Upload folder using huggingface_hub

53_128_e5_3e-5/checkpoint-1076/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

53_128_e5_3e-5/checkpoint-1076/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": [
+    "q_proj",
+    "k_proj",
+    "down_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "v_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

53_128_e5_3e-5/checkpoint-1076/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:722d6683c01da3cacbfa2ea502233df5c20ed3334ad62c31380210b84f4d43c7
+size 791751704

53_128_e5_3e-5/checkpoint-1076/latest

@@ -0,0 +1 @@
+global_step1076

53_128_e5_3e-5/checkpoint-1076/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:672148f9de1b976452c8aa1ee80fde87f1b51b108e42594599d5593fea706c2c
+size 15920

53_128_e5_3e-5/checkpoint-1076/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:81244619f77827de775573611f1fcf8809463b5f0a810a5e3a72754fec92b056
+size 15920

53_128_e5_3e-5/checkpoint-1076/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0b6c5553b60ee4b04b5d7969929f21932df2cdbfa41a4b3cfcfb5e486e38d40f
+size 15920

53_128_e5_3e-5/checkpoint-1076/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0ddc1cd426ab1e124127abc919ba5a46f0e46ba7092bf2fb75a1464935e9956c
+size 15920

53_128_e5_3e-5/checkpoint-1076/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3ae739f55c022cddbd60bfcbf67a2870d66ff1ff3f640e903a23bbd4338468a1
+size 15920

53_128_e5_3e-5/checkpoint-1076/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:95ecbddcd25915fdd5894fb16cefd00de649fecb17785406b23a119f80aa22c7
+size 15920

53_128_e5_3e-5/checkpoint-1076/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80c2a3abbe01b69cbb4543da2558866b2cd5a22e7abe1701681abaf572b69708
+size 15920

53_128_e5_3e-5/checkpoint-1076/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f0565ce576f5b5792914d0ffa2462d310003aa58643162af5f1a214d24be477e
+size 15920

53_128_e5_3e-5/checkpoint-1076/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f5a0c1bdd5cc5a7618c13893a27545b25095b373dce1b7e087bca766e790756c
+size 1064

53_128_e5_3e-5/checkpoint-1076/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
+  }
+}

53_128_e5_3e-5/checkpoint-1076/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
+}

