Upload folder using huggingface_hub

51_128_e5_3e-5/checkpoint-1182/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

51_128_e5_3e-5/checkpoint-1182/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": [
+    "down_proj",
+    "q_proj",
+    "up_proj",
+    "gate_proj",
+    "k_proj",
+    "o_proj",
+    "v_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

51_128_e5_3e-5/checkpoint-1182/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:454f03013903741741b6ea15aa6f1552b1aa207d6e04b0e8e1c7e381eb604d37
+size 791751704

51_128_e5_3e-5/checkpoint-1182/latest

@@ -0,0 +1 @@
+global_step1182

51_128_e5_3e-5/checkpoint-1182/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4a33804ff74024a37689e8e34691351e4c1c09e9325d762e125d86ec682752f2
+size 15920

51_128_e5_3e-5/checkpoint-1182/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1c63cb0be7b72e9bffdfb2306d419b40e90e2bf8555d7b92b575e6758690bb6b
+size 15920

51_128_e5_3e-5/checkpoint-1182/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0ab081d4d2f6ad29eaa71d5921f141812405eabddf06ad453b748e0e4f13dbd4
+size 15920

51_128_e5_3e-5/checkpoint-1182/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:95cd3bd3e873242038771f847610c0b97e7e9023462083186b1c3330257436d7
+size 15920

51_128_e5_3e-5/checkpoint-1182/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:06d22556bff59f8e8f3bdedd28ddfcce8acf01c83062a690cc05e8059d70485d
+size 15920

51_128_e5_3e-5/checkpoint-1182/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eb6ed2af573625633793ffda4ffad55603544a35e36b27f56ec0929917b8b87f
+size 15920

51_128_e5_3e-5/checkpoint-1182/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a2a3a4991ebcf4e9013d84194ee51c8d361dd4d01545d9c0c87b236a745c9605
+size 15920

51_128_e5_3e-5/checkpoint-1182/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a899d7c23b8221e0684c6e1f87148839cb1b556bbc308e63f9bc7e6de0299028
+size 15920

51_128_e5_3e-5/checkpoint-1182/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:421e144c0cdf386822bf7ef9f527ced3bcfe855d2d8605fac888c3630e01ba3b
+size 1064

51_128_e5_3e-5/checkpoint-1182/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
+  }
+}

51_128_e5_3e-5/checkpoint-1182/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
+}

