_norm": 1.0134680271148682,
+      "learning_rate": 8.44797317448447e-06,
+      "loss": 0.1292,
+      "step": 1255
+    },
+    {
+      "epoch": 3.324538258575198,
+      "grad_norm": 0.9296014308929443,
+      "learning_rate": 8.330472312226091e-06,
+      "loss": 0.1224,
+      "step": 1260
+    },
+    {
+      "epoch": 3.337730870712401,
+      "grad_norm": 1.0140485763549805,
+      "learning_rate": 8.213479360323258e-06,
+      "loss": 0.133,
+      "step": 1265
+    },
+    {
+      "epoch": 3.350923482849604,
+      "grad_norm": 1.0728709697723389,
+      "learning_rate": 8.097003228242227e-06,
+      "loss": 0.1104,
+      "step": 1270
+    },
+    {
+      "epoch": 3.3641160949868074,
+      "grad_norm": 1.0014153718948364,
+      "learning_rate": 7.981052786091401e-06,
+      "loss": 0.1169,
+      "step": 1275
+    },
+    {
+      "epoch": 3.3773087071240107,
+      "grad_norm": 0.9745768904685974,
+      "learning_rate": 7.865636863945869e-06,
+      "loss": 0.1133,
+      "step": 1280
+    },
+    {
+      "epoch": 3.390501319261214,
+      "grad_norm": 0.9238005876541138,
+      "learning_rate": 7.750764251174964e-06,
+      "loss": 0.119,
+      "step": 1285
+    },
+    {
+      "epoch": 3.4036939313984167,
+      "grad_norm": 1.0301371812820435,
+      "learning_rate": 7.636443695772888e-06,
+      "loss": 0.1245,
+      "step": 1290
+    },
+    {
+      "epoch": 3.41688654353562,
+      "grad_norm": 0.9262042045593262,
+      "learning_rate": 7.522683903692547e-06,
+      "loss": 0.1095,
+      "step": 1295
+    },
+    {
+      "epoch": 3.430079155672823,
+      "grad_norm": 1.1189404726028442,
+      "learning_rate": 7.409493538182546e-06,
+      "loss": 0.1245,
+      "step": 1300
+    },
+    {
+      "epoch": 3.4432717678100264,
+      "grad_norm": 1.0064700841903687,
+      "learning_rate": 7.296881219127453e-06,
+      "loss": 0.1255,
+      "step": 1305
+    },
+    {
+      "epoch": 3.4564643799472297,
+      "grad_norm": 1.1412367820739746,
+      "learning_rate": 7.1848555223913595e-06,
+      "loss": 0.1203,
+      "step": 1310
+    },
+    {
+      "epoch": 3.469656992084433,
+      "grad_norm": 0.9582564830780029,
+      "learning_rate": 7.0734249791647945e-06,
+      "loss": 0.1196,
+      "step": 1315
+    },
+    {
+      "epoch": 3.4828496042216357,
+      "grad_norm": 0.8669373393058777,
+      "learning_rate": 6.962598075315047e-06,
+      "loss": 0.1115,
+      "step": 1320
+    },
+    {
+      "epoch": 3.496042216358839,
+      "grad_norm": 1.074630618095398,
+      "learning_rate": 6.852383250739939e-06,
+      "loss": 0.1105,
+      "step": 1325
+    },
+    {
+      "epoch": 3.509234828496042,
+      "grad_norm": 0.9790440201759338,
+      "learning_rate": 6.742788898725066e-06,
+      "loss": 0.1372,
+      "step": 1330
+    },
+    {
+      "epoch": 3.5224274406332454,
+      "grad_norm": 0.8940548896789551,
+      "learning_rate": 6.633823365304649e-06,
+      "loss": 0.1345,
+      "step": 1335
+    },
+    {
+      "epoch": 3.5356200527704487,
+      "grad_norm": 0.9322503805160522,
+      "learning_rate": 6.525494948625932e-06,
+      "loss": 0.1108,
+      "step": 1340
+    },
+    {
+      "epoch": 3.5488126649076515,
+      "grad_norm": 0.8120478987693787,
+      "learning_rate": 6.41781189831726e-06,
+      "loss": 0.1113,
+      "step": 1345
+    },
+    {
+      "epoch": 3.5620052770448547,
+      "grad_norm": 0.9190356731414795,
+      "learning_rate": 6.310782414859821e-06,
+      "loss": 0.1042,
+      "step": 1350
+    },
+    {
+      "epoch": 3.575197889182058,
+      "grad_norm": 1.1531295776367188,