53_128_e5_3e-5/checkpoint-1076/trainer_state.json

@@ -0,0 +1,1539 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 1076,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0186219739292365,
+      "grad_norm": 1.397485375404358,
+      "learning_rate": 1.7647058823529412e-06,
+      "loss": 1.3045,
+      "step": 5
+    },
+    {
+      "epoch": 0.037243947858473,
+      "grad_norm": 0.9164593815803528,
+      "learning_rate": 3.970588235294118e-06,
+      "loss": 1.3238,
+      "step": 10
+    },
+    {
+      "epoch": 0.055865921787709494,
+      "grad_norm": 0.7285172343254089,
+      "learning_rate": 6.176470588235294e-06,
+      "loss": 1.3402,
+      "step": 15
+    },
+    {
+      "epoch": 0.074487895716946,
+      "grad_norm": 0.7771405577659607,
+      "learning_rate": 8.382352941176472e-06,
+      "loss": 1.2444,
+      "step": 20
+    },
+    {
+      "epoch": 0.0931098696461825,
+      "grad_norm": 0.590904951095581,
+      "learning_rate": 1.0588235294117648e-05,
+      "loss": 1.2908,
+      "step": 25
+    },
+    {
+      "epoch": 0.11173184357541899,
+      "grad_norm": 0.68792724609375,
+      "learning_rate": 1.2794117647058824e-05,
+      "loss": 1.2953,
+      "step": 30
+    },
+    {
+      "epoch": 0.1303538175046555,
+      "grad_norm": 0.6625350117683411,
+      "learning_rate": 1.5e-05,
+      "loss": 1.192,
+      "step": 35
+    },
+    {
+      "epoch": 0.148975791433892,
+      "grad_norm": 0.5564359426498413,
+      "learning_rate": 1.7205882352941175e-05,
+      "loss": 1.1406,
+      "step": 40
+    },
+    {
+      "epoch": 0.16759776536312848,
+      "grad_norm": 0.4591367542743683,
+      "learning_rate": 1.9411764705882355e-05,
+      "loss": 1.181,
+      "step": 45
+    },
+    {
+      "epoch": 0.186219739292365,
+      "grad_norm": 0.5679731369018555,
+      "learning_rate": 2.161764705882353e-05,
+      "loss": 1.1935,
+      "step": 50
+    },
+    {
+      "epoch": 0.2048417132216015,
+      "grad_norm": 0.4359203577041626,
+      "learning_rate": 2.3823529411764704e-05,
+      "loss": 1.1527,
+      "step": 55
+    },
+    {
+      "epoch": 0.22346368715083798,
+      "grad_norm": 0.49340277910232544,
+      "learning_rate": 2.6029411764705883e-05,
+      "loss": 1.1635,
+      "step": 60
+    },
+    {
+      "epoch": 0.24208566108007448,
+      "grad_norm": 0.4740353226661682,
+      "learning_rate": 2.823529411764706e-05,
+      "loss": 1.1883,
+      "step": 65
+    },
+    {
+      "epoch": 0.260707635009311,
+      "grad_norm": 0.6028391122817993,
+      "learning_rate": 2.9999954608033783e-05,
+      "loss": 1.1328,
+      "step": 70
+    },
+    {
+      "epoch": 0.27932960893854747,
+      "grad_norm": 0.5162932872772217,
+      "learning_rate": 2.9998365918062082e-05,
+      "loss": 1.0934,
+      "step": 75
+    },
+    {
+      "epoch": 0.297951582867784,
+      "grad_norm": 0.5382624864578247,
+      "learning_rate": 2.9994507904496206e-05,
+      "loss": 1.1019,
+      "step": 80
+    },
+    {
+      "epoch": 0.3165735567970205,
+      "grad_norm": 0.5565810799598694,
+      "learning_rate": 2.998838115107183e-05,
+      "loss": 1.0906,
+      "step": 85
+    },
+    {
+      "epoch": 0.33519553072625696,
+      "grad_norm": 0.5212501287460327,
+      "learning_rate": 2.997998658479568e-05,
+      "loss": 1.0724,
+      "step": 90
+    },
+    {
+      "epoch": 0.3538175046554935,
+      "grad_norm": 0.5489919185638428,
+      "learning_rate": 2.9969325475805274e-05,
+      "loss": 1.0658,
+      "step": 95
+    },
+    {
+      "epoch": 0.37243947858473,
+      "grad_norm": 0.6241554021835327,
+      "learning_rate": 2.995639943717676e-05,
+      "loss": 1.059,
+      "step": 100
+    },
+    {
+      "epoch": 0.39106145251396646,
+      "grad_norm": 0.563462495803833,
+      "learning_rate": 2.9941210424680813e-05,
+      "loss": 1.057,
+      "step": 105
+    },
+    {
+      "epoch": 0.409683426443203,
+      "grad_norm": 0.5747496485710144,
+      "learning_rate": 2.9923760736486766e-05,
+      "loss": 1.0564,
+      "step": 110
+    },
+    {
+      "epoch": 0.42830540037243947,
+      "grad_norm": 0.5738973021507263,
+      "learning_rate": 2.9904053012814848e-05,
+      "loss": 1.0316,
+      "step": 115
+    },
+    {
+      "epoch": 0.44692737430167595,
+      "grad_norm": 0.6159545183181763,
+      "learning_rate": 2.988209023553672e-05,
+      "loss": 0.995,
+      "step": 120
+    },
+    {
+      "epoch": 0.4655493482309125,
+      "grad_norm": 0.6447931528091431,
+      "learning_rate": 2.9857875727724304e-05,
+      "loss": 0.9692,
+      "step": 125
+    },
+    {
+      "epoch": 0.48417132216014896,
+      "grad_norm": 0.7060043215751648,
+      "learning_rate": 2.9831413153146988e-05,
+      "loss": 0.9569,
+      "step": 130
+    },
+    {
+      "epoch": 0.5027932960893855,
+      "grad_norm": 0.7400388717651367,
+      "learning_rate": 2.9802706515717272e-05,
+      "loss": 0.9378,
+      "step": 135
+    },
+    {
+      "epoch": 0.521415270018622,
+      "grad_norm": 0.7127872705459595,
+      "learning_rate": 2.9771760158884972e-05,
+      "loss": 0.8923,
+      "step": 140
+    },
+    {
+      "epoch": 0.5400372439478585,
+      "grad_norm": 0.8123269081115723,
+      "learning_rate": 2.9738578764980025e-05,
+      "loss": 0.9285,
+      "step": 145
+    },
+    {
+      "epoch": 0.5586592178770949,
+      "grad_norm": 0.8044330477714539,
+      "learning_rate": 2.9703167354504027e-05,
+      "loss": 0.885,
+      "step": 150
+    },
+    {
+      "epoch": 0.5772811918063314,
+      "grad_norm": 0.7073454856872559,
+      "learning_rate": 2.966553128537062e-05,
+      "loss": 0.9022,
+      "step": 155
+    },
+    {
+      "epoch": 0.595903165735568,
+      "grad_norm": 0.725287914276123,
+      "learning_rate": 2.9625676252094797e-05,
+      "loss": 0.8863,
+      "step": 160
+    },
+    {
+      "epoch": 0.6145251396648045,
+      "grad_norm": 0.843565046787262,
+      "learning_rate": 2.9583608284931317e-05,
+      "loss": 0.928,
+      "step": 165
+    },
+    {
+      "epoch": 0.633147113594041,
+      "grad_norm": 0.8186341524124146,
+      "learning_rate": 2.953933374896227e-05,
+      "loss": 0.8967,
+      "step": 170
+    },
+    {
+      "epoch": 0.6517690875232774,
+      "grad_norm": 0.8052655458450317,
+      "learning_rate": 2.949285934313405e-05,
+      "loss": 0.8766,
+      "step": 175
+    },
+    {
+      "epoch": 0.6703910614525139,
+      "grad_norm": 1.142082691192627,
+      "learning_rate": 2.9444192099243733e-05,
+      "loss": 0.8402,
+      "step": 180
+    },
+    {
+      "epoch": 0.6890130353817505,
+      "grad_norm": 0.9169373512268066,
+      "learning_rate": 2.939333938087515e-05,
+      "loss": 0.8426,
+      "step": 185
+    },
+    {
+      "epoch": 0.707635009310987,
+      "grad_norm": 0.8522142171859741,
+      "learning_rate": 2.9340308882284747e-05,
+      "loss": 0.8288,
+      "step": 190
+    },
+    {
+      "epoch": 0.7262569832402235,
+      "grad_norm": 0.8226320147514343,
+      "learning_rate": 2.92851086272374e-05,
+      "loss": 0.8252,
+      "step": 195
+    },
+    {
+      "epoch": 0.74487895716946,
+      "grad_norm": 0.840149462223053,
+      "learning_rate": 2.9227746967792392e-05,
+      "loss": 0.7888,
+      "step": 200
+    },
+    {
+      "epoch": 0.7635009310986964,
+      "grad_norm": 1.0344595909118652,
+      "learning_rate": 2.916823258303968e-05,
+      "loss": 0.7889,
+      "step": 205
+    },
+    {
+      "epoch": 0.7821229050279329,
+      "grad_norm": 0.9850447177886963,
+      "learning_rate": 2.9106574477786748e-05,
+      "loss": 0.7634,
+      "step": 210
+    },
+    {
+      "epoch": 0.8007448789571695,
+      "grad_norm": 0.9077547192573547,
+      "learning_rate": 2.9042781981196095e-05,
+      "loss": 0.7372,
+      "step": 215
+    },
+    {
+      "epoch": 0.819366852886406,
+      "grad_norm": 0.8616479635238647,
+      "learning_rate": 2.897686474537373e-05,
+      "loss": 0.7238,
+      "step": 220
+    },
+    {
+      "epoch": 0.8379888268156425,
+      "grad_norm": 0.9395949244499207,
+      "learning_rate": 2.890883274390872e-05,
+      "loss": 0.6952,
+      "step": 225
+    },
+    {
+      "epoch": 0.8566108007448789,
+      "grad_norm": 0.9580796957015991,
+      "learning_rate": 2.8838696270364183e-05,
+      "loss": 0.6983,
+      "step": 230
+    },
+    {
+      "epoch": 0.8752327746741154,
+      "grad_norm": 0.8633882403373718,
+      "learning_rate": 2.8766465936719785e-05,
+      "loss": 0.7479,
+      "step": 235
+    },
+    {
+      "epoch": 0.8938547486033519,
+      "grad_norm": 1.0677725076675415,
+      "learning_rate": 2.869215267176612e-05,
+      "loss": 0.7132,
+      "step": 240
+    },
+    {
+      "epoch": 0.9124767225325885,
+      "grad_norm": 1.0318955183029175,
+      "learning_rate": 2.8615767719451125e-05,
+      "loss": 0.6744,
+      "step": 245
+    },
+    {
+      "epoch": 0.931098696461825,
+      "grad_norm": 0.8933286070823669,
+      "learning_rate": 2.8537322637178816e-05,
+      "loss": 0.705,
+      "step": 250
+    },
+    {
+      "epoch": 0.9497206703910615,
+      "grad_norm": 0.8829292058944702,
+      "learning_rate": 2.8456829294060608e-05,
+      "loss": 0.7091,
+      "step": 255
+    },
+    {
+      "epoch": 0.9683426443202979,
+      "grad_norm": 1.1101434230804443,
+      "learning_rate": 2.837429986911944e-05,
+      "loss": 0.6532,
+      "step": 260
+    },
+    {
+      "epoch": 0.9869646182495344,
+      "grad_norm": 1.070557713508606,
+      "learning_rate": 2.828974684944707e-05,
+      "loss": 0.6493,
+      "step": 265
+    },
+    {
+      "epoch": 1.0037243947858474,
+      "grad_norm": 1.0655916929244995,
+      "learning_rate": 2.820318302831467e-05,
+      "loss": 0.6279,
+      "step": 270
+    },
+    {
+      "epoch": 1.0223463687150838,
+      "grad_norm": 1.0981754064559937,
+      "learning_rate": 2.8114621503237186e-05,
+      "loss": 0.5955,
+      "step": 275
+    },
+    {
+      "epoch": 1.0409683426443204,
+      "grad_norm": 1.099587082862854,
+      "learning_rate": 2.8024075673991597e-05,
+      "loss": 0.6128,
+      "step": 280
+    },
+    {
+      "epoch": 1.0595903165735567,
+      "grad_norm": 1.0760383605957031,
+      "learning_rate": 2.7931559240589487e-05,
+      "loss": 0.5807,
+      "step": 285
+    },
+    {
+      "epoch": 1.0782122905027933,
+      "grad_norm": 1.028070330619812,
+      "learning_rate": 2.7837086201204128e-05,
+      "loss": 0.5667,
+      "step": 290
+    },
+    {
+      "epoch": 1.0968342644320297,
+      "grad_norm": 1.0326428413391113,
+      "learning_rate": 2.7740670850052543e-05,
+      "loss": 0.5615,
+      "step": 295
+    },
+    {
+      "epoch": 1.1154562383612663,
+      "grad_norm": 1.1273221969604492,
+      "learning_rate": 2.764232777523269e-05,
+      "loss": 0.5597,
+      "step": 300
+    },
+    {
+      "epoch": 1.1340782122905029,
+      "grad_norm": 0.9928714036941528,
+      "learning_rate": 2.7542071856516215e-05,
+      "loss": 0.5426,
+      "step": 305
+    },
+    {
+      "epoch": 1.1527001862197392,
+      "grad_norm": 1.1522924900054932,
+      "learning_rate": 2.7439918263097097e-05,
+      "loss": 0.5586,
+      "step": 310
+    },
+    {
+      "epoch": 1.1713221601489758,
+      "grad_norm": 1.0707367658615112,
+      "learning_rate": 2.7335882451296438e-05,
+      "loss": 0.5964,
+      "step": 315
+    },
+    {
+      "epoch": 1.1899441340782122,
+      "grad_norm": 1.0582071542739868,
+      "learning_rate": 2.7229980162223876e-05,
+      "loss": 0.5463,
+      "step": 320
+    },
+    {
+      "epoch": 1.2085661080074488,
+      "grad_norm": 1.109670639038086,
+      "learning_rate": 2.7122227419395885e-05,
+      "loss": 0.5033,
+      "step": 325
+    },
+    {
+      "epoch": 1.2271880819366854,
+      "grad_norm": 1.0573663711547852,
+      "learning_rate": 2.7012640526311324e-05,
+      "loss": 0.5443,
+      "step": 330
+    },
+    {
+      "epoch": 1.2458100558659218,
+      "grad_norm": 1.0600390434265137,
+      "learning_rate": 2.6901236063984646e-05,
+      "loss": 0.5143,
+      "step": 335
+    },
+    {
+      "epoch": 1.2644320297951583,
+      "grad_norm": 1.339310646057129,
+      "learning_rate": 2.6788030888437118e-05,
+      "loss": 0.5324,
+      "step": 340
+    },
+    {
+      "epoch": 1.2830540037243947,
+      "grad_norm": 1.5570026636123657,
+      "learning_rate": 2.6673042128146414e-05,
+      "loss": 0.5467,
+      "step": 345
+    },
+    {
+      "epoch": 1.3016759776536313,
+      "grad_norm": 1.0688986778259277,
+      "learning_rate": 2.6556287181455013e-05,
+      "loss": 0.4999,
+      "step": 350
+    },
+    {
+      "epoch": 1.3202979515828677,
+      "grad_norm": 1.2187882661819458,
+      "learning_rate": 2.643778371393773e-05,
+      "loss": 0.507,
+      "step": 355
+    },
+    {
+      "epoch": 1.3389199255121043,
+      "grad_norm": 1.149132490158081,