51_128_e5_3e-5/checkpoint-1182/trainer_state.json

@@ -0,0 +1,1686 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 1182,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.012706480304955527,
+      "grad_norm": 1.0249743461608887,
+      "learning_rate": 1.2121212121212122e-06,
+      "loss": 1.2769,
+      "step": 5
+    },
+    {
+      "epoch": 0.025412960609911054,
+      "grad_norm": 0.894174337387085,
+      "learning_rate": 2.7272727272727272e-06,
+      "loss": 1.3117,
+      "step": 10
+    },
+    {
+      "epoch": 0.03811944091486658,
+      "grad_norm": 0.6720679998397827,
+      "learning_rate": 4.242424242424242e-06,
+      "loss": 1.33,
+      "step": 15
+    },
+    {
+      "epoch": 0.05082592121982211,
+      "grad_norm": 0.7012405395507812,
+      "learning_rate": 5.757575757575758e-06,
+      "loss": 1.3123,
+      "step": 20
+    },
+    {
+      "epoch": 0.06353240152477764,
+      "grad_norm": 0.8811719417572021,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 1.2455,
+      "step": 25
+    },
+    {
+      "epoch": 0.07623888182973317,
+      "grad_norm": 0.5775105357170105,
+      "learning_rate": 8.787878787878788e-06,
+      "loss": 1.2503,
+      "step": 30
+    },
+    {
+      "epoch": 0.08894536213468869,
+      "grad_norm": 0.6601129770278931,
+      "learning_rate": 1.0303030303030302e-05,
+      "loss": 1.2622,
+      "step": 35
+    },
+    {
+      "epoch": 0.10165184243964422,
+      "grad_norm": 0.563920795917511,
+      "learning_rate": 1.1818181818181819e-05,
+      "loss": 1.2293,
+      "step": 40
+    },
+    {
+      "epoch": 0.11435832274459974,
+      "grad_norm": 0.6447675824165344,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.1266,
+      "step": 45
+    },
+    {
+      "epoch": 0.12706480304955528,
+      "grad_norm": 0.6008123755455017,
+      "learning_rate": 1.484848484848485e-05,
+      "loss": 1.1518,
+      "step": 50
+    },
+    {
+      "epoch": 0.1397712833545108,
+      "grad_norm": 0.5885509848594666,
+      "learning_rate": 1.6363636363636363e-05,
+      "loss": 1.2858,
+      "step": 55
+    },
+    {
+      "epoch": 0.15247776365946633,
+      "grad_norm": 0.47118306159973145,
+      "learning_rate": 1.7878787878787877e-05,
+      "loss": 1.1583,
+      "step": 60
+    },
+    {
+      "epoch": 0.16518424396442186,
+      "grad_norm": 0.5283555388450623,
+      "learning_rate": 1.9393939393939395e-05,
+      "loss": 1.2376,
+      "step": 65
+    },
+    {
+      "epoch": 0.17789072426937738,
+      "grad_norm": 0.5695042014122009,
+      "learning_rate": 2.090909090909091e-05,
+      "loss": 1.2086,
+      "step": 70
+    },
+    {
+      "epoch": 0.1905972045743329,
+      "grad_norm": 0.7871887683868408,
+      "learning_rate": 2.2424242424242424e-05,
+      "loss": 1.178,
+      "step": 75
+    },
+    {
+      "epoch": 0.20330368487928843,
+      "grad_norm": 0.5051509141921997,
+      "learning_rate": 2.3939393939393942e-05,
+      "loss": 1.1244,
+      "step": 80
+    },
+    {
+      "epoch": 0.21601016518424396,
+      "grad_norm": 0.6592188477516174,
+      "learning_rate": 2.5454545454545457e-05,
+      "loss": 1.1719,
+      "step": 85
+    },
+    {
+      "epoch": 0.22871664548919948,
+      "grad_norm": 0.5462698340415955,
+      "learning_rate": 2.696969696969697e-05,
+      "loss": 1.1111,
+      "step": 90
+    },
+    {
+      "epoch": 0.241423125794155,
+      "grad_norm": 0.5268920063972473,
+      "learning_rate": 2.8484848484848486e-05,
+      "loss": 1.1048,
+      "step": 95
+    },
+    {
+      "epoch": 0.25412960609911056,
+      "grad_norm": 0.5009660124778748,
+      "learning_rate": 3e-05,
+      "loss": 1.1586,
+      "step": 100
+    },
+    {
+      "epoch": 0.2668360864040661,
+      "grad_norm": 0.6100103259086609,
+      "learning_rate": 2.999947137111298e-05,
+      "loss": 1.1005,
+      "step": 105
+    },
+    {
+      "epoch": 0.2795425667090216,
+      "grad_norm": 0.5694701075553894,
+      "learning_rate": 2.9997885521711738e-05,
+      "loss": 1.0613,
+      "step": 110
+    },
+    {
+      "epoch": 0.29224904701397714,
+      "grad_norm": 0.5084468722343445,
+      "learning_rate": 2.9995242563573035e-05,
+      "loss": 1.0681,
+      "step": 115
+    },
+    {
+      "epoch": 0.30495552731893266,
+      "grad_norm": 0.5901311635971069,
+      "learning_rate": 2.9991542682982747e-05,
+      "loss": 1.0505,
+      "step": 120
+    },
+    {
+      "epoch": 0.3176620076238882,
+      "grad_norm": 0.5251779556274414,
+      "learning_rate": 2.99867861407227e-05,
+      "loss": 1.0543,
+      "step": 125
+    },
+    {
+      "epoch": 0.3303684879288437,
+      "grad_norm": 0.5326008796691895,
+      "learning_rate": 2.9980973272052328e-05,
+      "loss": 1.0744,
+      "step": 130
+    },
+    {
+      "epoch": 0.34307496823379924,
+      "grad_norm": 0.5696294903755188,
+      "learning_rate": 2.9974104486684988e-05,
+      "loss": 1.0599,
+      "step": 135
+    },
+    {
+      "epoch": 0.35578144853875476,
+      "grad_norm": 0.5977292656898499,
+      "learning_rate": 2.996618026875914e-05,
+      "loss": 1.071,
+      "step": 140
+    },
+    {
+      "epoch": 0.3684879288437103,
+      "grad_norm": 0.5755591988563538,
+      "learning_rate": 2.995720117680417e-05,
+      "loss": 1.0214,
+      "step": 145
+    },
+    {
+      "epoch": 0.3811944091486658,
+      "grad_norm": 0.8020675778388977,
+      "learning_rate": 2.994716784370108e-05,
+      "loss": 1.0587,
+      "step": 150
+    },
+    {
+      "epoch": 0.39390088945362134,
+      "grad_norm": 0.6691463589668274,
+      "learning_rate": 2.9936080976637823e-05,
+      "loss": 0.9774,
+      "step": 155
+    },
+    {
+      "epoch": 0.40660736975857686,
+      "grad_norm": 0.6647286415100098,
+      "learning_rate": 2.992394135705949e-05,
+      "loss": 1.0251,
+      "step": 160
+    },
+    {
+      "epoch": 0.4193138500635324,
+      "grad_norm": 0.7553662061691284,
+      "learning_rate": 2.9910749840613233e-05,
+      "loss": 0.9626,
+      "step": 165
+    },
+    {
+      "epoch": 0.4320203303684879,
+      "grad_norm": 0.6578956842422485,
+      "learning_rate": 2.9896507357087928e-05,
+      "loss": 0.9649,
+      "step": 170
+    },
+    {
+      "epoch": 0.44472681067344344,
+      "grad_norm": 0.6698135137557983,
+      "learning_rate": 2.988121491034868e-05,
+      "loss": 0.9372,
+      "step": 175
+    },
+    {
+      "epoch": 0.45743329097839897,
+      "grad_norm": 0.7247338891029358,
+      "learning_rate": 2.9864873578266034e-05,
+      "loss": 0.9995,
+      "step": 180
+    },
+    {
+      "epoch": 0.4701397712833545,
+      "grad_norm": 0.7628229856491089,
+      "learning_rate": 2.9847484512640018e-05,
+      "loss": 0.9421,
+      "step": 185
+    },
+    {
+      "epoch": 0.48284625158831,
+      "grad_norm": 0.7909829020500183,
+      "learning_rate": 2.9829048939118944e-05,
+      "loss": 0.9142,
+      "step": 190
+    },
+    {
+      "epoch": 0.49555273189326554,
+      "grad_norm": 0.7012385725975037,
+      "learning_rate": 2.9809568157113047e-05,
+      "loss": 0.9399,