+      "learning_rate": 6.2044146489631595e-06,
+      "loss": 0.1267,
+      "step": 1355
+    },
+    {
+      "epoch": 3.588390501319261,
+      "grad_norm": 0.8819787502288818,
+      "learning_rate": 6.09871670094448e-06,
+      "loss": 0.1042,
+      "step": 1360
+    },
+    {
+      "epoch": 3.6015831134564644,
+      "grad_norm": 1.1059197187423706,
+      "learning_rate": 5.993696620111741e-06,
+      "loss": 0.119,
+      "step": 1365
+    },
+    {
+      "epoch": 3.6147757255936677,
+      "grad_norm": 0.9151285886764526,
+      "learning_rate": 5.8893624041507036e-06,
+      "loss": 0.0929,
+      "step": 1370
+    },
+    {
+      "epoch": 3.627968337730871,
+      "grad_norm": 0.8357903957366943,
+      "learning_rate": 5.7857219985158514e-06,
+      "loss": 0.1032,
+      "step": 1375
+    },
+    {
+      "epoch": 3.641160949868074,
+      "grad_norm": 1.0349687337875366,
+      "learning_rate": 5.682783295825346e-06,
+      "loss": 0.1145,
+      "step": 1380
+    },
+    {
+      "epoch": 3.654353562005277,
+      "grad_norm": 1.0447582006454468,
+      "learning_rate": 5.580554135259933e-06,
+      "loss": 0.098,
+      "step": 1385
+    },
+    {
+      "epoch": 3.66754617414248,
+      "grad_norm": 0.9989781975746155,
+      "learning_rate": 5.479042301965988e-06,
+      "loss": 0.1087,
+      "step": 1390
+    },
+    {
+      "epoch": 3.6807387862796834,
+      "grad_norm": 1.128767490386963,
+      "learning_rate": 5.378255526462631e-06,
+      "loss": 0.1105,
+      "step": 1395
+    },
+    {
+      "epoch": 3.6939313984168867,
+      "grad_norm": 0.8822545409202576,
+      "learning_rate": 5.2782014840530366e-06,
+      "loss": 0.1111,
+      "step": 1400
+    },
+    {
+      "epoch": 3.7071240105540895,
+      "grad_norm": 0.9544588923454285,
+      "learning_rate": 5.178887794239904e-06,
+      "loss": 0.1186,
+      "step": 1405
+    },
+    {
+      "epoch": 3.7203166226912927,
+      "grad_norm": 0.8241655230522156,
+      "learning_rate": 5.080322020145225e-06,
+      "loss": 0.1067,
+      "step": 1410
+    },
+    {
+      "epoch": 3.733509234828496,
+      "grad_norm": 0.9154494404792786,
+      "learning_rate": 4.982511667934303e-06,
+      "loss": 0.1046,
+      "step": 1415
+    },
+    {
+      "epoch": 3.746701846965699,
+      "grad_norm": 0.9264531135559082,
+      "learning_rate": 4.885464186244154e-06,
+      "loss": 0.112,
+      "step": 1420
+    },
+    {
+      "epoch": 3.7598944591029024,
+      "grad_norm": 0.9805530309677124,
+      "learning_rate": 4.789186965616233e-06,
+      "loss": 0.0889,
+      "step": 1425
+    },
+    {
+      "epoch": 3.7730870712401057,
+      "grad_norm": 1.0407724380493164,
+      "learning_rate": 4.693687337933657e-06,
+      "loss": 0.1017,
+      "step": 1430
+    },
+    {
+      "epoch": 3.786279683377309,
+      "grad_norm": 0.8391036987304688,
+      "learning_rate": 4.5989725758628035e-06,
+      "loss": 0.1015,
+      "step": 1435
+    },
+    {
+      "epoch": 3.7994722955145117,
+      "grad_norm": 0.94080650806427,
+      "learning_rate": 4.505049892299517e-06,
+      "loss": 0.1173,
+      "step": 1440
+    },
+    {
+      "epoch": 3.812664907651715,
+      "grad_norm": 0.8695018887519836,
+      "learning_rate": 4.411926439819785e-06,
+      "loss": 0.1141,
+      "step": 1445
+    },
+    {
+      "epoch": 3.825857519788918,
+      "grad_norm": 0.8716533780097961,
+      "learning_rate": 4.3196093101350545e-06,
+      "loss": 0.0994,
+      "step": 1450
+    },
+    {
+      "epoch": 3.8390501319261214,
+      "grad_norm": 0.9262160062789917,