+      "learning_rate": 2.6317549655728846e-05,
+      "loss": 0.4686,
+      "step": 360
+    },
+    {
+      "epoch": 1.3575418994413408,
+      "grad_norm": 1.230714201927185,
+      "learning_rate": 2.6195603198809175e-05,
+      "loss": 0.488,
+      "step": 365
+    },
+    {
+      "epoch": 1.3761638733705772,
+      "grad_norm": 1.1980931758880615,
+      "learning_rate": 2.6071962794253548e-05,
+      "loss": 0.5051,
+      "step": 370
+    },
+    {
+      "epoch": 1.3947858472998138,
+      "grad_norm": 1.1617096662521362,
+      "learning_rate": 2.594664714943908e-05,
+      "loss": 0.4775,
+      "step": 375
+    },
+    {
+      "epoch": 1.4134078212290504,
+      "grad_norm": 1.2936596870422363,
+      "learning_rate": 2.5819675225214644e-05,
+      "loss": 0.4976,
+      "step": 380
+    },
+    {
+      "epoch": 1.4320297951582868,
+      "grad_norm": 1.230708122253418,
+      "learning_rate": 2.5691066233032e-05,
+      "loss": 0.4756,
+      "step": 385
+    },
+    {
+      "epoch": 1.4506517690875234,
+      "grad_norm": 0.9902650713920593,
+      "learning_rate": 2.5560839632039032e-05,
+      "loss": 0.4514,
+      "step": 390
+    },
+    {
+      "epoch": 1.4692737430167597,
+      "grad_norm": 1.0236945152282715,
+      "learning_rate": 2.5429015126135482e-05,
+      "loss": 0.4197,
+      "step": 395
+    },
+    {
+      "epoch": 1.4878957169459963,
+      "grad_norm": 1.0958354473114014,
+      "learning_rate": 2.5295612660991632e-05,
+      "loss": 0.4475,
+      "step": 400
+    },
+    {
+      "epoch": 1.5065176908752327,
+      "grad_norm": 1.2085777521133423,
+      "learning_rate": 2.516065242103047e-05,
+      "loss": 0.4189,
+      "step": 405
+    },
+    {
+      "epoch": 1.5251396648044693,
+      "grad_norm": 1.070729374885559,
+      "learning_rate": 2.5024154826373655e-05,
+      "loss": 0.4804,
+      "step": 410
+    },
+    {
+      "epoch": 1.5437616387337059,
+      "grad_norm": 1.1738492250442505,
+      "learning_rate": 2.488614052975188e-05,
+      "loss": 0.4061,
+      "step": 415
+    },
+    {
+      "epoch": 1.5623836126629422,
+      "grad_norm": 1.128957986831665,
+      "learning_rate": 2.474663041338e-05,
+      "loss": 0.4125,
+      "step": 420
+    },
+    {
+      "epoch": 1.5810055865921788,
+      "grad_norm": 1.1465480327606201,
+      "learning_rate": 2.4605645585797476e-05,
+      "loss": 0.4208,
+      "step": 425
+    },
+    {
+      "epoch": 1.5996275605214154,
+      "grad_norm": 1.1640135049819946,
+      "learning_rate": 2.4463207378674525e-05,
+      "loss": 0.4578,
+      "step": 430
+    },
+    {
+      "epoch": 1.6182495344506518,
+      "grad_norm": 1.242473840713501,
+      "learning_rate": 2.431933734358456e-05,
+      "loss": 0.406,
+      "step": 435
+    },
+    {
+      "epoch": 1.6368715083798882,
+      "grad_norm": 1.2133651971817017,
+      "learning_rate": 2.417405724874333e-05,
+      "loss": 0.4357,
+      "step": 440
+    },
+    {
+      "epoch": 1.6554934823091247,
+      "grad_norm": 1.2609267234802246,
+      "learning_rate": 2.4027389075715272e-05,
+      "loss": 0.427,
+      "step": 445
+    },
+    {
+      "epoch": 1.6741154562383613,
+      "grad_norm": 1.1197216510772705,
+      "learning_rate": 2.3879355016087623e-05,
+      "loss": 0.4025,
+      "step": 450
+    },
+    {
+      "epoch": 1.6927374301675977,
+      "grad_norm": 1.1437435150146484,
+      "learning_rate": 2.3729977468112696e-05,
+      "loss": 0.3899,
+      "step": 455
+    },
+    {
+      "epoch": 1.7113594040968343,
+      "grad_norm": 1.1721247434616089,
+      "learning_rate": 2.357927903331895e-05,
+      "loss": 0.4361,
+      "step": 460
+    },
+    {
+      "epoch": 1.7299813780260709,
+      "grad_norm": 1.2531391382217407,
+      "learning_rate": 2.342728251309126e-05,
+      "loss": 0.3918,
+      "step": 465
+    },
+    {
+      "epoch": 1.7486033519553073,
+      "grad_norm": 1.160923957824707,
+      "learning_rate": 2.327401090522094e-05,
+      "loss": 0.3994,
+      "step": 470
+    },
+    {
+      "epoch": 1.7672253258845436,
+      "grad_norm": 1.3061710596084595,
+      "learning_rate": 2.3119487400426097e-05,
+      "loss": 0.4156,
+      "step": 475
+    },
+    {
+      "epoch": 1.7858472998137802,
+      "grad_norm": 1.0728996992111206,
+      "learning_rate": 2.2963735378842757e-05,
+      "loss": 0.3827,
+      "step": 480
+    },
+    {
+      "epoch": 1.8044692737430168,
+      "grad_norm": 1.153071403503418,
+      "learning_rate": 2.2806778406487337e-05,
+      "loss": 0.399,
+      "step": 485
+    },
+    {
+      "epoch": 1.8230912476722532,
+      "grad_norm": 1.1256935596466064,
+      "learning_rate": 2.2648640231690987e-05,
+      "loss": 0.3721,
+      "step": 490
+    },
+    {
+      "epoch": 1.8417132216014898,
+      "grad_norm": 1.1391295194625854,
+      "learning_rate": 2.2489344781506375e-05,
+      "loss": 0.3625,
+      "step": 495
+    },
+    {
+      "epoch": 1.8603351955307263,
+      "grad_norm": 1.235418438911438,
+      "learning_rate": 2.2328916158087396e-05,
+      "loss": 0.3652,
+      "step": 500
+    },
+    {
+      "epoch": 1.8789571694599627,
+      "grad_norm": 1.1151893138885498,
+      "learning_rate": 2.2167378635042414e-05,
+      "loss": 0.3526,
+      "step": 505
+    },
+    {
+      "epoch": 1.8975791433891993,
+      "grad_norm": 1.1720143556594849,
+      "learning_rate": 2.2004756653761533e-05,
+      "loss": 0.363,
+      "step": 510
+    },
+    {
+      "epoch": 1.916201117318436,
+      "grad_norm": 1.219879388809204,
+      "learning_rate": 2.184107481971851e-05,
+      "loss": 0.3879,
+      "step": 515
+    },
+    {
+      "epoch": 1.9348230912476723,
+      "grad_norm": 1.3815174102783203,
+      "learning_rate": 2.1676357898747836e-05,
+      "loss": 0.3744,
+      "step": 520
+    },
+    {
+      "epoch": 1.9534450651769086,
+      "grad_norm": 1.2076925039291382,
+      "learning_rate": 2.1510630813297545e-05,
+      "loss": 0.3371,
+      "step": 525
+    },
+    {
+      "epoch": 1.9720670391061452,
+      "grad_norm": 1.265386700630188,
+      "learning_rate": 2.134391863865832e-05,
+      "loss": 0.35,
+      "step": 530
+    },
+    {
+      "epoch": 1.9906890130353818,
+      "grad_norm": 1.168981671333313,
+      "learning_rate": 2.1176246599169492e-05,
+      "loss": 0.3677,
+      "step": 535
+    },
+    {
+      "epoch": 2.007448789571695,
+      "grad_norm": 1.2176551818847656,
+      "learning_rate": 2.1007640064402474e-05,
+      "loss": 0.3187,
+      "step": 540
+    },
+    {
+      "epoch": 2.026070763500931,
+      "grad_norm": 1.2926204204559326,
+      "learning_rate": 2.083812454532224e-05,
+      "loss": 0.2928,
+      "step": 545
+    },
+    {
+      "epoch": 2.0446927374301676,
+      "grad_norm": 1.0775201320648193,
+      "learning_rate": 2.0667725690427374e-05,
+      "loss": 0.2886,
+      "step": 550
+    },
+    {
+      "epoch": 2.063314711359404,
+      "grad_norm": 1.205060362815857,
+      "learning_rate": 2.0496469281869337e-05,
+      "loss": 0.2741,
+      "step": 555
+    },
+    {
+      "epoch": 2.0819366852886407,
+      "grad_norm": 1.238979458808899,
+      "learning_rate": 2.0324381231551516e-05,
+      "loss": 0.256,
+      "step": 560
+    },
+    {
+      "epoch": 2.100558659217877,
+      "grad_norm": 1.0161023139953613,
+      "learning_rate": 2.0151487577208613e-05,
+      "loss": 0.2865,
+      "step": 565
+    },
+    {
+      "epoch": 2.1191806331471135,
+      "grad_norm": 1.2487982511520386,
+      "learning_rate": 1.997781447846705e-05,
+      "loss": 0.2548,
+      "step": 570
+    },
+    {
+      "epoch": 2.1378026070763503,
+      "grad_norm": 1.2972549200057983,
+      "learning_rate": 1.9803388212886828e-05,
+      "loss": 0.2343,
+      "step": 575
+    },
+    {
+      "epoch": 2.1564245810055866,
+      "grad_norm": 1.1097617149353027,
+      "learning_rate": 1.9628235171985688e-05,
+      "loss": 0.2774,
+      "step": 580
+    },
+    {
+      "epoch": 2.175046554934823,
+      "grad_norm": 1.1447429656982422,
+      "learning_rate": 1.9452381857245876e-05,
+      "loss": 0.24,
+      "step": 585
+    },
+    {
+      "epoch": 2.1936685288640594,
+      "grad_norm": 1.177545189857483,
+      "learning_rate": 1.92758548761044e-05,
+      "loss": 0.2804,
+      "step": 590
+    },
+    {
+      "epoch": 2.212290502793296,
+      "grad_norm": 1.2133318185806274,
+      "learning_rate": 1.9098680937927164e-05,
+      "loss": 0.2409,
+      "step": 595
+    },
+    {
+      "epoch": 2.2309124767225326,
+      "grad_norm": 0.9909643530845642,
+      "learning_rate": 1.8920886849967725e-05,
+      "loss": 0.278,
+      "step": 600
+    },
+    {
+      "epoch": 2.249534450651769,
+      "grad_norm": 1.1698384284973145,
+      "learning_rate": 1.8742499513311242e-05,
+      "loss": 0.2409,
+      "step": 605
+    },
+    {
+      "epoch": 2.2681564245810057,
+      "grad_norm": 1.11681067943573,
+      "learning_rate": 1.856354591880419e-05,
+      "loss": 0.2377,
+      "step": 610
+    },
+    {
+      "epoch": 2.286778398510242,
+      "grad_norm": 1.2661349773406982,
+      "learning_rate": 1.838405314297055e-05,
+      "loss": 0.2382,
+      "step": 615
+    },
+    {
+      "epoch": 2.3054003724394785,
+      "grad_norm": 1.0915417671203613,
+      "learning_rate": 1.8204048343914962e-05,
+      "loss": 0.22,
+      "step": 620
+    },
+    {
+      "epoch": 2.3240223463687153,
+      "grad_norm": 1.3376933336257935,
+      "learning_rate": 1.8023558757213604e-05,
+      "loss": 0.269,
+      "step": 625
+    },
+    {
+      "epoch": 2.3426443202979517,
+      "grad_norm": 1.1464945077896118,
+      "learning_rate": 1.7842611691793334e-05,
+      "loss": 0.2431,
+      "step": 630
+    },
+    {
+      "epoch": 2.361266294227188,
+      "grad_norm": 1.135488748550415,
+      "learning_rate": 1.766123452579967e-05,
+      "loss": 0.2293,
+      "step": 635
+    },
+    {
+      "epoch": 2.3798882681564244,
+      "grad_norm": 1.2150588035583496,
+      "learning_rate": 1.74794547024544e-05,
+      "loss": 0.2244,
+      "step": 640
+    },
+    {
+      "epoch": 2.398510242085661,
+      "grad_norm": 1.1374517679214478,
+      "learning_rate": 1.729729972590326e-05,
+      "loss": 0.2364,
+      "step": 645
+    },
+    {
+      "epoch": 2.4171322160148976,
+      "grad_norm": 1.2967171669006348,
+      "learning_rate": 1.711479715705444e-05,
+      "loss": 0.2482,
+      "step": 650
+    },
+    {
+      "epoch": 2.435754189944134,
+      "grad_norm": 1.166526436805725,
+      "learning_rate": 1.693197460940849e-05,
+      "loss": 0.2463,
+      "step": 655
+    },
+    {
+      "epoch": 2.4543761638733708,
+      "grad_norm": 1.0751850605010986,
+      "learning_rate": 1.6748859744880255e-05,
+      "loss": 0.2579,
+      "step": 660
+    },
+    {
+      "epoch": 2.472998137802607,
+      "grad_norm": 1.1875776052474976,
+      "learning_rate": 1.6565480269613522e-05,
+      "loss": 0.2306,
+      "step": 665
+    },
+    {
+      "epoch": 2.4916201117318435,
+      "grad_norm": 1.264696478843689,
+      "learning_rate": 1.638186392978891e-05,
+      "loss": 0.1954,
+      "step": 670
+    },
+    {
+      "epoch": 2.51024208566108,
+      "grad_norm": 1.1798068284988403,
+      "learning_rate": 1.6198038507425777e-05,
+      "loss": 0.222,
+      "step": 675
+    },
+    {
+      "epoch": 2.5288640595903167,
+      "grad_norm": 1.2695372104644775,
+      "learning_rate": 1.6014031816178655e-05,
+      "loss": 0.2006,
+      "step": 680
+    },
+    {
+      "epoch": 2.547486033519553,
+      "grad_norm": 1.154645562171936,
+      "learning_rate": 1.582987169712891e-05,
+      "loss": 0.2405,
+      "step": 685
+    },
+    {
+      "epoch": 2.5661080074487894,
+      "grad_norm": 1.175594687461853,
+      "learning_rate": 1.564558601457227e-05,
+      "loss": 0.2436,
+      "step": 690
+    },
+    {
+      "epoch": 2.5847299813780262,
+      "grad_norm": 1.191159725189209,
+      "learning_rate": 1.5461202651802802e-05,
+      "loss": 0.2027,
+      "step": 695
+    },
+    {
+      "epoch": 2.6033519553072626,
+      "grad_norm": 1.2153513431549072,
+      "learning_rate": 1.5276749506894056e-05,
+      "loss": 0.2384,
+      "step": 700
+    },
+    {
+      "epoch": 2.621973929236499,
+      "grad_norm": 1.1220816373825073,
+      "learning_rate": 1.5092254488477951e-05,
+      "loss": 0.2117,
+      "step": 705
+    },
+    {
+      "epoch": 2.6405959031657353,
+      "grad_norm": 1.1654083728790283,
+      "learning_rate": 1.4907745511522056e-05,
+      "loss": 0.2474,
+      "step": 710
+    },
+    {
+      "epoch": 2.659217877094972,
+      "grad_norm": 1.138382911682129,
+      "learning_rate": 1.4723250493105948e-05,
+      "loss": 0.1977,
+      "step": 715
+    },
+    {
+      "epoch": 2.6778398510242085,
+      "grad_norm": 1.1752201318740845,
+      "learning_rate": 1.4538797348197202e-05,