+      "step": 195
+    },
+    {
+      "epoch": 0.5082592121982211,
+      "grad_norm": 0.7712706327438354,
+      "learning_rate": 2.9789043539702875e-05,
+      "loss": 0.9837,
+      "step": 200
+    },
+    {
+      "epoch": 0.5209656925031766,
+      "grad_norm": 0.7406087517738342,
+      "learning_rate": 2.9767476533542513e-05,
+      "loss": 0.9507,
+      "step": 205
+    },
+    {
+      "epoch": 0.5336721728081322,
+      "grad_norm": 0.7858076095581055,
+      "learning_rate": 2.9744868658757628e-05,
+      "loss": 0.9529,
+      "step": 210
+    },
+    {
+      "epoch": 0.5463786531130876,
+      "grad_norm": 0.7638741731643677,
+      "learning_rate": 2.9721221508838302e-05,
+      "loss": 0.9502,
+      "step": 215
+    },
+    {
+      "epoch": 0.5590851334180432,
+      "grad_norm": 0.9117738604545593,
+      "learning_rate": 2.9696536750526748e-05,
+      "loss": 0.8913,
+      "step": 220
+    },
+    {
+      "epoch": 0.5717916137229987,
+      "grad_norm": 0.8238304853439331,
+      "learning_rate": 2.9670816123699812e-05,
+      "loss": 0.8704,
+      "step": 225
+    },
+    {
+      "epoch": 0.5844980940279543,
+      "grad_norm": 0.7925108075141907,
+      "learning_rate": 2.9644061441246323e-05,
+      "loss": 0.8565,
+      "step": 230
+    },
+    {
+      "epoch": 0.5972045743329097,
+      "grad_norm": 0.7725275754928589,
+      "learning_rate": 2.9616274588939364e-05,
+      "loss": 0.8292,
+      "step": 235
+    },
+    {
+      "epoch": 0.6099110546378653,
+      "grad_norm": 0.8474175333976746,
+      "learning_rate": 2.9587457525303305e-05,
+      "loss": 0.8822,
+      "step": 240
+    },
+    {
+      "epoch": 0.6226175349428208,
+      "grad_norm": 1.0394459962844849,
+      "learning_rate": 2.9557612281475776e-05,
+      "loss": 0.8602,
+      "step": 245
+    },
+    {
+      "epoch": 0.6353240152477764,
+      "grad_norm": 0.9201862812042236,
+      "learning_rate": 2.9526740961064516e-05,
+      "loss": 0.8374,
+      "step": 250
+    },
+    {
+      "epoch": 0.6480304955527318,
+      "grad_norm": 0.9244139790534973,
+      "learning_rate": 2.9494845739999103e-05,
+      "loss": 0.8361,
+      "step": 255
+    },
+    {
+      "epoch": 0.6607369758576874,
+      "grad_norm": 0.9039394855499268,
+      "learning_rate": 2.9461928866377553e-05,
+      "loss": 0.8261,
+      "step": 260
+    },
+    {
+      "epoch": 0.6734434561626429,
+      "grad_norm": 0.7772687673568726,
+      "learning_rate": 2.942799266030791e-05,
+      "loss": 0.8687,
+      "step": 265
+    },
+    {
+      "epoch": 0.6861499364675985,
+      "grad_norm": 1.1102648973464966,
+      "learning_rate": 2.9393039513744684e-05,
+      "loss": 0.8003,
+      "step": 270
+    },
+    {
+      "epoch": 0.6988564167725541,
+      "grad_norm": 0.8548868298530579,
+      "learning_rate": 2.9357071890320262e-05,
+      "loss": 0.8169,
+      "step": 275
+    },
+    {
+      "epoch": 0.7115628970775095,
+      "grad_norm": 0.8160343766212463,
+      "learning_rate": 2.9320092325171292e-05,
+      "loss": 0.7822,
+      "step": 280
+    },
+    {
+      "epoch": 0.7242693773824651,
+      "grad_norm": 0.825340986251831,
+      "learning_rate": 2.9282103424759935e-05,
+      "loss": 0.7904,
+      "step": 285
+    },
+    {
+      "epoch": 0.7369758576874206,
+      "grad_norm": 0.7895886898040771,
+      "learning_rate": 2.924310786669023e-05,
+      "loss": 0.8006,
+      "step": 290
+    },
+    {
+      "epoch": 0.7496823379923762,
+      "grad_norm": 1.0612972974777222,
+      "learning_rate": 2.9203108399519295e-05,
+      "loss": 0.7724,
+      "step": 295
+    },
+    {
+      "epoch": 0.7623888182973316,
+      "grad_norm": 0.937221884727478,
+      "learning_rate": 2.9162107842563645e-05,
+      "loss": 0.7597,
+      "step": 300
+    },
+    {
+      "epoch": 0.7750952986022872,
+      "grad_norm": 0.9658941626548767,
+      "learning_rate": 2.9120109085700443e-05,
+      "loss": 0.7786,
+      "step": 305
+    },
+    {
+      "epoch": 0.7878017789072427,
+      "grad_norm": 1.1432039737701416,
+      "learning_rate": 2.9077115089163842e-05,
+      "loss": 0.8014,
+      "step": 310
+    },
+    {
+      "epoch": 0.8005082592121983,
+      "grad_norm": 0.9009749293327332,
+      "learning_rate": 2.903312888333631e-05,
+      "loss": 0.8,
+      "step": 315
+    },
+    {
+      "epoch": 0.8132147395171537,
+      "grad_norm": 1.0406774282455444,
+      "learning_rate": 2.8988153568535053e-05,
+      "loss": 0.743,
+      "step": 320
+    },
+    {
+      "epoch": 0.8259212198221093,
+      "grad_norm": 1.0573517084121704,
+      "learning_rate": 2.8942192314793486e-05,
+      "loss": 0.746,
+      "step": 325
+    },
+    {
+      "epoch": 0.8386277001270648,
+      "grad_norm": 0.9490292072296143,
+      "learning_rate": 2.8895248361637795e-05,
+      "loss": 0.7417,
+      "step": 330
+    },
+    {
+      "epoch": 0.8513341804320204,
+      "grad_norm": 1.0302095413208008,
+      "learning_rate": 2.8847325017858608e-05,
+      "loss": 0.7573,
+      "step": 335
+    },
+    {
+      "epoch": 0.8640406607369758,
+      "grad_norm": 0.8958389759063721,
+      "learning_rate": 2.879842566127778e-05,
+      "loss": 0.7489,
+      "step": 340
+    },
+    {
+      "epoch": 0.8767471410419314,
+      "grad_norm": 0.9970207214355469,
+      "learning_rate": 2.8748553738510296e-05,
+      "loss": 0.6952,
+      "step": 345
+    },
+    {
+      "epoch": 0.8894536213468869,
+      "grad_norm": 1.0113465785980225,
+      "learning_rate": 2.869771276472137e-05,
+      "loss": 0.6985,
+      "step": 350
+    },
+    {
+      "epoch": 0.9021601016518425,
+      "grad_norm": 0.9318073391914368,
+      "learning_rate": 2.8645906323378642e-05,
+      "loss": 0.7184,
+      "step": 355
+    },
+    {
+      "epoch": 0.9148665819567979,
+      "grad_norm": 0.9506980180740356,
+      "learning_rate": 2.8593138065999648e-05,
+      "loss": 0.6907,
+      "step": 360
+    },
+    {
+      "epoch": 0.9275730622617535,
+      "grad_norm": 0.9164155721664429,
+      "learning_rate": 2.8539411711894397e-05,
+      "loss": 0.7064,
+      "step": 365
+    },
+    {
+      "epoch": 0.940279542566709,
+      "grad_norm": 1.1175841093063354,
+      "learning_rate": 2.8484731047903274e-05,
+      "loss": 0.6761,
+      "step": 370
+    },
+    {
+      "epoch": 0.9529860228716646,
+      "grad_norm": 0.9533513188362122,
+      "learning_rate": 2.842909992813007e-05,
+      "loss": 0.6527,
+      "step": 375
+    },
+    {
+      "epoch": 0.96569250317662,
+      "grad_norm": 1.088073968887329,
+      "learning_rate": 2.8372522273670386e-05,
+      "loss": 0.7256,
+      "step": 380
+    },
+    {
+      "epoch": 0.9783989834815756,
+      "grad_norm": 0.9813941121101379,
+      "learning_rate": 2.8315002072335216e-05,
+      "loss": 0.6904,
+      "step": 385
+    },
+    {
+      "epoch": 0.9911054637865311,
+      "grad_norm": 1.0815173387527466,
+      "learning_rate": 2.8256543378369906e-05,
+      "loss": 0.7024,
+      "step": 390
+    },
+    {
+      "epoch": 1.0025412960609912,
+      "grad_norm": 0.9916452169418335,