+      "learning_rate": 4.22810553355217e-06,
+      "loss": 0.1016,
+      "step": 1455
+    },
+    {
+      "epoch": 3.8522427440633247,
+      "grad_norm": 0.8679620623588562,
+      "learning_rate": 4.137422078437992e-06,
+      "loss": 0.1087,
+      "step": 1460
+    },
+    {
+      "epoch": 3.8654353562005275,
+      "grad_norm": 0.9702051281929016,
+      "learning_rate": 4.0475658506887144e-06,
+      "loss": 0.0874,
+      "step": 1465
+    },
+    {
+      "epoch": 3.8786279683377307,
+      "grad_norm": 0.927317202091217,
+      "learning_rate": 3.958543693203985e-06,
+      "loss": 0.1009,
+      "step": 1470
+    },
+    {
+      "epoch": 3.891820580474934,
+      "grad_norm": 0.7865936756134033,
+      "learning_rate": 3.870362385365755e-06,
+      "loss": 0.0938,
+      "step": 1475
+    },
+    {
+      "epoch": 3.905013192612137,
+      "grad_norm": 0.8493991494178772,
+      "learning_rate": 3.783028642522024e-06,
+      "loss": 0.085,
+      "step": 1480
+    },
+    {
+      "epoch": 3.9182058047493404,
+      "grad_norm": 0.8362450003623962,
+      "learning_rate": 3.6965491154754346e-06,
+      "loss": 0.0993,
+      "step": 1485
+    },
+    {
+      "epoch": 3.9313984168865437,
+      "grad_norm": 0.8434048295021057,
+      "learning_rate": 3.6109303899767883e-06,
+      "loss": 0.0906,
+      "step": 1490
+    },
+    {
+      "epoch": 3.944591029023747,
+      "grad_norm": 0.9962806701660156,
+      "learning_rate": 3.526178986223524e-06,
+      "loss": 0.1078,
+      "step": 1495
+    },
+    {
+      "epoch": 3.9577836411609497,
+      "grad_norm": 0.9092873930931091,
+      "learning_rate": 3.442301358363163e-06,
+      "loss": 0.0923,
+      "step": 1500
+    },
+    {
+      "epoch": 3.970976253298153,
+      "grad_norm": 0.7910172939300537,
+      "learning_rate": 3.3593038940018094e-06,
+      "loss": 0.0857,
+      "step": 1505
+    },
+    {
+      "epoch": 3.984168865435356,
+      "grad_norm": 0.825131356716156,
+      "learning_rate": 3.2771929137177174e-06,
+      "loss": 0.0993,
+      "step": 1510
+    },
+    {
+      "epoch": 3.9973614775725594,
+      "grad_norm": 0.9550831317901611,
+      "learning_rate": 3.1959746705799412e-06,
+      "loss": 0.114,
+      "step": 1515
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1895,
+  "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": 2.1880537050670694e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

9_128_e5_3e-5/checkpoint-1516/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f0c9334f6fbff83eadc1fc0cb18b7b9f0220f80a8560293b3502a56c8ccec92
+size 7736

9_128_e5_3e-5/checkpoint-1516/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)

9_128_e5_3e-5/checkpoint-1895/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

9_128_e5_3e-5/checkpoint-1895/adapter_config.json

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

9_128_e5_3e-5/checkpoint-1895/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1a02e11ebd9dd4ccb34097a1367aa5f5b2ac4b74eb852996a5e68497ca04e58c
+size 791751704

9_128_e5_3e-5/checkpoint-1895/latest

@@ -0,0 +1 @@
+global_step1895

9_128_e5_3e-5/checkpoint-1895/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:97d51258426f83ff1f6f097cd23e2b6acaab557dad6052ed7cf516ba951c1711
+size 15920

9_128_e5_3e-5/checkpoint-1895/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:080bf14b166a5ce7dca0bfe6d1583b1e308a18292dee9af43cf765058063a688
+size 15920

9_128_e5_3e-5/checkpoint-1895/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fd972415812da97118fb9c3657dea09c8a142379cd4847e025591c5e0d67941d
+size 15920