+      "loss": 0.2314,
+      "step": 720
+    },
+    {
+      "epoch": 2.6964618249534453,
+      "grad_norm": 1.0158238410949707,
+      "learning_rate": 1.4354413985427733e-05,
+      "loss": 0.2155,
+      "step": 725
+    },
+    {
+      "epoch": 2.7150837988826817,
+      "grad_norm": 1.4306650161743164,
+      "learning_rate": 1.4170128302871094e-05,
+      "loss": 0.1841,
+      "step": 730
+    },
+    {
+      "epoch": 2.733705772811918,
+      "grad_norm": 1.0107722282409668,
+      "learning_rate": 1.3985968183821347e-05,
+      "loss": 0.1647,
+      "step": 735
+    },
+    {
+      "epoch": 2.7523277467411544,
+      "grad_norm": 1.1376290321350098,
+      "learning_rate": 1.3801961492574229e-05,
+      "loss": 0.2009,
+      "step": 740
+    },
+    {
+      "epoch": 2.770949720670391,
+      "grad_norm": 1.2442623376846313,
+      "learning_rate": 1.361813607021109e-05,
+      "loss": 0.2114,
+      "step": 745
+    },
+    {
+      "epoch": 2.7895716945996276,
+      "grad_norm": 1.1218327283859253,
+      "learning_rate": 1.3434519730386483e-05,
+      "loss": 0.1969,
+      "step": 750
+    },
+    {
+      "epoch": 2.808193668528864,
+      "grad_norm": 1.1548234224319458,
+      "learning_rate": 1.3251140255119744e-05,
+      "loss": 0.2084,
+      "step": 755
+    },
+    {
+      "epoch": 2.826815642458101,
+      "grad_norm": 1.0886045694351196,
+      "learning_rate": 1.306802539059151e-05,
+      "loss": 0.2111,
+      "step": 760
+    },
+    {
+      "epoch": 2.845437616387337,
+      "grad_norm": 1.0762062072753906,
+      "learning_rate": 1.2885202842945563e-05,
+      "loss": 0.2178,
+      "step": 765
+    },
+    {
+      "epoch": 2.8640595903165735,
+      "grad_norm": 1.115242838859558,
+      "learning_rate": 1.2702700274096744e-05,
+      "loss": 0.2133,
+      "step": 770
+    },
+    {
+      "epoch": 2.88268156424581,
+      "grad_norm": 1.0876168012619019,
+      "learning_rate": 1.2520545297545603e-05,
+      "loss": 0.1729,
+      "step": 775
+    },
+    {
+      "epoch": 2.9013035381750467,
+      "grad_norm": 1.0145092010498047,
+      "learning_rate": 1.233876547420033e-05,
+      "loss": 0.1832,
+      "step": 780
+    },
+    {
+      "epoch": 2.919925512104283,
+      "grad_norm": 1.134987473487854,
+      "learning_rate": 1.2157388308206669e-05,
+      "loss": 0.2141,
+      "step": 785
+    },
+    {
+      "epoch": 2.9385474860335195,
+      "grad_norm": 1.1757721900939941,
+      "learning_rate": 1.1976441242786394e-05,
+      "loss": 0.206,
+      "step": 790
+    },
+    {
+      "epoch": 2.9571694599627563,
+      "grad_norm": 1.1044893264770508,
+      "learning_rate": 1.1795951656085042e-05,
+      "loss": 0.1804,
+      "step": 795
+    },
+    {
+      "epoch": 2.9757914338919926,
+      "grad_norm": 1.0237022638320923,
+      "learning_rate": 1.1615946857029453e-05,
+      "loss": 0.1853,
+      "step": 800
+    },
+    {
+      "epoch": 2.994413407821229,
+      "grad_norm": 1.2020282745361328,
+      "learning_rate": 1.1436454081195811e-05,
+      "loss": 0.1908,
+      "step": 805
+    },
+    {
+      "epoch": 3.011173184357542,
+      "grad_norm": 0.9971558451652527,
+      "learning_rate": 1.1257500486688762e-05,
+      "loss": 0.1511,
+      "step": 810
+    },
+    {
+      "epoch": 3.0297951582867784,
+      "grad_norm": 1.146209955215454,
+      "learning_rate": 1.1079113150032282e-05,
+      "loss": 0.1326,
+      "step": 815
+    },
+    {
+      "epoch": 3.0484171322160147,
+      "grad_norm": 0.9054371118545532,
+      "learning_rate": 1.0901319062072841e-05,
+      "loss": 0.132,
+      "step": 820
+    },
+    {
+      "epoch": 3.0670391061452515,
+      "grad_norm": 1.019821286201477,
+      "learning_rate": 1.0724145123895605e-05,
+      "loss": 0.158,
+      "step": 825
+    },
+    {
+      "epoch": 3.085661080074488,
+      "grad_norm": 1.2374272346496582,
+      "learning_rate": 1.0547618142754127e-05,
+      "loss": 0.1405,
+      "step": 830
+    },
+    {
+      "epoch": 3.1042830540037243,
+      "grad_norm": 1.1176825761795044,
+      "learning_rate": 1.0371764828014318e-05,
+      "loss": 0.1397,
+      "step": 835
+    },
+    {
+      "epoch": 3.122905027932961,
+      "grad_norm": 0.9421104788780212,
+      "learning_rate": 1.0196611787113173e-05,
+      "loss": 0.129,
+      "step": 840
+    },
+    {
+      "epoch": 3.1415270018621975,
+      "grad_norm": 1.510794997215271,
+      "learning_rate": 1.0022185521532952e-05,
+      "loss": 0.1438,
+      "step": 845
+    },
+    {
+      "epoch": 3.160148975791434,
+      "grad_norm": 1.0882729291915894,
+      "learning_rate": 9.848512422791386e-06,
+      "loss": 0.1364,
+      "step": 850
+    },
+    {
+      "epoch": 3.17877094972067,
+      "grad_norm": 1.111037015914917,
+      "learning_rate": 9.67561876844849e-06,
+      "loss": 0.1425,
+      "step": 855
+    },
+    {
+      "epoch": 3.197392923649907,
+      "grad_norm": 1.1784346103668213,
+      "learning_rate": 9.503530718130669e-06,
+      "loss": 0.1347,
+      "step": 860
+    },
+    {
+      "epoch": 3.2160148975791434,
+      "grad_norm": 1.0116277933120728,
+      "learning_rate": 9.332274309572628e-06,
+      "loss": 0.1368,
+      "step": 865
+    },
+    {
+      "epoch": 3.2346368715083798,
+      "grad_norm": 1.2434337139129639,
+      "learning_rate": 9.161875454677761e-06,
+      "loss": 0.1318,
+      "step": 870
+    },
+    {
+      "epoch": 3.2532588454376166,
+      "grad_norm": 1.1153370141983032,
+      "learning_rate": 8.992359935597526e-06,
+      "loss": 0.1269,
+      "step": 875
+    },
+    {
+      "epoch": 3.271880819366853,
+      "grad_norm": 0.9924963712692261,
+      "learning_rate": 8.823753400830514e-06,
+      "loss": 0.143,
+      "step": 880
+    },
+    {
+      "epoch": 3.2905027932960893,
+      "grad_norm": 0.8898366689682007,
+      "learning_rate": 8.656081361341687e-06,
+      "loss": 0.1156,
+      "step": 885
+    },
+    {
+      "epoch": 3.3091247672253257,
+      "grad_norm": 0.9309813380241394,
+      "learning_rate": 8.48936918670246e-06,
+      "loss": 0.1216,
+      "step": 890
+    },
+    {
+      "epoch": 3.3277467411545625,
+      "grad_norm": 1.0335708856582642,
+      "learning_rate": 8.323642101252165e-06,
+      "loss": 0.1386,
+      "step": 895
+    },
+    {
+      "epoch": 3.346368715083799,
+      "grad_norm": 0.956021249294281,
+      "learning_rate": 8.158925180281494e-06,
+      "loss": 0.1419,
+      "step": 900
+    },
+    {
+      "epoch": 3.364990689013035,
+      "grad_norm": 1.0639731884002686,
+      "learning_rate": 7.995243346238472e-06,
+      "loss": 0.147,
+      "step": 905
+    },
+    {
+      "epoch": 3.383612662942272,
+      "grad_norm": 1.1223934888839722,
+      "learning_rate": 7.83262136495759e-06,
+      "loss": 0.1319,
+      "step": 910
+    },
+    {
+      "epoch": 3.4022346368715084,
+      "grad_norm": 0.9859718084335327,
+      "learning_rate": 7.671083841912603e-06,
+      "loss": 0.1189,
+      "step": 915
+    },
+    {
+      "epoch": 3.4208566108007448,
+      "grad_norm": 0.8468735218048096,
+      "learning_rate": 7.510655218493627e-06,
+      "loss": 0.123,
+      "step": 920
+    },
+    {
+      "epoch": 3.439478584729981,
+      "grad_norm": 0.8232627511024475,
+      "learning_rate": 7.351359768309017e-06,
+      "loss": 0.1216,
+      "step": 925
+    },
+    {
+      "epoch": 3.458100558659218,
+      "grad_norm": 0.9500121474266052,
+      "learning_rate": 7.193221593512663e-06,
+      "loss": 0.1255,
+      "step": 930
+    },
+    {
+      "epoch": 3.4767225325884543,
+      "grad_norm": 1.1072030067443848,
+      "learning_rate": 7.036264621157241e-06,
+      "loss": 0.1173,
+      "step": 935
+    },
+    {
+      "epoch": 3.4953445065176907,
+      "grad_norm": 1.1253328323364258,
+      "learning_rate": 6.880512599573903e-06,
+      "loss": 0.1261,
+      "step": 940
+    },
+    {
+      "epoch": 3.5139664804469275,
+      "grad_norm": 1.1017041206359863,
+      "learning_rate": 6.725989094779063e-06,
+      "loss": 0.1202,
+      "step": 945
+    },
+    {
+      "epoch": 3.532588454376164,
+      "grad_norm": 1.0356934070587158,
+      "learning_rate": 6.572717486908738e-06,
+      "loss": 0.1152,
+      "step": 950
+    },
+    {
+      "epoch": 3.5512104283054002,
+      "grad_norm": 0.9527298808097839,
+      "learning_rate": 6.420720966681052e-06,
+      "loss": 0.1141,
+      "step": 955
+    },
+    {
+      "epoch": 3.5698324022346366,
+      "grad_norm": 0.8534860610961914,
+      "learning_rate": 6.270022531887305e-06,
+      "loss": 0.1157,
+      "step": 960
+    },
+    {
+      "epoch": 3.5884543761638734,
+      "grad_norm": 1.1221200227737427,
+      "learning_rate": 6.120644983912383e-06,
+      "loss": 0.109,
+      "step": 965
+    },
+    {
+      "epoch": 3.60707635009311,
+      "grad_norm": 1.075600028038025,
+      "learning_rate": 5.972610924284732e-06,
+      "loss": 0.1287,
+      "step": 970
+    },
+    {
+      "epoch": 3.6256983240223466,
+      "grad_norm": 1.0059423446655273,
+      "learning_rate": 5.825942751256677e-06,
+      "loss": 0.1256,
+      "step": 975
+    },
+    {
+      "epoch": 3.644320297951583,
+      "grad_norm": 1.0131361484527588,
+      "learning_rate": 5.680662656415441e-06,
+      "loss": 0.127,
+      "step": 980
+    },
+    {
+      "epoch": 3.6629422718808193,
+      "grad_norm": 0.9724953174591064,
+      "learning_rate": 5.536792621325479e-06,
+      "loss": 0.1285,
+      "step": 985
+    },
+    {
+      "epoch": 3.6815642458100557,
+      "grad_norm": 1.1040209531784058,
+      "learning_rate": 5.394354414202529e-06,
+      "loss": 0.1177,
+      "step": 990
+    },
+    {
+      "epoch": 3.7001862197392925,
+      "grad_norm": 0.9747113585472107,
+      "learning_rate": 5.253369586620004e-06,
+      "loss": 0.1071,
+      "step": 995
+    },
+    {
+      "epoch": 3.718808193668529,
+      "grad_norm": 0.9118765592575073,
+      "learning_rate": 5.113859470248123e-06,
+      "loss": 0.1176,
+      "step": 1000
+    },
+    {
+      "epoch": 3.7374301675977653,
+      "grad_norm": 0.9922019839286804,
+      "learning_rate": 4.975845173626351e-06,
+      "loss": 0.1011,
+      "step": 1005
+    },
+    {
+      "epoch": 3.756052141527002,
+      "grad_norm": 0.9839317798614502,
+      "learning_rate": 4.8393475789695325e-06,
+      "loss": 0.1086,
+      "step": 1010
+    },
+    {
+      "epoch": 3.7746741154562384,
+      "grad_norm": 1.049396276473999,
+      "learning_rate": 4.7043873390083654e-06,
+      "loss": 0.1114,
+      "step": 1015
+    },
+    {
+      "epoch": 3.793296089385475,
+      "grad_norm": 1.1213620901107788,
+      "learning_rate": 4.570984873864519e-06,
+      "loss": 0.0996,
+      "step": 1020
+    },
+    {
+      "epoch": 3.811918063314711,
+      "grad_norm": 0.9337913393974304,
+      "learning_rate": 4.439160367960966e-06,
+      "loss": 0.0884,
+      "step": 1025
+    },
+    {
+      "epoch": 3.830540037243948,
+      "grad_norm": 1.1189533472061157,
+      "learning_rate": 4.308933766968003e-06,
+      "loss": 0.1128,
+      "step": 1030
+    },
+    {
+      "epoch": 3.8491620111731844,
+      "grad_norm": 0.8592391014099121,
+      "learning_rate": 4.180324774785356e-06,
+      "loss": 0.1036,
+      "step": 1035
+    },
+    {
+      "epoch": 3.8677839851024207,
+      "grad_norm": 0.9491106271743774,
+      "learning_rate": 4.053352850560923e-06,
+      "loss": 0.1053,
+      "step": 1040
+    },
+    {
+      "epoch": 3.8864059590316575,
+      "grad_norm": 1.7115939855575562,
+      "learning_rate": 3.928037205746451e-06,
+      "loss": 0.1079,
+      "step": 1045
+    },
+    {
+      "epoch": 3.905027932960894,
+      "grad_norm": 0.965495228767395,
+      "learning_rate": 3.8043968011908277e-06,
+      "loss": 0.1084,
+      "step": 1050
+    },
+    {
+      "epoch": 3.9236499068901303,
+      "grad_norm": 0.9409112334251404,
+      "learning_rate": 3.682450344271157e-06,
+      "loss": 0.0952,
+      "step": 1055
+    },
+    {
+      "epoch": 3.9422718808193666,
+      "grad_norm": 0.8451502323150635,
+      "learning_rate": 3.562216286062271e-06,
+      "loss": 0.1157,
+      "step": 1060
+    },
+    {
+      "epoch": 3.9608938547486034,
+      "grad_norm": 1.0194039344787598,
+      "learning_rate": 3.4437128185449872e-06,
+      "loss": 0.0916,
+      "step": 1065
+    },
+    {
+      "epoch": 3.97951582867784,
+      "grad_norm": 0.8282318115234375,
+      "learning_rate": 3.326957871853587e-06,
+      "loss": 0.0934,
+      "step": 1070
+    },
+    {
+      "epoch": 3.998137802607076,
+      "grad_norm": 0.8793200850486755,
+      "learning_rate": 3.211969111562886e-06,
+      "loss": 0.1086,
+      "step": 1075
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1345,
+  "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": 1.5409152515228303e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