+      "learning_rate": 2.8197150312168353e-05,
+      "loss": 0.648,
+      "step": 395
+    },
+    {
+      "epoch": 1.0152477763659467,
+      "grad_norm": 0.9849523901939392,
+      "learning_rate": 2.813682705998263e-05,
+      "loss": 0.5599,
+      "step": 400
+    },
+    {
+      "epoch": 1.0279542566709021,
+      "grad_norm": 1.178002119064331,
+      "learning_rate": 2.8075577873627888e-05,
+      "loss": 0.6203,
+      "step": 405
+    },
+    {
+      "epoch": 1.0406607369758576,
+      "grad_norm": 1.0375463962554932,
+      "learning_rate": 2.8013407070182688e-05,
+      "loss": 0.593,
+      "step": 410
+    },
+    {
+      "epoch": 1.0533672172808133,
+      "grad_norm": 1.0772303342819214,
+      "learning_rate": 2.795031903168472e-05,
+      "loss": 0.5981,
+      "step": 415
+    },
+    {
+      "epoch": 1.0660736975857688,
+      "grad_norm": 0.9816545248031616,
+      "learning_rate": 2.7886318204821925e-05,
+      "loss": 0.5964,
+      "step": 420
+    },
+    {
+      "epoch": 1.0787801778907242,
+      "grad_norm": 1.0742219686508179,
+      "learning_rate": 2.7821409100619083e-05,
+      "loss": 0.5081,
+      "step": 425
+    },
+    {
+      "epoch": 1.0914866581956797,
+      "grad_norm": 0.9489808082580566,
+      "learning_rate": 2.7755596294119863e-05,
+      "loss": 0.5634,
+      "step": 430
+    },
+    {
+      "epoch": 1.1041931385006354,
+      "grad_norm": 0.8840591907501221,
+      "learning_rate": 2.768888442406435e-05,
+      "loss": 0.5896,
+      "step": 435
+    },
+    {
+      "epoch": 1.1168996188055909,
+      "grad_norm": 1.1017783880233765,
+      "learning_rate": 2.76212781925621e-05,
+      "loss": 0.5543,
+      "step": 440
+    },
+    {
+      "epoch": 1.1296060991105463,
+      "grad_norm": 0.9155241847038269,
+      "learning_rate": 2.7552782364760694e-05,
+      "loss": 0.5682,
+      "step": 445
+    },
+    {
+      "epoch": 1.1423125794155018,
+      "grad_norm": 1.0706830024719238,
+      "learning_rate": 2.74834017685099e-05,
+      "loss": 0.5214,
+      "step": 450
+    },
+    {
+      "epoch": 1.1550190597204575,
+      "grad_norm": 1.0417948961257935,
+      "learning_rate": 2.7413141294021368e-05,
+      "loss": 0.5803,
+      "step": 455
+    },
+    {
+      "epoch": 1.167725540025413,
+      "grad_norm": 1.1154874563217163,
+      "learning_rate": 2.734200589352395e-05,
+      "loss": 0.5837,
+      "step": 460
+    },
+    {
+      "epoch": 1.1804320203303684,
+      "grad_norm": 0.9265525937080383,
+      "learning_rate": 2.727000058091466e-05,
+      "loss": 0.509,
+      "step": 465
+    },
+    {
+      "epoch": 1.1931385006353241,
+      "grad_norm": 1.1180102825164795,
+      "learning_rate": 2.7197130431405274e-05,
+      "loss": 0.5551,
+      "step": 470
+    },
+    {
+      "epoch": 1.2058449809402796,
+      "grad_norm": 1.0591485500335693,
+      "learning_rate": 2.7123400581164584e-05,
+      "loss": 0.5185,
+      "step": 475
+    },
+    {
+      "epoch": 1.218551461245235,
+      "grad_norm": 1.030959129333496,
+      "learning_rate": 2.7048816226956418e-05,
+      "loss": 0.5607,
+      "step": 480
+    },
+    {
+      "epoch": 1.2312579415501905,
+      "grad_norm": 1.2631689310073853,
+      "learning_rate": 2.6973382625773334e-05,
+      "loss": 0.5098,
+      "step": 485
+    },
+    {
+      "epoch": 1.2439644218551462,
+      "grad_norm": 1.231269121170044,
+      "learning_rate": 2.689710509446608e-05,
+      "loss": 0.5228,
+      "step": 490
+    },
+    {
+      "epoch": 1.2566709021601017,
+      "grad_norm": 1.1229807138442993,
+      "learning_rate": 2.6819989009368856e-05,
+      "loss": 0.5284,
+      "step": 495
+    },
+    {
+      "epoch": 1.2693773824650572,
+      "grad_norm": 1.0254383087158203,
+      "learning_rate": 2.6742039805920352e-05,
+      "loss": 0.5224,
+      "step": 500
+    },
+    {
+      "epoch": 1.2820838627700126,
+      "grad_norm": 1.0726102590560913,
+      "learning_rate": 2.6663262978280667e-05,
+      "loss": 0.4937,
+      "step": 505
+    },
+    {
+      "epoch": 1.2947903430749683,
+      "grad_norm": 1.1195117235183716,
+      "learning_rate": 2.6583664078944025e-05,
+      "loss": 0.4933,
+      "step": 510
+    },
+    {
+      "epoch": 1.3074968233799238,
+      "grad_norm": 1.0920395851135254,
+      "learning_rate": 2.6503248718347438e-05,
+      "loss": 0.468,
+      "step": 515
+    },
+    {
+      "epoch": 1.3202033036848793,
+      "grad_norm": 1.142539381980896,
+      "learning_rate": 2.6422022564475244e-05,
+      "loss": 0.506,
+      "step": 520
+    },
+    {
+      "epoch": 1.3329097839898347,
+      "grad_norm": 1.1334837675094604,
+      "learning_rate": 2.6339991342459612e-05,
+      "loss": 0.4724,
+      "step": 525
+    },
+    {
+      "epoch": 1.3456162642947904,
+      "grad_norm": 1.0070760250091553,
+      "learning_rate": 2.625716083417704e-05,
+      "loss": 0.4973,
+      "step": 530
+    },
+    {
+      "epoch": 1.358322744599746,
+      "grad_norm": 1.0452746152877808,
+      "learning_rate": 2.6173536877840767e-05,
+      "loss": 0.4788,
+      "step": 535
+    },
+    {
+      "epoch": 1.3710292249047014,
+      "grad_norm": 1.0710910558700562,
+      "learning_rate": 2.6089125367589327e-05,
+      "loss": 0.4638,
+      "step": 540
+    },
+    {
+      "epoch": 1.383735705209657,
+      "grad_norm": 1.1262043714523315,
+      "learning_rate": 2.600393225307108e-05,
+      "loss": 0.454,
+      "step": 545
+    },
+    {
+      "epoch": 1.3964421855146125,
+      "grad_norm": 1.2925057411193848,
+      "learning_rate": 2.5917963539024873e-05,
+      "loss": 0.5065,
+      "step": 550
+    },
+    {
+      "epoch": 1.409148665819568,
+      "grad_norm": 1.1768943071365356,
+      "learning_rate": 2.583122528485678e-05,
+      "loss": 0.4812,
+      "step": 555
+    },
+    {
+      "epoch": 1.4218551461245235,
+      "grad_norm": 1.0473438501358032,
+      "learning_rate": 2.574372360421304e-05,
+      "loss": 0.531,
+      "step": 560
+    },
+    {
+      "epoch": 1.434561626429479,
+      "grad_norm": 1.1397316455841064,
+      "learning_rate": 2.565546466454913e-05,
+      "loss": 0.4855,
+      "step": 565
+    },
+    {
+      "epoch": 1.4472681067344346,
+      "grad_norm": 1.1195589303970337,
+      "learning_rate": 2.556645468669506e-05,
+      "loss": 0.4697,
+      "step": 570
+    },
+    {
+      "epoch": 1.45997458703939,
+      "grad_norm": 1.0572574138641357,
+      "learning_rate": 2.547669994441688e-05,
+      "loss": 0.5084,
+      "step": 575
+    },
+    {
+      "epoch": 1.4726810673443456,
+      "grad_norm": 1.121638298034668,
+      "learning_rate": 2.538620676397455e-05,
+      "loss": 0.4843,
+      "step": 580
+    },
+    {
+      "epoch": 1.4853875476493013,
+      "grad_norm": 1.0831444263458252,
+      "learning_rate": 2.5294981523675954e-05,
+      "loss": 0.4599,
+      "step": 585
+    },
+    {
+      "epoch": 1.4980940279542567,
+      "grad_norm": 1.2113792896270752,
+      "learning_rate": 2.52030306534274e-05,
+      "loss": 0.4739,
+      "step": 590
+    },
+    {
+      "epoch": 1.5108005082592122,