53_128_e5_3e-5/checkpoint-1076/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ef384a0e7f290964c504f65dcbd3ddbb9bba343263178891e1dfd052c620a7a4
+size 7736

53_128_e5_3e-5/checkpoint-1076/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)

53_128_e5_3e-5/checkpoint-1345/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

53_128_e5_3e-5/checkpoint-1345/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": [
+    "q_proj",
+    "k_proj",
+    "down_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "v_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

53_128_e5_3e-5/checkpoint-1345/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b76a3ea3c0dc3205531cfe680402e60109a7bd607c532c2ad74475994f14a690
+size 791751704

53_128_e5_3e-5/checkpoint-1345/latest

@@ -0,0 +1 @@
+global_step1345

53_128_e5_3e-5/checkpoint-1345/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:87c3880316957cf0878393eaa13b2330b73329dffb88089751b7b647f519702b
+size 15920

53_128_e5_3e-5/checkpoint-1345/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac76d77790dd62ff1fb1cfbca3d727894853aec8b24475ddf81d8001a10cff82
+size 15920

53_128_e5_3e-5/checkpoint-1345/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:470fca0ba4e47f78531a0538742d65b96262991883561e18b5004b6551faee5a
+size 15920

53_128_e5_3e-5/checkpoint-1345/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1652ecabd25a465efc884f3055afd7aac072898a99441b09f483bf15ba0ec191
+size 15920

53_128_e5_3e-5/checkpoint-1345/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1306c39e99472a90ed0edd9a423fd67a3ca254f9e1e31f4db82c510e1bec08c5
+size 15920

53_128_e5_3e-5/checkpoint-1345/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cea67841aecd3c465c51fcb0663a66411acd362ecfc393d8529b59b915765c67
+size 15920

53_128_e5_3e-5/checkpoint-1345/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9b2d2b83b5b73b10cfbd0b6804b36939232cd3ce4e5efd56ab393166de1093d1
+size 15920

53_128_e5_3e-5/checkpoint-1345/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:89c67e709d2b660b3f1a9b0ce51486dfe2d59864b180b7a7f057d968f7b5912c
+size 15920

53_128_e5_3e-5/checkpoint-1345/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0ebedccd0711d0b835e492d7965904368174ec8274bc06fda0e81c4e893a2ae1
+size 1064

53_128_e5_3e-5/checkpoint-1345/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
+  }
+}

53_128_e5_3e-5/checkpoint-1345/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
+}

53_128_e5_3e-5/checkpoint-1345/trainer_state.json

@@ -0,0 +1,1917 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 1345,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0186219739292365,
+      "grad_norm": 1.397485375404358,
+      "learning_rate": 1.7647058823529412e-06,
+      "loss": 1.3045,
+      "step": 5
+    },
+    {
+      "epoch": 0.037243947858473,
+      "grad_norm": 0.9164593815803528,
+      "learning_rate": 3.970588235294118e-06,
+      "loss": 1.3238,
+      "step": 10
+    },
+    {
+      "epoch": 0.055865921787709494,
+      "grad_norm": 0.7285172343254089,
+      "learning_rate": 6.176470588235294e-06,
+      "loss": 1.3402,
+      "step": 15
+    },
+    {
+      "epoch": 0.074487895716946,
+      "grad_norm": 0.7771405577659607,
+      "learning_rate": 8.382352941176472e-06,
+      "loss": 1.2444,
+      "step": 20
+    },
+    {
+      "epoch": 0.0931098696461825,
+      "grad_norm": 0.590904951095581,
+      "learning_rate": 1.0588235294117648e-05,
+      "loss": 1.2908,
+      "step": 25
+    },
+    {
+      "epoch": 0.11173184357541899,
+      "grad_norm": 0.68792724609375,
+      "learning_rate": 1.2794117647058824e-05,
+      "loss": 1.2953,
+      "step": 30
+    },
+    {
+      "epoch": 0.1303538175046555,
+      "grad_norm": 0.6625350117683411,
+      "learning_rate": 1.5e-05,
+      "loss": 1.192,
+      "step": 35
+    },
+    {
+      "epoch": 0.148975791433892,
+      "grad_norm": 0.5564359426498413,
+      "learning_rate": 1.7205882352941175e-05,
+      "loss": 1.1406,
+      "step": 40
+    },
+    {
+      "epoch": 0.16759776536312848,
+      "grad_norm": 0.4591367542743683,
+      "learning_rate": 1.9411764705882355e-05,
+      "loss": 1.181,
+      "step": 45
+    },
+    {
+      "epoch": 0.186219739292365,
+      "grad_norm": 0.5679731369018555,
+      "learning_rate": 2.161764705882353e-05,
+      "loss": 1.1935,
+      "step": 50
+    },
+    {
+      "epoch": 0.2048417132216015,
+      "grad_norm": 0.4359203577041626,
+      "learning_rate": 2.3823529411764704e-05,
+      "loss": 1.1527,
+      "step": 55
+    },
+    {
+      "epoch": 0.22346368715083798,
+      "grad_norm": 0.49340277910232544,
+      "learning_rate": 2.6029411764705883e-05,
+      "loss": 1.1635,
+      "step": 60
+    },
+    {
+      "epoch": 0.24208566108007448,
+      "grad_norm": 0.4740353226661682,
+      "learning_rate": 2.823529411764706e-05,
+      "loss": 1.1883,
+      "step": 65
+    },
+    {
+      "epoch": 0.260707635009311,
+      "grad_norm": 0.6028391122817993,
+      "learning_rate": 2.9999954608033783e-05,
+      "loss": 1.1328,
+      "step": 70
+    },
+    {
+      "epoch": 0.27932960893854747,
+      "grad_norm": 0.5162932872772217,
+      "learning_rate": 2.9998365918062082e-05,
+      "loss": 1.0934,
+      "step": 75
+    },
+    {
+      "epoch": 0.297951582867784,
+      "grad_norm": 0.5382624864578247,
+      "learning_rate": 2.9994507904496206e-05,
+      "loss": 1.1019,
+      "step": 80
+    },
+    {
+      "epoch": 0.3165735567970205,
+      "grad_norm": 0.5565810799598694,
+      "learning_rate": 2.998838115107183e-05,
+      "loss": 1.0906,
+      "step": 85
+    },
+    {
+      "epoch": 0.33519553072625696,
+      "grad_norm": 0.5212501287460327,
+      "learning_rate": 2.997998658479568e-05,
+      "loss": 1.0724,
+      "step": 90
+    },
+    {
+      "epoch": 0.3538175046554935,
+      "grad_norm": 0.5489919185638428,
+      "learning_rate": 2.9969325475805274e-05,
+      "loss": 1.0658,
+      "step": 95
+    },
+    {
+      "epoch": 0.37243947858473,
+      "grad_norm": 0.6241554021835327,
+      "learning_rate": 2.995639943717676e-05,
+      "loss": 1.059,
+      "step": 100
+    },
+    {
+      "epoch": 0.39106145251396646,
+      "grad_norm": 0.563462495803833,
+      "learning_rate": 2.9941210424680813e-05,
+      "loss": 1.057,
+      "step": 105
+    },
+    {
+      "epoch": 0.409683426443203,
+      "grad_norm": 0.5747496485710144,
+      "learning_rate": 2.9923760736486766e-05,
+      "loss": 1.0564,
+      "step": 110
+    },
+    {
+      "epoch": 0.42830540037243947,
+      "grad_norm": 0.5738973021507263,
+      "learning_rate": 2.9904053012814848e-05,
+      "loss": 1.0316,
+      "step": 115
+    },
+    {
+      "epoch": 0.44692737430167595,
+      "grad_norm": 0.6159545183181763,
+      "learning_rate": 2.988209023553672e-05,
+      "loss": 0.995,
+      "step": 120
+    },
+    {
+      "epoch": 0.4655493482309125,
+      "grad_norm": 0.6447931528091431,
+      "learning_rate": 2.9857875727724304e-05,
+      "loss": 0.9692,
+      "step": 125
+    },
+    {
+      "epoch": 0.48417132216014896,
+      "grad_norm": 0.7060043215751648,
+      "learning_rate": 2.9831413153146988e-05,
+      "loss": 0.9569,
+      "step": 130
+    },
+    {
+      "epoch": 0.5027932960893855,
+      "grad_norm": 0.7400388717651367,
+      "learning_rate": 2.9802706515717272e-05,
+      "loss": 0.9378,
+      "step": 135
+    },
+    {
+      "epoch": 0.521415270018622,
+      "grad_norm": 0.7127872705459595,
+      "learning_rate": 2.9771760158884972e-05,
+      "loss": 0.8923,
+      "step": 140
+    },
+    {
+      "epoch": 0.5400372439478585,
+      "grad_norm": 0.8123269081115723,
+      "learning_rate": 2.9738578764980025e-05,
+      "loss": 0.9285,
+      "step": 145
+    },
+    {
+      "epoch": 0.5586592178770949,
+      "grad_norm": 0.8044330477714539,
+      "learning_rate": 2.9703167354504027e-05,
+      "loss": 0.885,
+      "step": 150
+    },
+    {
+      "epoch": 0.5772811918063314,
+      "grad_norm": 0.7073454856872559,
+      "learning_rate": 2.966553128537062e-05,
+      "loss": 0.9022,
+      "step": 155
+    },
+    {
+      "epoch": 0.595903165735568,
+      "grad_norm": 0.725287914276123,
+      "learning_rate": 2.9625676252094797e-05,
+      "loss": 0.8863,
+      "step": 160
+    },
+    {
+      "epoch": 0.6145251396648045,
+      "grad_norm": 0.843565046787262,
+      "learning_rate": 2.9583608284931317e-05,
+      "loss": 0.928,
+      "step": 165
+    },
+    {
+      "epoch": 0.633147113594041,
+      "grad_norm": 0.8186341524124146,
+      "learning_rate": 2.953933374896227e-05,
+      "loss": 0.8967,
+      "step": 170
+    },
+    {
+      "epoch": 0.6517690875232774,
+      "grad_norm": 0.8052655458450317,
+      "learning_rate": 2.949285934313405e-05,
+      "loss": 0.8766,
+      "step": 175
+    },
+    {
+      "epoch": 0.6703910614525139,
+      "grad_norm": 1.142082691192627,
+      "learning_rate": 2.9444192099243733e-05,
+      "loss": 0.8402,
+      "step": 180
+    },
+    {
+      "epoch": 0.6890130353817505,
+      "grad_norm": 0.9169373512268066,
+      "learning_rate": 2.939333938087515e-05,
+      "loss": 0.8426,
+      "step": 185
+    },
+    {
+      "epoch": 0.707635009310987,
+      "grad_norm": 0.8522142171859741,
+      "learning_rate": 2.9340308882284747e-05,
+      "loss": 0.8288,
+      "step": 190
+    },
+    {
+      "epoch": 0.7262569832402235,
+      "grad_norm": 0.8226320147514343,
+      "learning_rate": 2.92851086272374e-05,
+      "loss": 0.8252,
+      "step": 195
+    },
+    {
+      "epoch": 0.74487895716946,
+      "grad_norm": 0.840149462223053,
+      "learning_rate": 2.9227746967792392e-05,
+      "loss": 0.7888,
+      "step": 200
+    },
+    {
+      "epoch": 0.7635009310986964,
+      "grad_norm": 1.0344595909118652,
+      "learning_rate": 2.916823258303968e-05,
+      "loss": 0.7889,
+      "step": 205
+    },
+    {
+      "epoch": 0.7821229050279329,
+      "grad_norm": 0.9850447177886963,
+      "learning_rate": 2.9106574477786748e-05,
+      "loss": 0.7634,
+      "step": 210
+    },
+    {
+      "epoch": 0.8007448789571695,
+      "grad_norm": 0.9077547192573547,
+      "learning_rate": 2.9042781981196095e-05,
+      "loss": 0.7372,
+      "step": 215
+    },
+    {
+      "epoch": 0.819366852886406,
+      "grad_norm": 0.8616479635238647,
+      "learning_rate": 2.897686474537373e-05,
+      "loss": 0.7238,
+      "step": 220
+    },
+    {
+      "epoch": 0.8379888268156425,
+      "grad_norm": 0.9395949244499207,
+      "learning_rate": 2.890883274390872e-05,
+      "loss": 0.6952,
+      "step": 225
+    },
+    {
+      "epoch": 0.8566108007448789,
+      "grad_norm": 0.9580796957015991,
+      "learning_rate": 2.8838696270364183e-05,
+      "loss": 0.6983,
+      "step": 230
+    },
+    {
+      "epoch": 0.8752327746741154,
+      "grad_norm": 0.8633882403373718,
+      "learning_rate": 2.8766465936719785e-05,
+      "loss": 0.7479,
+      "step": 235
+    },
+    {
+      "epoch": 0.8938547486033519,
+      "grad_norm": 1.0677725076675415,
+      "learning_rate": 2.869215267176612e-05,
+      "loss": 0.7132,
+      "step": 240
+    },
+    {
+      "epoch": 0.9124767225325885,
+      "grad_norm": 1.0318955183029175,
+      "learning_rate": 2.8615767719451125e-05,
+      "loss": 0.6744,
+      "step": 245
+    },
+    {
+      "epoch": 0.931098696461825,
+      "grad_norm": 0.8933286070823669,
+      "learning_rate": 2.8537322637178816e-05,
+      "loss": 0.705,
+      "step": 250
+    },
+    {
+      "epoch": 0.9497206703910615,
+      "grad_norm": 0.8829292058944702,
+      "learning_rate": 2.8456829294060608e-05,
+      "loss": 0.7091,
+      "step": 255
+    },
+    {
+      "epoch": 0.9683426443202979,
+      "grad_norm": 1.1101434230804443,
+      "learning_rate": 2.837429986911944e-05,
+      "loss": 0.6532,
+      "step": 260
+    },
+    {
+      "epoch": 0.9869646182495344,
+      "grad_norm": 1.070557713508606,
+      "learning_rate": 2.828974684944707e-05,
+      "loss": 0.6493,
+      "step": 265
+    },
+    {
+      "epoch": 1.0037243947858474,
+      "grad_norm": 1.0655916929244995,
+      "learning_rate": 2.820318302831467e-05,
+      "loss": 0.6279,
+      "step": 270
+    },
+    {
+      "epoch": 1.0223463687150838,
+      "grad_norm": 1.0981754064559937,
+      "learning_rate": 2.8114621503237186e-05,
+      "loss": 0.5955,
+      "step": 275
+    },
+    {
+      "epoch": 1.0409683426443204,
+      "grad_norm": 1.099587082862854,
+      "learning_rate": 2.8024075673991597e-05,
+      "loss": 0.6128,
+      "step": 280
+    },
+    {
+      "epoch": 1.0595903165735567,
+      "grad_norm": 1.0760383605957031,
+      "learning_rate": 2.7931559240589487e-05,
+      "loss": 0.5807,
+      "step": 285
+    },
+    {
+      "epoch": 1.0782122905027933,
+      "grad_norm": 1.028070330619812,
+      "learning_rate": 2.7837086201204128e-05,
+      "loss": 0.5667,
+      "step": 290
+    },
+    {
+      "epoch": 1.0968342644320297,
+      "grad_norm": 1.0326428413391113,
+      "learning_rate": 2.7740670850052543e-05,
+      "loss": 0.5615,
+      "step": 295
+    },
+    {
+      "epoch": 1.1154562383612663,
+      "grad_norm": 1.1273221969604492,
+      "learning_rate": 2.764232777523269e-05,
+      "loss": 0.5597,
+      "step": 300
+    },
+    {
+      "epoch": 1.1340782122905029,
+      "grad_norm": 0.9928714036941528,
+      "learning_rate": 2.7542071856516215e-05,
+      "loss": 0.5426,
+      "step": 305
+    },
+    {
+      "epoch": 1.1527001862197392,
+      "grad_norm": 1.1522924900054932,
+      "learning_rate": 2.7439918263097097e-05,
+      "loss": 0.5586,
+      "step": 310
+    },
+    {
+      "epoch": 1.1713221601489758,
+      "grad_norm": 1.0707367658615112,
+      "learning_rate": 2.7335882451296438e-05,
+      "loss": 0.5964,
+      "step": 315
+    },
+    {
+      "epoch": 1.1899441340782122,
+      "grad_norm": 1.0582071542739868,
+      "learning_rate": 2.7229980162223876e-05,
+      "loss": 0.5463,
+      "step": 320
+    },
+    {
+      "epoch": 1.2085661080074488,
+      "grad_norm": 1.109670639038086,
+      "learning_rate": 2.7122227419395885e-05,
+      "loss": 0.5033,
+      "step": 325
+    },
+    {
+      "epoch": 1.2271880819366854,
+      "grad_norm": 1.0573663711547852,
+      "learning_rate": 2.7012640526311324e-05,
+      "loss": 0.5443,
+      "step": 330
+    },
+    {
+      "epoch": 1.2458100558659218,
+      "grad_norm": 1.0600390434265137,
+      "learning_rate": 2.6901236063984646e-05,
+      "loss": 0.5143,
+      "step": 335
+    },
+    {
+      "epoch": 1.2644320297951583,
+      "grad_norm": 1.339310646057129,
+      "learning_rate": 2.6788030888437118e-05,
+      "loss": 0.5324,
+      "step": 340
+    },
+    {
+      "epoch": 1.2830540037243947,
+      "grad_norm": 1.5570026636123657,
+      "learning_rate": 2.6673042128146414e-05,
+      "loss": 0.5467,
+      "step": 345
+    },
+    {
+      "epoch": 1.3016759776536313,
+      "grad_norm": 1.0688986778259277,
+      "learning_rate": 2.6556287181455013e-05,
+      "loss": 0.4999,
+      "step": 350
+    },
+    {
+      "epoch": 1.3202979515828677,
+      "grad_norm": 1.2187882661819458,
+      "learning_rate": 2.643778371393773e-05,
+      "loss": 0.507,
+      "step": 355
+    },
+    {
+      "epoch": 1.3389199255121043,
+      "grad_norm": 1.149132490158081,
+      "learning_rate": 2.6317549655728846e-05,
+      "loss": 0.4686,
+      "step": 360
+    },
+    {
+      "epoch": 1.3575418994413408,
+      "grad_norm": 1.230714201927185,
+      "learning_rate": 2.6195603198809175e-05,
+      "loss": 0.488,
+      "step": 365
+    },
+    {
+      "epoch": 1.3761638733705772,
+      "grad_norm": 1.1980931758880615,
+      "learning_rate": 2.6071962794253548e-05,
+      "loss": 0.5051,
+      "step": 370
+    },
+    {
+      "epoch": 1.3947858472998138,
+      "grad_norm": 1.1617096662521362,
+      "learning_rate": 2.594664714943908e-05,
+      "loss": 0.4775,
+      "step": 375
+    },
+    {
+      "epoch": 1.4134078212290504,
+      "grad_norm": 1.2936596870422363,
+      "learning_rate": 2.5819675225214644e-05,
+      "loss": 0.4976,
+      "step": 380
+    },
+    {
+      "epoch": 1.4320297951582868,
+      "grad_norm": 1.230708122253418,
+      "learning_rate": 2.5691066233032e-05,
+      "loss": 0.4756,
+      "step": 385
+    },
+    {
+      "epoch": 1.4506517690875234,
+      "grad_norm": 0.9902650713920593,
+      "learning_rate": 2.5560839632039032e-05,
+      "loss": 0.4514,
+      "step": 390
+    },
+    {
+      "epoch": 1.4692737430167597,
+      "grad_norm": 1.0236945152282715,
+      "learning_rate": 2.5429015126135482e-05,
+      "loss": 0.4197,
+      "step": 395
+    },
+    {
+      "epoch": 1.4878957169459963,
+      "grad_norm": 1.0958354473114014,
+      "learning_rate": 2.5295612660991632e-05,
+      "loss": 0.4475,
+      "step": 400
+    },
+    {
+      "epoch": 1.5065176908752327,
+      "grad_norm": 1.2085777521133423,
+      "learning_rate": 2.516065242103047e-05,
+      "loss": 0.4189,
+      "step": 405
+    },
+    {
+      "epoch": 1.5251396648044693,
+      "grad_norm": 1.070729374885559,
+      "learning_rate": 2.5024154826373655e-05,
+      "loss": 0.4804,
+      "step": 410
+    },
+    {
+      "epoch": 1.5437616387337059,
+      "grad_norm": 1.1738492250442505,
+      "learning_rate": 2.488614052975188e-05,
+      "loss": 0.4061,
+      "step": 415
+    },
+    {
+      "epoch": 1.5623836126629422,
+      "grad_norm": 1.128957986831665,
+      "learning_rate": 2.474663041338e-05,
+      "loss": 0.4125,
+      "step": 420
+    },
+    {
+      "epoch": 1.5810055865921788,
+      "grad_norm": 1.1465480327606201,
+      "learning_rate": 2.4605645585797476e-05,
+      "loss": 0.4208,
+      "step": 425
+    },
+    {
+      "epoch": 1.5996275605214154,
+      "grad_norm": 1.1640135049819946,
+      "learning_rate": 2.4463207378674525e-05,
+      "loss": 0.4578,
+      "step": 430
+    },
+    {
+      "epoch": 1.6182495344506518,
+      "grad_norm": 1.242473840713501,
+      "learning_rate": 2.431933734358456e-05,
+      "loss": 0.406,
+      "step": 435
+    },
+    {
+      "epoch": 1.6368715083798882,
+      "grad_norm": 1.2133651971817017,
+      "learning_rate": 2.417405724874333e-05,
+      "loss": 0.4357,
+      "step": 440
+    },
+    {
+      "epoch": 1.6554934823091247,
+      "grad_norm": 1.2609267234802246,
+      "learning_rate": 2.4027389075715272e-05,
+      "loss": 0.427,
+      "step": 445
+    },
+    {
+      "epoch": 1.6741154562383613,
+      "grad_norm": 1.1197216510772705,
+      "learning_rate": 2.3879355016087623e-05,
+      "loss": 0.4025,
+      "step": 450
+    },
+    {
+      "epoch": 1.6927374301675977,
+      "grad_norm": 1.1437435150146484,
+      "learning_rate": 2.3729977468112696e-05,
+      "loss": 0.3899,
+      "step": 455
+    },
+    {
+      "epoch": 1.7113594040968343,
+      "grad_norm": 1.1721247434616089,
+      "learning_rate": 2.357927903331895e-05,
+      "loss": 0.4361,
+      "step": 460