+      "grad_norm": 1.015008568763733,
+      "learning_rate": 2.5110360634280375e-05,
+      "loss": 0.4583,
+      "step": 595
+    },
+    {
+      "epoch": 1.5235069885641677,
+      "grad_norm": 1.171127200126648,
+      "learning_rate": 2.5016977997974762e-05,
+      "loss": 0.4776,
+      "step": 600
+    },
+    {
+      "epoch": 1.5362134688691231,
+      "grad_norm": 1.1218819618225098,
+      "learning_rate": 2.492288932647844e-05,
+      "loss": 0.4267,
+      "step": 605
+    },
+    {
+      "epoch": 1.5489199491740788,
+      "grad_norm": 1.0391136407852173,
+      "learning_rate": 2.4828101251523367e-05,
+      "loss": 0.4188,
+      "step": 610
+    },
+    {
+      "epoch": 1.5616264294790343,
+      "grad_norm": 0.9929266571998596,
+      "learning_rate": 2.4732620454138147e-05,
+      "loss": 0.4219,
+      "step": 615
+    },
+    {
+      "epoch": 1.57433290978399,
+      "grad_norm": 1.246825098991394,
+      "learning_rate": 2.4636453664177147e-05,
+      "loss": 0.4374,
+      "step": 620
+    },
+    {
+      "epoch": 1.5870393900889455,
+      "grad_norm": 1.1008881330490112,
+      "learning_rate": 2.4539607659846103e-05,
+      "loss": 0.458,
+      "step": 625
+    },
+    {
+      "epoch": 1.599745870393901,
+      "grad_norm": 1.341241717338562,
+      "learning_rate": 2.444208926722443e-05,
+      "loss": 0.4315,
+      "step": 630
+    },
+    {
+      "epoch": 1.6124523506988564,
+      "grad_norm": 1.101243019104004,
+      "learning_rate": 2.4343905359784027e-05,
+      "loss": 0.4466,
+      "step": 635
+    },
+    {
+      "epoch": 1.6251588310038119,
+      "grad_norm": 1.1305433511734009,
+      "learning_rate": 2.424506285790486e-05,
+      "loss": 0.4303,
+      "step": 640
+    },
+    {
+      "epoch": 1.6378653113087673,
+      "grad_norm": 1.2521634101867676,
+      "learning_rate": 2.4145568728387162e-05,
+      "loss": 0.4058,
+      "step": 645
+    },
+    {
+      "epoch": 1.650571791613723,
+      "grad_norm": 1.2680494785308838,
+      "learning_rate": 2.404542998396041e-05,
+      "loss": 0.4248,
+      "step": 650
+    },
+    {
+      "epoch": 1.6632782719186785,
+      "grad_norm": 1.1652700901031494,
+      "learning_rate": 2.394465368278898e-05,
+      "loss": 0.3996,
+      "step": 655
+    },
+    {
+      "epoch": 1.6759847522236342,
+      "grad_norm": 1.0949989557266235,
+      "learning_rate": 2.384324692797474e-05,
+      "loss": 0.4128,
+      "step": 660
+    },
+    {
+      "epoch": 1.6886912325285897,
+      "grad_norm": 1.1956212520599365,
+      "learning_rate": 2.374121686705635e-05,
+      "loss": 0.4542,
+      "step": 665
+    },
+    {
+      "epoch": 1.7013977128335451,
+      "grad_norm": 1.0765902996063232,
+      "learning_rate": 2.3638570691505486e-05,
+      "loss": 0.3978,
+      "step": 670
+    },
+    {
+      "epoch": 1.7141041931385006,
+      "grad_norm": 1.2470359802246094,
+      "learning_rate": 2.3535315636219948e-05,
+      "loss": 0.4004,
+      "step": 675
+    },
+    {
+      "epoch": 1.726810673443456,
+      "grad_norm": 1.1302058696746826,
+      "learning_rate": 2.3431458979013727e-05,
+      "loss": 0.4148,
+      "step": 680
+    },
+    {
+      "epoch": 1.7395171537484115,
+      "grad_norm": 1.3136945962905884,
+      "learning_rate": 2.332700804010404e-05,
+      "loss": 0.4071,
+      "step": 685
+    },
+    {
+      "epoch": 1.7522236340533672,
+      "grad_norm": 1.0668866634368896,
+      "learning_rate": 2.3221970181595367e-05,
+      "loss": 0.3577,
+      "step": 690
+    },
+    {
+      "epoch": 1.7649301143583227,
+      "grad_norm": 1.1561533212661743,
+      "learning_rate": 2.3116352806960533e-05,
+      "loss": 0.3903,
+      "step": 695
+    },
+    {
+      "epoch": 1.7776365946632784,
+      "grad_norm": 1.2595140933990479,
+      "learning_rate": 2.301016336051891e-05,
+      "loss": 0.3801,
+      "step": 700
+    },
+    {
+      "epoch": 1.7903430749682339,
+      "grad_norm": 1.0591890811920166,
+      "learning_rate": 2.2903409326911674e-05,
+      "loss": 0.37,
+      "step": 705
+    },
+    {
+      "epoch": 1.8030495552731893,
+      "grad_norm": 1.1846657991409302,
+      "learning_rate": 2.279609823057429e-05,
+      "loss": 0.4001,
+      "step": 710
+    },
+    {
+      "epoch": 1.8157560355781448,
+      "grad_norm": 0.9889497756958008,
+      "learning_rate": 2.2688237635206147e-05,
+      "loss": 0.3738,
+      "step": 715
+    },
+    {
+      "epoch": 1.8284625158831003,
+      "grad_norm": 1.1138595342636108,
+      "learning_rate": 2.257983514323744e-05,
+      "loss": 0.4148,
+      "step": 720
+    },
+    {
+      "epoch": 1.841168996188056,
+      "grad_norm": 1.2167162895202637,
+      "learning_rate": 2.2470898395293332e-05,
+      "loss": 0.4046,
+      "step": 725
+    },
+    {
+      "epoch": 1.8538754764930114,
+      "grad_norm": 1.0153459310531616,
+      "learning_rate": 2.236143506965539e-05,
+      "loss": 0.4015,
+      "step": 730
+    },
+    {
+      "epoch": 1.8665819567979671,
+      "grad_norm": 1.159151554107666,
+      "learning_rate": 2.2251452881720435e-05,
+      "loss": 0.3683,
+      "step": 735
+    },
+    {
+      "epoch": 1.8792884371029226,
+      "grad_norm": 1.088222861289978,
+      "learning_rate": 2.2140959583456655e-05,
+      "loss": 0.356,
+      "step": 740
+    },
+    {
+      "epoch": 1.891994917407878,
+      "grad_norm": 1.2370007038116455,
+      "learning_rate": 2.20299629628573e-05,
+      "loss": 0.3809,
+      "step": 745
+    },
+    {
+      "epoch": 1.9047013977128335,
+      "grad_norm": 1.3320659399032593,
+      "learning_rate": 2.1918470843391703e-05,
+      "loss": 0.3802,
+      "step": 750
+    },
+    {
+      "epoch": 1.917407878017789,
+      "grad_norm": 1.0436190366744995,
+      "learning_rate": 2.1806491083453868e-05,
+      "loss": 0.3626,
+      "step": 755
+    },
+    {
+      "epoch": 1.9301143583227445,
+      "grad_norm": 1.369911789894104,
+      "learning_rate": 2.1694031575808577e-05,
+      "loss": 0.3418,
+      "step": 760
+    },
+    {
+      "epoch": 1.9428208386277002,
+      "grad_norm": 1.1903291940689087,
+      "learning_rate": 2.1581100247035074e-05,
+      "loss": 0.3619,
+      "step": 765
+    },
+    {
+      "epoch": 1.9555273189326556,
+      "grad_norm": 1.119796872138977,
+      "learning_rate": 2.146770505696838e-05,
+      "loss": 0.3274,
+      "step": 770
+    },
+    {
+      "epoch": 1.9682337992376113,
+      "grad_norm": 1.1661344766616821,
+      "learning_rate": 2.1353853998138252e-05,
+      "loss": 0.3933,
+      "step": 775
+    },
+    {
+      "epoch": 1.9809402795425668,
+      "grad_norm": 1.1047519445419312,
+      "learning_rate": 2.1239555095205815e-05,
+      "loss": 0.3464,
+      "step": 780
+    },
+    {
+      "epoch": 1.9936467598475223,
+      "grad_norm": 1.3075625896453857,
+      "learning_rate": 2.1124816404397983e-05,
+      "loss": 0.365,
+      "step": 785
+    },
+    {
+      "epoch": 2.0050825921219824,
+      "grad_norm": 1.1474101543426514,
+      "learning_rate": 2.1009646012939613e-05,
+      "loss": 0.3349,
+      "step": 790
+    },
+    {
+      "epoch": 2.017789072426938,
+      "grad_norm": 1.1730916500091553,
+      "learning_rate": 2.0894052038483494e-05,
+      "loss": 0.2965,
+      "step": 795
+    },
+    {
+      "epoch": 2.0304955527318933,
+      "grad_norm": 1.0433058738708496,
+      "learning_rate": 2.077804262853815e-05,
+      "loss": 0.2934,
+      "step": 800
+    },
+    {
+      "epoch": 2.0432020330368488,
+      "grad_norm": 1.1516586542129517,
+      "learning_rate": 2.066162595989363e-05,
+      "loss": 0.2743,
+      "step": 805
+    },
+    {
+      "epoch": 2.0559085133418042,
+      "grad_norm": 1.1744905710220337,
+      "learning_rate": 2.0544810238045123e-05,
+      "loss": 0.2614,
+      "step": 810
+    },
+    {
+      "epoch": 2.0686149936467597,
+      "grad_norm": 1.276524543762207,
+      "learning_rate": 2.0427603696614643e-05,
+      "loss": 0.2489,
+      "step": 815
+    },
+    {
+      "epoch": 2.081321473951715,
+      "grad_norm": 1.121077060699463,
+      "learning_rate": 2.031001459677065e-05,
+      "loss": 0.3004,
+      "step": 820
+    },
+    {
+      "epoch": 2.094027954256671,
+      "grad_norm": 1.1395784616470337,
+      "learning_rate": 2.019205122664582e-05,
+      "loss": 0.2813,
+      "step": 825
+    },
+    {
+      "epoch": 2.1067344345616266,
+      "grad_norm": 1.3531075716018677,
+      "learning_rate": 2.007372190075281e-05,
+      "loss": 0.2607,
+      "step": 830
+    },
+    {
+      "epoch": 2.119440914866582,
+      "grad_norm": 1.1465169191360474,
+      "learning_rate": 1.995503495939829e-05,
+      "loss": 0.2719,
+      "step": 835
+    },
+    {
+      "epoch": 2.1321473951715375,
+      "grad_norm": 1.0055954456329346,
+      "learning_rate": 1.9835998768095e-05,
+      "loss": 0.2677,
+      "step": 840
+    },
+    {
+      "epoch": 2.144853875476493,
+      "grad_norm": 1.3007574081420898,
+      "learning_rate": 1.9716621716972198e-05,
+      "loss": 0.2536,
+      "step": 845
+    },
+    {
+      "epoch": 2.1575603557814484,
+      "grad_norm": 1.222320795059204,
+      "learning_rate": 1.959691222018423e-05,
+      "loss": 0.252,
+      "step": 850
+    },
+    {
+      "epoch": 2.170266836086404,
+      "grad_norm": 1.1874042749404907,
+      "learning_rate": 1.947687871531751e-05,
+      "loss": 0.2424,
+      "step": 855
+    },
+    {
+      "epoch": 2.1829733163913594,
+      "grad_norm": 1.0278013944625854,
+      "learning_rate": 1.9356529662795776e-05,
+      "loss": 0.2641,
+      "step": 860
+    },
+    {
+      "epoch": 2.1956797966963153,
+      "grad_norm": 1.111448049545288,
+      "learning_rate": 1.9235873545283788e-05,
+      "loss": 0.2312,
+      "step": 865
+    },
+    {
+      "epoch": 2.2083862770012708,
+      "grad_norm": 1.1844691038131714,
+      "learning_rate": 1.9114918867089435e-05,
+      "loss": 0.2715,
+      "step": 870
+    },
+    {
+      "epoch": 2.2210927573062262,
+      "grad_norm": 1.1054425239562988,
+      "learning_rate": 1.8993674153564295e-05,
+      "loss": 0.2275,
+      "step": 875
+    },
+    {
+      "epoch": 2.2337992376111817,
+      "grad_norm": 1.1780076026916504,
+      "learning_rate": 1.887214795050277e-05,
+      "loss": 0.2619,
+      "step": 880
+    },
+    {
+      "epoch": 2.246505717916137,
+      "grad_norm": 1.0730093717575073,
+      "learning_rate": 1.875034882353972e-05,
+      "loss": 0.2644,
+      "step": 885
+    },
+    {
+      "epoch": 2.2592121982210926,
+      "grad_norm": 1.374300479888916,
+      "learning_rate": 1.8628285357546737e-05,
+      "loss": 0.2707,
+      "step": 890
+    },
+    {
+      "epoch": 2.271918678526048,
+      "grad_norm": 1.0528799295425415,
+      "learning_rate": 1.8505966156027042e-05,
+      "loss": 0.2281,
+      "step": 895
+    },
+    {
+      "epoch": 2.2846251588310036,
+      "grad_norm": 1.0651166439056396,
+      "learning_rate": 1.8383399840509084e-05,
+      "loss": 0.2373,
+      "step": 900
+    },
+    {
+      "epoch": 2.2973316391359595,
+      "grad_norm": 1.1552715301513672,
+      "learning_rate": 1.8260595049938856e-05,
+      "loss": 0.2387,
+      "step": 905
+    },
+    {
+      "epoch": 2.310038119440915,
+      "grad_norm": 1.1984320878982544,
+      "learning_rate": 1.8137560440070992e-05,
+      "loss": 0.2518,
+      "step": 910
+    },
+    {
+      "epoch": 2.3227445997458704,
+      "grad_norm": 1.0798629522323608,
+      "learning_rate": 1.8014304682858686e-05,
+      "loss": 0.2187,
+      "step": 915
+    },
+    {
+      "epoch": 2.335451080050826,
+      "grad_norm": 1.1161442995071411,
+      "learning_rate": 1.789083646584242e-05,
+      "loss": 0.2546,
+      "step": 920
+    },
+    {
+      "epoch": 2.3481575603557814,
+      "grad_norm": 1.2970261573791504,
+      "learning_rate": 1.7767164491537698e-05,
+      "loss": 0.2515,
+      "step": 925
+    },
+    {
+      "epoch": 2.360864040660737,
+      "grad_norm": 1.1087594032287598,
+      "learning_rate": 1.7643297476821588e-05,
+      "loss": 0.25,
+      "step": 930
+    },
+    {
+      "epoch": 2.3735705209656923,
+      "grad_norm": 1.2234079837799072,
+      "learning_rate": 1.7519244152318388e-05,
+      "loss": 0.2095,
+      "step": 935
+    },
+    {
+      "epoch": 2.3862770012706482,
+      "grad_norm": 1.1008849143981934,
+      "learning_rate": 1.7395013261784207e-05,
+      "loss": 0.2202,
+      "step": 940
+    },
+    {
+      "epoch": 2.3989834815756037,
+      "grad_norm": 1.0901923179626465,
+      "learning_rate": 1.7270613561490696e-05,
+      "loss": 0.2265,
+      "step": 945
+    },
+    {
+      "epoch": 2.411689961880559,
+      "grad_norm": 1.0975196361541748,
+      "learning_rate": 1.7146053819607867e-05,
+      "loss": 0.2256,
+      "step": 950
+    },
+    {
+      "epoch": 2.4243964421855146,
+      "grad_norm": 1.086605191230774,
+      "learning_rate": 1.7021342815586094e-05,
+      "loss": 0.2058,
+      "step": 955
+    },
+    {
+      "epoch": 2.43710292249047,
+      "grad_norm": 1.2081586122512817,
+      "learning_rate": 1.6896489339537267e-05,
+      "loss": 0.2541,
+      "step": 960
+    },
+    {
+      "epoch": 2.4498094027954256,
+      "grad_norm": 1.2829140424728394,
+      "learning_rate": 1.6771502191615264e-05,
+      "loss": 0.2367,
+      "step": 965
+    },
+    {
+      "epoch": 2.462515883100381,
+      "grad_norm": 0.9739468693733215,
+      "learning_rate": 1.6646390181395676e-05,
+      "loss": 0.2596,
+      "step": 970
+    },
+    {
+      "epoch": 2.4752223634053365,
+      "grad_norm": 1.1967625617980957,
+      "learning_rate": 1.6521162127254865e-05,
+      "loss": 0.2666,
+      "step": 975
+    },
+    {
+      "epoch": 2.4879288437102924,
+      "grad_norm": 1.1611639261245728,
+      "learning_rate": 1.6395826855748412e-05,
+      "loss": 0.1999,
+      "step": 980
+    },
+    {
+      "epoch": 2.500635324015248,
+      "grad_norm": 1.1321150064468384,
+      "learning_rate": 1.6270393200989e-05,
+      "loss": 0.2429,
+      "step": 985
+    },
+    {
+      "epoch": 2.5133418043202034,
+      "grad_norm": 0.9815230965614319,