+    },
+    {
+      "epoch": 1.7299813780260709,
+      "grad_norm": 1.2531391382217407,
+      "learning_rate": 2.342728251309126e-05,
+      "loss": 0.3918,
+      "step": 465
+    },
+    {
+      "epoch": 1.7486033519553073,
+      "grad_norm": 1.160923957824707,
+      "learning_rate": 2.327401090522094e-05,
+      "loss": 0.3994,
+      "step": 470
+    },
+    {
+      "epoch": 1.7672253258845436,
+      "grad_norm": 1.3061710596084595,
+      "learning_rate": 2.3119487400426097e-05,
+      "loss": 0.4156,
+      "step": 475
+    },
+    {
+      "epoch": 1.7858472998137802,
+      "grad_norm": 1.0728996992111206,
+      "learning_rate": 2.2963735378842757e-05,
+      "loss": 0.3827,
+      "step": 480
+    },
+    {
+      "epoch": 1.8044692737430168,
+      "grad_norm": 1.153071403503418,
+      "learning_rate": 2.2806778406487337e-05,
+      "loss": 0.399,
+      "step": 485
+    },
+    {
+      "epoch": 1.8230912476722532,
+      "grad_norm": 1.1256935596466064,
+      "learning_rate": 2.2648640231690987e-05,
+      "loss": 0.3721,
+      "step": 490
+    },
+    {
+      "epoch": 1.8417132216014898,
+      "grad_norm": 1.1391295194625854,
+      "learning_rate": 2.2489344781506375e-05,
+      "loss": 0.3625,
+      "step": 495
+    },
+    {
+      "epoch": 1.8603351955307263,
+      "grad_norm": 1.235418438911438,
+      "learning_rate": 2.2328916158087396e-05,
+      "loss": 0.3652,
+      "step": 500
+    },
+    {
+      "epoch": 1.8789571694599627,
+      "grad_norm": 1.1151893138885498,
+      "learning_rate": 2.2167378635042414e-05,
+      "loss": 0.3526,
+      "step": 505
+    },
+    {
+      "epoch": 1.8975791433891993,
+      "grad_norm": 1.1720143556594849,
+      "learning_rate": 2.2004756653761533e-05,
+      "loss": 0.363,
+      "step": 510
+    },
+    {
+      "epoch": 1.916201117318436,
+      "grad_norm": 1.219879388809204,
+      "learning_rate": 2.184107481971851e-05,
+      "loss": 0.3879,
+      "step": 515
+    },
+    {
+      "epoch": 1.9348230912476723,
+      "grad_norm": 1.3815174102783203,
+      "learning_rate": 2.1676357898747836e-05,
+      "loss": 0.3744,
+      "step": 520
+    },
+    {
+      "epoch": 1.9534450651769086,
+      "grad_norm": 1.2076925039291382,
+      "learning_rate": 2.1510630813297545e-05,
+      "loss": 0.3371,
+      "step": 525
+    },
+    {
+      "epoch": 1.9720670391061452,
+      "grad_norm": 1.265386700630188,
+      "learning_rate": 2.134391863865832e-05,
+      "loss": 0.35,
+      "step": 530
+    },
+    {
+      "epoch": 1.9906890130353818,
+      "grad_norm": 1.168981671333313,
+      "learning_rate": 2.1176246599169492e-05,
+      "loss": 0.3677,
+      "step": 535
+    },
+    {
+      "epoch": 2.007448789571695,
+      "grad_norm": 1.2176551818847656,
+      "learning_rate": 2.1007640064402474e-05,
+      "loss": 0.3187,
+      "step": 540
+    },
+    {
+      "epoch": 2.026070763500931,
+      "grad_norm": 1.2926204204559326,
+      "learning_rate": 2.083812454532224e-05,
+      "loss": 0.2928,
+      "step": 545
+    },
+    {
+      "epoch": 2.0446927374301676,
+      "grad_norm": 1.0775201320648193,
+      "learning_rate": 2.0667725690427374e-05,
+      "loss": 0.2886,
+      "step": 550
+    },
+    {
+      "epoch": 2.063314711359404,
+      "grad_norm": 1.205060362815857,
+      "learning_rate": 2.0496469281869337e-05,
+      "loss": 0.2741,
+      "step": 555
+    },
+    {
+      "epoch": 2.0819366852886407,
+      "grad_norm": 1.238979458808899,
+      "learning_rate": 2.0324381231551516e-05,
+      "loss": 0.256,
+      "step": 560
+    },
+    {
+      "epoch": 2.100558659217877,
+      "grad_norm": 1.0161023139953613,
+      "learning_rate": 2.0151487577208613e-05,
+      "loss": 0.2865,
+      "step": 565
+    },
+    {
+      "epoch": 2.1191806331471135,
+      "grad_norm": 1.2487982511520386,
+      "learning_rate": 1.997781447846705e-05,
+      "loss": 0.2548,
+      "step": 570
+    },
+    {
+      "epoch": 2.1378026070763503,
+      "grad_norm": 1.2972549200057983,
+      "learning_rate": 1.9803388212886828e-05,
+      "loss": 0.2343,
+      "step": 575
+    },
+    {
+      "epoch": 2.1564245810055866,
+      "grad_norm": 1.1097617149353027,
+      "learning_rate": 1.9628235171985688e-05,
+      "loss": 0.2774,
+      "step": 580
+    },
+    {
+      "epoch": 2.175046554934823,
+      "grad_norm": 1.1447429656982422,
+      "learning_rate": 1.9452381857245876e-05,
+      "loss": 0.24,
+      "step": 585
+    },
+    {
+      "epoch": 2.1936685288640594,
+      "grad_norm": 1.177545189857483,
+      "learning_rate": 1.92758548761044e-05,
+      "loss": 0.2804,
+      "step": 590
+    },
+    {
+      "epoch": 2.212290502793296,
+      "grad_norm": 1.2133318185806274,
+      "learning_rate": 1.9098680937927164e-05,
+      "loss": 0.2409,
+      "step": 595
+    },
+    {
+      "epoch": 2.2309124767225326,
+      "grad_norm": 0.9909643530845642,
+      "learning_rate": 1.8920886849967725e-05,
+      "loss": 0.278,
+      "step": 600
+    },
+    {
+      "epoch": 2.249534450651769,
+      "grad_norm": 1.1698384284973145,
+      "learning_rate": 1.8742499513311242e-05,
+      "loss": 0.2409,
+      "step": 605
+    },
+    {
+      "epoch": 2.2681564245810057,
+      "grad_norm": 1.11681067943573,
+      "learning_rate": 1.856354591880419e-05,
+      "loss": 0.2377,
+      "step": 610
+    },
+    {
+      "epoch": 2.286778398510242,
+      "grad_norm": 1.2661349773406982,
+      "learning_rate": 1.838405314297055e-05,
+      "loss": 0.2382,
+      "step": 615
+    },
+    {
+      "epoch": 2.3054003724394785,
+      "grad_norm": 1.0915417671203613,
+      "learning_rate": 1.8204048343914962e-05,
+      "loss": 0.22,
+      "step": 620
+    },
+    {
+      "epoch": 2.3240223463687153,
+      "grad_norm": 1.3376933336257935,
+      "learning_rate": 1.8023558757213604e-05,
+      "loss": 0.269,
+      "step": 625
+    },
+    {
+      "epoch": 2.3426443202979517,
+      "grad_norm": 1.1464945077896118,
+      "learning_rate": 1.7842611691793334e-05,
+      "loss": 0.2431,
+      "step": 630
+    },
+    {
+      "epoch": 2.361266294227188,
+      "grad_norm": 1.135488748550415,
+      "learning_rate": 1.766123452579967e-05,
+      "loss": 0.2293,
+      "step": 635
+    },
+    {
+      "epoch": 2.3798882681564244,
+      "grad_norm": 1.2150588035583496,
+      "learning_rate": 1.74794547024544e-05,
+      "loss": 0.2244,
+      "step": 640
+    },
+    {
+      "epoch": 2.398510242085661,
+      "grad_norm": 1.1374517679214478,
+      "learning_rate": 1.729729972590326e-05,
+      "loss": 0.2364,
+      "step": 645
+    },
+    {
+      "epoch": 2.4171322160148976,
+      "grad_norm": 1.2967171669006348,
+      "learning_rate": 1.711479715705444e-05,
+      "loss": 0.2482,
+      "step": 650
+    },
+    {
+      "epoch": 2.435754189944134,
+      "grad_norm": 1.166526436805725,
+      "learning_rate": 1.693197460940849e-05,
+      "loss": 0.2463,
+      "step": 655
+    },
+    {
+      "epoch": 2.4543761638733708,
+      "grad_norm": 1.0751850605010986,
+      "learning_rate": 1.6748859744880255e-05,
+      "loss": 0.2579,
+      "step": 660
+    },
+    {
+      "epoch": 2.472998137802607,
+      "grad_norm": 1.1875776052474976,
+      "learning_rate": 1.6565480269613522e-05,
+      "loss": 0.2306,
+      "step": 665
+    },
+    {
+      "epoch": 2.4916201117318435,
+      "grad_norm": 1.264696478843689,
+      "learning_rate": 1.638186392978891e-05,
+      "loss": 0.1954,
+      "step": 670
+    },
+    {
+      "epoch": 2.51024208566108,
+      "grad_norm": 1.1798068284988403,
+      "learning_rate": 1.6198038507425777e-05,
+      "loss": 0.222,
+      "step": 675
+    },
+    {
+      "epoch": 2.5288640595903167,
+      "grad_norm": 1.2695372104644775,
+      "learning_rate": 1.6014031816178655e-05,
+      "loss": 0.2006,
+      "step": 680
+    },
+    {
+      "epoch": 2.547486033519553,
+      "grad_norm": 1.154645562171936,
+      "learning_rate": 1.582987169712891e-05,
+      "loss": 0.2405,
+      "step": 685
+    },
+    {
+      "epoch": 2.5661080074487894,
+      "grad_norm": 1.175594687461853,
+      "learning_rate": 1.564558601457227e-05,
+      "loss": 0.2436,
+      "step": 690
+    },
+    {
+      "epoch": 2.5847299813780262,
+      "grad_norm": 1.191159725189209,
+      "learning_rate": 1.5461202651802802e-05,
+      "loss": 0.2027,
+      "step": 695
+    },
+    {
+      "epoch": 2.6033519553072626,
+      "grad_norm": 1.2153513431549072,
+      "learning_rate": 1.5276749506894056e-05,
+      "loss": 0.2384,
+      "step": 700
+    },
+    {
+      "epoch": 2.621973929236499,
+      "grad_norm": 1.1220816373825073,
+      "learning_rate": 1.5092254488477951e-05,
+      "loss": 0.2117,
+      "step": 705
+    },
+    {
+      "epoch": 2.6405959031657353,
+      "grad_norm": 1.1654083728790283,
+      "learning_rate": 1.4907745511522056e-05,
+      "loss": 0.2474,
+      "step": 710
+    },
+    {
+      "epoch": 2.659217877094972,
+      "grad_norm": 1.138382911682129,
+      "learning_rate": 1.4723250493105948e-05,
+      "loss": 0.1977,
+      "step": 715
+    },
+    {
+      "epoch": 2.6778398510242085,
+      "grad_norm": 1.1752201318740845,
+      "learning_rate": 1.4538797348197202e-05,
+      "loss": 0.2314,
+      "step": 720
+    },
+    {
+      "epoch": 2.6964618249534453,
+      "grad_norm": 1.0158238410949707,
+      "learning_rate": 1.4354413985427733e-05,
+      "loss": 0.2155,
+      "step": 725
+    },
+    {
+      "epoch": 2.7150837988826817,
+      "grad_norm": 1.4306650161743164,
+      "learning_rate": 1.4170128302871094e-05,
+      "loss": 0.1841,
+      "step": 730
+    },
+    {
+      "epoch": 2.733705772811918,
+      "grad_norm": 1.0107722282409668,
+      "learning_rate": 1.3985968183821347e-05,
+      "loss": 0.1647,
+      "step": 735
+    },
+    {
+      "epoch": 2.7523277467411544,
+      "grad_norm": 1.1376290321350098,
+      "learning_rate": 1.3801961492574229e-05,
+      "loss": 0.2009,
+      "step": 740
+    },
+    {
+      "epoch": 2.770949720670391,
+      "grad_norm": 1.2442623376846313,
+      "learning_rate": 1.361813607021109e-05,
+      "loss": 0.2114,
+      "step": 745
+    },
+    {
+      "epoch": 2.7895716945996276,
+      "grad_norm": 1.1218327283859253,
+      "learning_rate": 1.3434519730386483e-05,
+      "loss": 0.1969,
+      "step": 750
+    },
+    {
+      "epoch": 2.808193668528864,
+      "grad_norm": 1.1548234224319458,
+      "learning_rate": 1.3251140255119744e-05,
+      "loss": 0.2084,
+      "step": 755
+    },
+    {
+      "epoch": 2.826815642458101,
+      "grad_norm": 1.0886045694351196,
+      "learning_rate": 1.306802539059151e-05,
+      "loss": 0.2111,
+      "step": 760
+    },
+    {
+      "epoch": 2.845437616387337,
+      "grad_norm": 1.0762062072753906,
+      "learning_rate": 1.2885202842945563e-05,
+      "loss": 0.2178,
+      "step": 765
+    },
+    {
+      "epoch": 2.8640595903165735,
+      "grad_norm": 1.115242838859558,
+      "learning_rate": 1.2702700274096744e-05,
+      "loss": 0.2133,
+      "step": 770
+    },
+    {
+      "epoch": 2.88268156424581,
+      "grad_norm": 1.0876168012619019,
+      "learning_rate": 1.2520545297545603e-05,
+      "loss": 0.1729,
+      "step": 775
+    },
+    {
+      "epoch": 2.9013035381750467,
+      "grad_norm": 1.0145092010498047,
+      "learning_rate": 1.233876547420033e-05,
+      "loss": 0.1832,
+      "step": 780
+    },
+    {
+      "epoch": 2.919925512104283,
+      "grad_norm": 1.134987473487854,
+      "learning_rate": 1.2157388308206669e-05,
+      "loss": 0.2141,
+      "step": 785
+    },
+    {
+      "epoch": 2.9385474860335195,
+      "grad_norm": 1.1757721900939941,
+      "learning_rate": 1.1976441242786394e-05,
+      "loss": 0.206,
+      "step": 790
+    },
+    {
+      "epoch": 2.9571694599627563,
+      "grad_norm": 1.1044893264770508,
+      "learning_rate": 1.1795951656085042e-05,
+      "loss": 0.1804,
+      "step": 795
+    },
+    {
+      "epoch": 2.9757914338919926,
+      "grad_norm": 1.0237022638320923,
+      "learning_rate": 1.1615946857029453e-05,
+      "loss": 0.1853,
+      "step": 800
+    },
+    {
+      "epoch": 2.994413407821229,
+      "grad_norm": 1.2020282745361328,
+      "learning_rate": 1.1436454081195811e-05,
+      "loss": 0.1908,
+      "step": 805
+    },
+    {
+      "epoch": 3.011173184357542,
+      "grad_norm": 0.9971558451652527,
+      "learning_rate": 1.1257500486688762e-05,
+      "loss": 0.1511,
+      "step": 810
+    },
+    {
+      "epoch": 3.0297951582867784,
+      "grad_norm": 1.146209955215454,
+      "learning_rate": 1.1079113150032282e-05,
+      "loss": 0.1326,
+      "step": 815
+    },
+    {
+      "epoch": 3.0484171322160147,
+      "grad_norm": 0.9054371118545532,
+      "learning_rate": 1.0901319062072841e-05,
+      "loss": 0.132,
+      "step": 820
+    },
+    {
+      "epoch": 3.0670391061452515,
+      "grad_norm": 1.019821286201477,
+      "learning_rate": 1.0724145123895605e-05,
+      "loss": 0.158,
+      "step": 825
+    },
+    {
+      "epoch": 3.085661080074488,
+      "grad_norm": 1.2374272346496582,
+      "learning_rate": 1.0547618142754127e-05,
+      "loss": 0.1405,
+      "step": 830
+    },
+    {
+      "epoch": 3.1042830540037243,
+      "grad_norm": 1.1176825761795044,
+      "learning_rate": 1.0371764828014318e-05,
+      "loss": 0.1397,
+      "step": 835
+    },
+    {
+      "epoch": 3.122905027932961,
+      "grad_norm": 0.9421104788780212,
+      "learning_rate": 1.0196611787113173e-05,
+      "loss": 0.129,
+      "step": 840
+    },
+    {
+      "epoch": 3.1415270018621975,
+      "grad_norm": 1.510794997215271,
+      "learning_rate": 1.0022185521532952e-05,
+      "loss": 0.1438,
+      "step": 845
+    },
+    {
+      "epoch": 3.160148975791434,
+      "grad_norm": 1.0882729291915894,
+      "learning_rate": 9.848512422791386e-06,
+      "loss": 0.1364,
+      "step": 850
+    },
+    {
+      "epoch": 3.17877094972067,
+      "grad_norm": 1.111037015914917,
+      "learning_rate": 9.67561876844849e-06,
+      "loss": 0.1425,
+      "step": 855
+    },
+    {
+      "epoch": 3.197392923649907,
+      "grad_norm": 1.1784346103668213,
+      "learning_rate": 9.503530718130669e-06,
+      "loss": 0.1347,
+      "step": 860
+    },
+    {
+      "epoch": 3.2160148975791434,
+      "grad_norm": 1.0116277933120728,
+      "learning_rate": 9.332274309572628e-06,
+      "loss": 0.1368,
+      "step": 865
+    },
+    {
+      "epoch": 3.2346368715083798,
+      "grad_norm": 1.2434337139129639,
+      "learning_rate": 9.161875454677761e-06,
+      "loss": 0.1318,
+      "step": 870
+    },
+    {
+      "epoch": 3.2532588454376166,
+      "grad_norm": 1.1153370141983032,
+      "learning_rate": 8.992359935597526e-06,
+      "loss": 0.1269,
+      "step": 875
+    },
+    {
+      "epoch": 3.271880819366853,
+      "grad_norm": 0.9924963712692261,
+      "learning_rate": 8.823753400830514e-06,
+      "loss": 0.143,
+      "step": 880
+    },
+    {
+      "epoch": 3.2905027932960893,
+      "grad_norm": 0.8898366689682007,
+      "learning_rate": 8.656081361341687e-06,
+      "loss": 0.1156,
+      "step": 885
+    },
+    {
+      "epoch": 3.3091247672253257,
+      "grad_norm": 0.9309813380241394,
+      "learning_rate": 8.48936918670246e-06,
+      "loss": 0.1216,
+      "step": 890
+    },
+    {
+      "epoch": 3.3277467411545625,
+      "grad_norm": 1.0335708856582642,
+      "learning_rate": 8.323642101252165e-06,
+      "loss": 0.1386,
+      "step": 895
+    },
+    {
+      "epoch": 3.346368715083799,
+      "grad_norm": 0.956021249294281,
+      "learning_rate": 8.158925180281494e-06,
+      "loss": 0.1419,
+      "step": 900
+    },
+    {
+      "epoch": 3.364990689013035,
+      "grad_norm": 1.0639731884002686,
+      "learning_rate": 7.995243346238472e-06,
+      "loss": 0.147,
+      "step": 905
+    },
+    {
+      "epoch": 3.383612662942272,
+      "grad_norm": 1.1223934888839722,
+      "learning_rate": 7.83262136495759e-06,
+      "loss": 0.1319,
+      "step": 910
+    },
+    {
+      "epoch": 3.4022346368715084,
+      "grad_norm": 0.9859718084335327,
+      "learning_rate": 7.671083841912603e-06,
+      "loss": 0.1189,
+      "step": 915
+    },
+    {
+      "epoch": 3.4208566108007448,
+      "grad_norm": 0.8468735218048096,
+      "learning_rate": 7.510655218493627e-06,
+      "loss": 0.123,
+      "step": 920
+    },
+    {
+      "epoch": 3.439478584729981,
+      "grad_norm": 0.8232627511024475,
+      "learning_rate": 7.351359768309017e-06,
+      "loss": 0.1216,
+      "step": 925
+    },
+    {
+      "epoch": 3.458100558659218,
+      "grad_norm": 0.9500121474266052,
+      "learning_rate": 7.193221593512663e-06,
+      "loss": 0.1255,
+      "step": 930
+    },
+    {
+      "epoch": 3.4767225325884543,
+      "grad_norm": 1.1072030067443848,
+      "learning_rate": 7.036264621157241e-06,
+      "loss": 0.1173,
+      "step": 935
+    },
+    {
+      "epoch": 3.4953445065176907,
+      "grad_norm": 1.1253328323364258,
+      "learning_rate": 6.880512599573903e-06,
+      "loss": 0.1261,
+      "step": 940
+    },
+    {
+      "epoch": 3.5139664804469275,
+      "grad_norm": 1.1017041206359863,
+      "learning_rate": 6.725989094779063e-06,
+      "loss": 0.1202,
+      "step": 945
+    },
+    {
+      "epoch": 3.532588454376164,
+      "grad_norm": 1.0356934070587158,
+      "learning_rate": 6.572717486908738e-06,
+      "loss": 0.1152,
+      "step": 950
+    },
+    {
+      "epoch": 3.5512104283054002,
+      "grad_norm": 0.9527298808097839,
+      "learning_rate": 6.420720966681052e-06,
+      "loss": 0.1141,
+      "step": 955
+    },
+    {
+      "epoch": 3.5698324022346366,
+      "grad_norm": 0.8534860610961914,
+      "learning_rate": 6.270022531887305e-06,
+      "loss": 0.1157,
+      "step": 960
+    },
+    {
+      "epoch": 3.5884543761638734,
+      "grad_norm": 1.1221200227737427,
+      "learning_rate": 6.120644983912383e-06,
+      "loss": 0.109,
+      "step": 965
+    },
+    {
+      "epoch": 3.60707635009311,
+      "grad_norm": 1.075600028038025,
+      "learning_rate": 5.972610924284732e-06,
+      "loss": 0.1287,
+      "step": 970
+    },
+    {
+      "epoch": 3.6256983240223466,
+      "grad_norm": 1.0059423446655273,
+      "learning_rate": 5.825942751256677e-06,
+      "loss": 0.1256,
+      "step": 975
+    },
+    {
+      "epoch": 3.644320297951583,
+      "grad_norm": 1.0131361484527588,
+      "learning_rate": 5.680662656415441e-06,
+      "loss": 0.127,
+      "step": 980
+    },
+    {
+      "epoch": 3.6629422718808193,
+      "grad_norm": 0.9724953174591064,
+      "learning_rate": 5.536792621325479e-06,
+      "loss": 0.1285,
+      "step": 985
+    },
+    {
+      "epoch": 3.6815642458100557,
+      "grad_norm": 1.1040209531784058,
+      "learning_rate": 5.394354414202529e-06,
+      "loss": 0.1177,
+      "step": 990
+    },
+    {
+      "epoch": 3.7001862197392925,
+      "grad_norm": 0.9747113585472107,
+      "learning_rate": 5.253369586620004e-06,
+      "loss": 0.1071,
+      "step": 995
+    },
+    {
+      "epoch": 3.718808193668529,
+      "grad_norm": 0.9118765592575073,
+      "learning_rate": 5.113859470248123e-06,
+      "loss": 0.1176,
+      "step": 1000
+    },
+    {
+      "epoch": 3.7374301675977653,
+      "grad_norm": 0.9922019839286804,
+      "learning_rate": 4.975845173626351e-06,
+      "loss": 0.1011,
+      "step": 1005
+    },
+    {
+      "epoch": 3.756052141527002,
+      "grad_norm": 0.9839317798614502,
+      "learning_rate": 4.8393475789695325e-06,
+      "loss": 0.1086,
+      "step": 1010
+    },
+    {
+      "epoch": 3.7746741154562384,
+      "grad_norm": 1.049396276473999,
+      "learning_rate": 4.7043873390083654e-06,
+      "loss": 0.1114,
+      "step": 1015
+    },
+    {
+      "epoch": 3.793296089385475,
+      "grad_norm": 1.1213620901107788,
+      "learning_rate": 4.570984873864519e-06,
+      "loss": 0.0996,
+      "step": 1020
+    },
+    {
+      "epoch": 3.811918063314711,
+      "grad_norm": 0.9337913393974304,
+      "learning_rate": 4.439160367960966e-06,
+      "loss": 0.0884,
+      "step": 1025
+    },
+    {
+      "epoch": 3.830540037243948,
+      "grad_norm": 1.1189533472061157,
+      "learning_rate": 4.308933766968003e-06,
+      "loss": 0.1128,
+      "step": 1030
+    },
+    {
+      "epoch": 3.8491620111731844,
+      "grad_norm": 0.8592391014099121,
+      "learning_rate": 4.180324774785356e-06,
+      "loss": 0.1036,
+      "step": 1035
+    },
+    {
+      "epoch": 3.8677839851024207,
+      "grad_norm": 0.9491106271743774,
+      "learning_rate": 4.053352850560923e-06,
+      "loss": 0.1053,
+      "step": 1040
+    },
+    {
+      "epoch": 3.8864059590316575,
+      "grad_norm": 1.7115939855575562,
+      "learning_rate": 3.928037205746451e-06,
+      "loss": 0.1079,
+      "step": 1045
+    },
+    {
+      "epoch": 3.905027932960894,
+      "grad_norm": 0.965495228767395,
+      "learning_rate": 3.8043968011908277e-06,
+      "loss": 0.1084,
+      "step": 1050
+    },
+    {
+      "epoch": 3.9236499068901303,
+      "grad_norm": 0.9409112334251404,
+      "learning_rate": 3.682450344271157e-06,
+      "loss": 0.0952,
+      "step": 1055
+    },
+    {
+      "epoch": 3.9422718808193666,
+      "grad_norm": 0.8451502323150635,
+      "learning_rate": 3.562216286062271e-06,
+      "loss": 0.1157,
+      "step": 1060
+    },
+    {
+      "epoch": 3.9608938547486034,
+      "grad_norm": 1.0194039344787598,
+      "learning_rate": 3.4437128185449872e-06,
+      "loss": 0.0916,
+      "step": 1065
+    },
+    {
+      "epoch": 3.97951582867784,
+      "grad_norm": 0.8282318115234375,
+      "learning_rate": 3.326957871853587e-06,
+      "loss": 0.0934,
+      "step": 1070
+    },
+    {
+      "epoch": 3.998137802607076,
+      "grad_norm": 0.8793200850486755,
+      "learning_rate": 3.211969111562886e-06,
+      "loss": 0.1086,
+      "step": 1075
+    },
+    {
+      "epoch": 4.01489757914339,
+      "grad_norm": 0.8343816995620728,
+      "learning_rate": 3.0987639360153587e-06,
+      "loss": 0.092,
+      "step": 1080
+    },
+    {
+      "epoch": 4.033519553072626,
+      "grad_norm": 0.7905257344245911,
+      "learning_rate": 2.9873594736886766e-06,
+      "loss": 0.0832,
+      "step": 1085
+    },
+    {
+      "epoch": 4.052141527001862,
+      "grad_norm": 0.8123075366020203,
+      "learning_rate": 2.8777725806041194e-06,
+      "loss": 0.0813,
+      "step": 1090
+    },
+    {
+      "epoch": 4.070763500931099,
+      "grad_norm": 0.6700873374938965,
+      "learning_rate": 2.7700198377761266e-06,
+      "loss": 0.0914,
+      "step": 1095
+    },
+    {
+      "epoch": 4.089385474860335,
+      "grad_norm": 0.7350034713745117,
+      "learning_rate": 2.664117548703564e-06,
+      "loss": 0.0802,
+      "step": 1100
+    },
+    {
+      "epoch": 4.1080074487895715,
+      "grad_norm": 0.7391467690467834,
+      "learning_rate": 2.560081736902907e-06,
+      "loss": 0.0765,
+      "step": 1105
+    },
+    {
+      "epoch": 4.126629422718808,
+      "grad_norm": 0.8788180351257324,
+      "learning_rate": 2.4579281434837836e-06,
+      "loss": 0.0876,
+      "step": 1110
+    },
+    {
+      "epoch": 4.145251396648045,
+      "grad_norm": 0.8143097162246704,
+      "learning_rate": 2.3576722247673127e-06,
+      "loss": 0.0785,
+      "step": 1115
+    },
+    {
+      "epoch": 4.1638733705772815,
+      "grad_norm": 0.8672139644622803,
+      "learning_rate": 2.259329149947457e-06,
+      "loss": 0.0915,
+      "step": 1120
+    },
+    {
+      "epoch": 4.182495344506518,
+      "grad_norm": 0.8274207711219788,
+      "learning_rate": 2.1629137987958768e-06,
+      "loss": 0.0827,
+      "step": 1125
+    },
+    {
+      "epoch": 4.201117318435754,
+      "grad_norm": 0.8411132097244263,
+      "learning_rate": 2.0684407594105163e-06,
+      "loss": 0.0914,
+      "step": 1130
+    },
+    {
+      "epoch": 4.219739292364991,
+      "grad_norm": 0.7280264496803284,
+      "learning_rate": 1.9759243260084025e-06,
+      "loss": 0.0799,
+      "step": 1135
+    },
+    {
+      "epoch": 4.238361266294227,
+      "grad_norm": 0.7131537199020386,
+      "learning_rate": 1.8853784967628158e-06,
+      "loss": 0.0901,
+      "step": 1140
+    },
+    {
+      "epoch": 4.256983240223463,
+      "grad_norm": 0.7776678204536438,
+      "learning_rate": 1.7968169716853328e-06,
+      "loss": 0.0845,
+      "step": 1145
+    },
+    {
+      "epoch": 4.275605214152701,
+      "grad_norm": 0.788881778717041,
+      "learning_rate": 1.7102531505529317e-06,
+      "loss": 0.1044,
+      "step": 1150
+    },
+    {
+      "epoch": 4.294227188081937,
+      "grad_norm": 0.8455144762992859,
+      "learning_rate": 1.6257001308805602e-06,
+      "loss": 0.0733,
+      "step": 1155
+    },
+    {
+      "epoch": 4.312849162011173,
+      "grad_norm": 0.8313959240913391,
+      "learning_rate": 1.5431707059393973e-06,
+      "loss": 0.0823,
+      "step": 1160
+    },
+    {
+      "epoch": 4.33147113594041,
+      "grad_norm": 0.7532757520675659,
+      "learning_rate": 1.462677362821187e-06,
+      "loss": 0.0829,
+      "step": 1165
+    },
+    {
+      "epoch": 4.350093109869646,
+      "grad_norm": 0.7443646192550659,
+      "learning_rate": 1.384232280548876e-06,
+      "loss": 0.0808,
+      "step": 1170
+    },
+    {
+      "epoch": 4.368715083798882,
+      "grad_norm": 0.7649634480476379,
+      "learning_rate": 1.3078473282338848e-06,
+      "loss": 0.0717,
+      "step": 1175
+    },
+    {
+      "epoch": 4.387337057728119,
+      "grad_norm": 0.7778022885322571,
+      "learning_rate": 1.2335340632802173e-06,
+      "loss": 0.091,
+      "step": 1180
+    },
+    {
+      "epoch": 4.405959031657356,
+      "grad_norm": 0.7728734612464905,
+      "learning_rate": 1.1613037296358175e-06,
+      "loss": 0.0855,
+      "step": 1185
+    },
+    {
+      "epoch": 4.424581005586592,
+      "grad_norm": 0.7408890724182129,
+      "learning_rate": 1.0911672560912816e-06,
+      "loss": 0.0822,
+      "step": 1190
+    },
+    {
+      "epoch": 4.443202979515829,
+      "grad_norm": 0.8574674725532532,
+      "learning_rate": 1.0231352546262728e-06,
+      "loss": 0.0831,
+      "step": 1195
+    },
+    {
+      "epoch": 4.461824953445065,
+      "grad_norm": 0.8668495416641235,
+      "learning_rate": 9.572180188039065e-07,
+      "loss": 0.0777,
+      "step": 1200
+    },
+    {
+      "epoch": 4.4804469273743015,
+      "grad_norm": 0.7216169834136963,
+      "learning_rate": 8.93425522213257e-07,
+      "loss": 0.0847,
+      "step": 1205
+    },
+    {
+      "epoch": 4.499068901303538,
+      "grad_norm": 0.75110924243927,
+      "learning_rate": 8.317674169603223e-07,
+      "loss": 0.08,
+      "step": 1210
+    },
+    {
+      "epoch": 4.517690875232775,
+      "grad_norm": 0.7196640372276306,
+      "learning_rate": 7.722530322076099e-07,
+      "loss": 0.0945,
+      "step": 1215
+    },
+    {
+      "epoch": 4.5363128491620115,
+      "grad_norm": 0.7442631125450134,
+      "learning_rate": 7.148913727625994e-07,
+      "loss": 0.0804,
+      "step": 1220
+    },
+    {
+      "epoch": 4.554934823091248,
+      "grad_norm": 0.826667308807373,
+      "learning_rate": 6.596911177152559e-07,
+      "loss": 0.0899,
+      "step": 1225
+    },
+    {
+      "epoch": 4.573556797020484,
+      "grad_norm": 0.675247848033905,
+      "learning_rate": 6.066606191248536e-07,
+      "loss": 0.0788,
+      "step": 1230
+    },
+    {
+      "epoch": 4.592178770949721,
+      "grad_norm": 0.5857364535331726,
+      "learning_rate": 5.558079007562683e-07,
+      "loss": 0.0738,
+      "step": 1235
+    },
+    {
+      "epoch": 4.610800744878957,
+      "grad_norm": 0.7541478276252747,
+      "learning_rate": 5.071406568659486e-07,
+      "loss": 0.0787,
+      "step": 1240
+    },
+    {
+      "epoch": 4.629422718808193,
+      "grad_norm": 0.8422206044197083,
+      "learning_rate": 4.606662510377274e-07,
+      "loss": 0.0785,
+      "step": 1245
+    },
+    {
+      "epoch": 4.648044692737431,
+      "grad_norm": 0.7363956570625305,
+      "learning_rate": 4.163917150686858e-07,
+      "loss": 0.0812,
+      "step": 1250
+    },
+    {
+      "epoch": 4.666666666666667,
+      "grad_norm": 0.7241252660751343,
+      "learning_rate": 3.743237479052014e-07,
+      "loss": 0.0797,
+      "step": 1255
+    },
+    {
+      "epoch": 4.685288640595903,
+      "grad_norm": 0.7271686792373657,
+      "learning_rate": 3.344687146293807e-07,
+      "loss": 0.0679,
+      "step": 1260
+    },
+    {
+      "epoch": 4.70391061452514,
+      "grad_norm": 0.8053618669509888,
+      "learning_rate": 2.968326454959708e-07,
+      "loss": 0.0778,
+      "step": 1265
+    },
+    {
+      "epoch": 4.722532588454376,
+      "grad_norm": 0.7725048065185547,
+      "learning_rate": 2.614212350199746e-07,
+      "loss": 0.0738,
+      "step": 1270
+    },
+    {
+      "epoch": 4.741154562383612,
+      "grad_norm": 0.6539868712425232,
+      "learning_rate": 2.2823984111502828e-07,
+      "loss": 0.0764,
+      "step": 1275
+    },
+    {
+      "epoch": 4.759776536312849,
+      "grad_norm": 0.6823570132255554,
+      "learning_rate": 1.9729348428272954e-07,
+      "loss": 0.0878,
+      "step": 1280
+    },
+    {
+      "epoch": 4.778398510242086,
+      "grad_norm": 0.6665881872177124,
+      "learning_rate": 1.6858684685301718e-07,
+      "loss": 0.0804,
+      "step": 1285
+    },
+    {
+      "epoch": 4.797020484171322,
+      "grad_norm": 0.6775912642478943,
+      "learning_rate": 1.4212427227569934e-07,
+      "loss": 0.0874,
+      "step": 1290
+    },
+    {
+      "epoch": 4.815642458100559,
+      "grad_norm": 0.6975992918014526,
+      "learning_rate": 1.1790976446328094e-07,
+      "loss": 0.0716,
+      "step": 1295
+    },
+    {
+      "epoch": 4.834264432029795,
+      "grad_norm": 0.7351179122924805,
+      "learning_rate": 9.594698718515149e-08,
+      "loss": 0.066,
+      "step": 1300
+    },
+    {
+      "epoch": 4.8528864059590315,
+      "grad_norm": 0.8980118632316589,
+      "learning_rate": 7.623926351323197e-08,
+      "loss": 0.0852,
+      "step": 1305
+    },
+    {
+      "epoch": 4.871508379888268,
+      "grad_norm": 0.6633474826812744,
+      "learning_rate": 5.8789575319185274e-08,
+      "loss": 0.0735,
+      "step": 1310
+    },
+    {
+      "epoch": 4.890130353817504,
+      "grad_norm": 0.7721505761146545,
+      "learning_rate": 4.360056282324221e-08,
+      "loss": 0.0758,
+      "step": 1315
+    },
+    {
+      "epoch": 4.9087523277467415,
+      "grad_norm": 0.8143950700759888,
+      "learning_rate": 3.067452419472594e-08,
+      "loss": 0.0831,
+      "step": 1320
+    },
+    {
+      "epoch": 4.927374301675978,
+      "grad_norm": 0.7624742984771729,
+      "learning_rate": 2.0013415204321872e-08,
+      "loss": 0.0685,
+      "step": 1325
+    },
+    {
+      "epoch": 4.945996275605214,
+      "grad_norm": 0.8198992013931274,
+      "learning_rate": 1.1618848928172664e-08,
+      "loss": 0.0787,
+      "step": 1330
+    },
+    {
+      "epoch": 4.964618249534451,
+      "grad_norm": 0.8066611289978027,
+      "learning_rate": 5.492095503795147e-09,
+      "loss": 0.08,
+      "step": 1335
+    },
+    {
+      "epoch": 4.983240223463687,
+      "grad_norm": 0.7737635970115662,
+      "learning_rate": 1.6340819379190296e-09,
+      "loss": 0.0748,
+      "step": 1340
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.9816650152206421,
+      "learning_rate": 4.539196621577979e-11,
+      "loss": 0.0686,
+      "step": 1345
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1345,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 2000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.9262091408588145e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

53_128_e5_3e-5/checkpoint-1345/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ef384a0e7f290964c504f65dcbd3ddbb9bba343263178891e1dfd052c620a7a4
+size 7736

53_128_e5_3e-5/checkpoint-1345/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)

53_128_e5_3e-5/checkpoint-269/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

53_128_e5_3e-5/checkpoint-269/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": [
+    "q_proj",
+    "k_proj",
+    "down_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "v_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

53_128_e5_3e-5/checkpoint-269/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bff29f9654af2beaa4a681fa6f326802aa26ef1e12a7e924637cda47e3619485
+size 791751704

53_128_e5_3e-5/checkpoint-269/latest

@@ -0,0 +1 @@
+global_step269

53_128_e5_3e-5/checkpoint-269/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2b32ca30a6e6194d57d133dc195f188e7f34878a5833934071725534e10410bc
+size 15920

53_128_e5_3e-5/checkpoint-269/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b68282405de5bad04d73a3f1c2aa19db7fdd36b2ba8c2d5e0588cc15a97d03ce
+size 15920

53_128_e5_3e-5/checkpoint-269/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:62fd5ce88f54d558e06fd1ea31643607d7eecb12801ea5bd67246c38e775cb78
+size 15920