+      "learning_rate": 1.614487000402373e-05,
+      "loss": 0.2242,
+      "step": 990
+    },
+    {
+      "epoch": 2.526048284625159,
+      "grad_norm": 1.096049427986145,
+      "learning_rate": 1.6019266112210996e-05,
+      "loss": 0.2343,
+      "step": 995
+    },
+    {
+      "epoch": 2.5387547649301143,
+      "grad_norm": 1.083471417427063,
+      "learning_rate": 1.5893590378596862e-05,
+      "loss": 0.2297,
+      "step": 1000
+    },
+    {
+      "epoch": 2.55146124523507,
+      "grad_norm": 1.1834745407104492,
+      "learning_rate": 1.5767851661291095e-05,
+      "loss": 0.1976,
+      "step": 1005
+    },
+    {
+      "epoch": 2.5641677255400253,
+      "grad_norm": 1.2436003684997559,
+      "learning_rate": 1.5642058822842773e-05,
+      "loss": 0.2048,
+      "step": 1010
+    },
+    {
+      "epoch": 2.576874205844981,
+      "grad_norm": 1.1318174600601196,
+      "learning_rate": 1.5516220729615672e-05,
+      "loss": 0.2309,
+      "step": 1015
+    },
+    {
+      "epoch": 2.5895806861499366,
+      "grad_norm": 1.167190670967102,
+      "learning_rate": 1.5390346251163263e-05,
+      "loss": 0.2133,
+      "step": 1020
+    },
+    {
+      "epoch": 2.602287166454892,
+      "grad_norm": 1.095115303993225,
+      "learning_rate": 1.526444425960362e-05,
+      "loss": 0.2285,
+      "step": 1025
+    },
+    {
+      "epoch": 2.6149936467598476,
+      "grad_norm": 1.158190369606018,
+      "learning_rate": 1.5138523628994016e-05,
+      "loss": 0.2152,
+      "step": 1030
+    },
+    {
+      "epoch": 2.627700127064803,
+      "grad_norm": 1.2040417194366455,
+      "learning_rate": 1.5012593234705515e-05,
+      "loss": 0.2215,
+      "step": 1035
+    },
+    {
+      "epoch": 2.6404066073697585,
+      "grad_norm": 1.102165937423706,
+      "learning_rate": 1.488666195279732e-05,
+      "loss": 0.1956,
+      "step": 1040
+    },
+    {
+      "epoch": 2.653113087674714,
+      "grad_norm": 1.3835619688034058,
+      "learning_rate": 1.4760738659391227e-05,
+      "loss": 0.2218,
+      "step": 1045
+    },
+    {
+      "epoch": 2.6658195679796695,
+      "grad_norm": 1.0833338499069214,
+      "learning_rate": 1.463483223004596e-05,
+      "loss": 0.2127,
+      "step": 1050
+    },
+    {
+      "epoch": 2.678526048284625,
+      "grad_norm": 0.9585152268409729,
+      "learning_rate": 1.4508951539131596e-05,
+      "loss": 0.2299,
+      "step": 1055
+    },
+    {
+      "epoch": 2.691232528589581,
+      "grad_norm": 1.0633128881454468,
+      "learning_rate": 1.4383105459204087e-05,
+      "loss": 0.2013,
+      "step": 1060
+    },
+    {
+      "epoch": 2.7039390088945363,
+      "grad_norm": 1.189265489578247,
+      "learning_rate": 1.4257302860379837e-05,
+      "loss": 0.2009,
+      "step": 1065
+    },
+    {
+      "epoch": 2.716645489199492,
+      "grad_norm": 1.2080296277999878,
+      "learning_rate": 1.4131552609710564e-05,
+      "loss": 0.2142,
+      "step": 1070
+    },
+    {
+      "epoch": 2.7293519695044473,
+      "grad_norm": 1.0356550216674805,
+      "learning_rate": 1.400586357055827e-05,
+      "loss": 0.1941,
+      "step": 1075
+    },
+    {
+      "epoch": 2.7420584498094027,
+      "grad_norm": 0.926860511302948,
+      "learning_rate": 1.3880244601970549e-05,
+      "loss": 0.2016,
+      "step": 1080
+    },
+    {
+      "epoch": 2.754764930114358,
+      "grad_norm": 1.039290189743042,
+      "learning_rate": 1.3754704558056124e-05,
+      "loss": 0.1964,
+      "step": 1085
+    },
+    {
+      "epoch": 2.767471410419314,
+      "grad_norm": 1.125374674797058,
+      "learning_rate": 1.3629252287360832e-05,
+      "loss": 0.2288,
+      "step": 1090
+    },
+    {
+      "epoch": 2.7801778907242696,
+      "grad_norm": 1.1088508367538452,
+      "learning_rate": 1.3503896632243902e-05,
+      "loss": 0.2034,
+      "step": 1095
+    },
+    {
+      "epoch": 2.792884371029225,
+      "grad_norm": 0.9535992741584778,
+      "learning_rate": 1.337864642825473e-05,
+      "loss": 0.2041,
+      "step": 1100
+    },
+    {
+      "epoch": 2.8055908513341805,
+      "grad_norm": 1.1275798082351685,
+      "learning_rate": 1.32535105035101e-05,
+      "loss": 0.2139,
+      "step": 1105
+    },
+    {
+      "epoch": 2.818297331639136,
+      "grad_norm": 1.090336561203003,
+      "learning_rate": 1.3128497678071961e-05,
+      "loss": 0.189,
+      "step": 1110
+    },
+    {
+      "epoch": 2.8310038119440915,
+      "grad_norm": 1.122253179550171,
+      "learning_rate": 1.3003616763325759e-05,
+      "loss": 0.2302,
+      "step": 1115
+    },
+    {
+      "epoch": 2.843710292249047,
+      "grad_norm": 1.1841496229171753,
+      "learning_rate": 1.2878876561359353e-05,
+      "loss": 0.1944,
+      "step": 1120
+    },
+    {
+      "epoch": 2.8564167725540024,
+      "grad_norm": 1.0980600118637085,
+      "learning_rate": 1.2754285864342624e-05,
+      "loss": 0.1969,
+      "step": 1125
+    },
+    {
+      "epoch": 2.869123252858958,
+      "grad_norm": 1.0629725456237793,
+      "learning_rate": 1.2629853453907769e-05,
+      "loss": 0.1788,
+      "step": 1130
+    },
+    {
+      "epoch": 2.8818297331639133,
+      "grad_norm": 1.1231316328048706,
+      "learning_rate": 1.2505588100530346e-05,
+      "loss": 0.2011,
+      "step": 1135
+    },
+    {
+      "epoch": 2.8945362134688692,
+      "grad_norm": 1.427032470703125,
+      "learning_rate": 1.2381498562911084e-05,
+      "loss": 0.1809,
+      "step": 1140
+    },
+    {
+      "epoch": 2.9072426937738247,
+      "grad_norm": 1.0919214487075806,
+      "learning_rate": 1.2257593587358531e-05,
+      "loss": 0.1898,
+      "step": 1145
+    },
+    {
+      "epoch": 2.91994917407878,
+      "grad_norm": 1.0702979564666748,
+      "learning_rate": 1.2133881907172599e-05,
+      "loss": 0.2107,
+      "step": 1150
+    },
+    {
+      "epoch": 2.9326556543837357,
+      "grad_norm": 1.2244259119033813,
+      "learning_rate": 1.2010372242028992e-05,
+      "loss": 0.1744,
+      "step": 1155
+    },
+    {
+      "epoch": 2.945362134688691,
+      "grad_norm": 1.0888618230819702,
+      "learning_rate": 1.188707329736463e-05,
+      "loss": 0.2126,
+      "step": 1160
+    },
+    {
+      "epoch": 2.9580686149936466,
+      "grad_norm": 1.0948479175567627,
+      "learning_rate": 1.176399376376402e-05,
+      "loss": 0.1866,
+      "step": 1165
+    },
+    {
+      "epoch": 2.9707750952986025,
+      "grad_norm": 1.1298916339874268,
+      "learning_rate": 1.1641142316346747e-05,
+      "loss": 0.1745,
+      "step": 1170
+    },
+    {
+      "epoch": 2.983481575603558,
+      "grad_norm": 1.0410146713256836,
+      "learning_rate": 1.1518527614156001e-05,
+      "loss": 0.1866,
+      "step": 1175
+    },
+    {
+      "epoch": 2.9961880559085134,
+      "grad_norm": 1.108626365661621,
+      "learning_rate": 1.1396158299548263e-05,
+      "loss": 0.1883,
+      "step": 1180
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1970,
+  "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": 1.6441611660516393e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

51_128_e5_3e-5/checkpoint-1182/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:245714c11904f4ba207f93b00b1888364e827c9293be5f8cd77d776439fa70ff
+size 7736

51_128_e5_3e-5/checkpoint-1182/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)

51_128_e5_3e-5/checkpoint-1576/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

51_128_e5_3e-5/checkpoint-1576/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": [
+    "down_proj",
+    "q_proj",
+    "up_proj",
+    "gate_proj",
+    "k_proj",
+    "o_proj",
+    "v_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

51_128_e5_3e-5/checkpoint-1576/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:86e614d6bc3f2815c62b6be83b3280134ef155b06f68c5123dd726ab0534b56f
+size 791751704

51_128_e5_3e-5/checkpoint-1576/latest

@@ -0,0 +1 @@
+global_step1576

51_128_e5_3e-5/checkpoint-1576/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9bda6eeb3c8bffb5db31a4acbde346f7d3d241e68b8653a31491784309a44e4d
+size 15920

51_128_e5_3e-5/checkpoint-1576/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b9128620509a51475744d82722a66b6b933a3f30f9724eeff426b9a20669861
+size 15920

51_128_e5_3e-5/checkpoint-1576/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:79999813cfe5c1e6f022c5fd3143622a926eb4ca3893808f533276de489c92fc
+size 15920

51_128_e5_3e-5/checkpoint-1576/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e82f9f3375bdade821eb48eb8adcfbc26eeccc489d9b3575b5d58138e93467f0
+size 15920

51_128_e5_3e-5/checkpoint-1576/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9893e6ff8aff255097a13c2246d20002c27402325afbc65d070272d5ee9fa85c
+size 15920

51_128_e5_3e-5/checkpoint-1576/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:584879dc1a09d02054f687c3448cdd6fe8a96338dbc1bb54cac908c81b96464d
+size 15920

51_128_e5_3e-5/checkpoint-1576/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0e2742e932f13f1690b63a69e257f1fd2ec112703eb193ce0f13ae7a91f64e52
+size 15920

51_128_e5_3e-5/checkpoint-1576/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:41565112754dd193db445c3fa3729a51594bbd6f31af4ea85fdce911fbcf3bfd
+size 15920

51_128_e5_3e-5/checkpoint-1576/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:356407294a682f7883269c98859d036df27a586328b517d949197ff4cee80f5a
+size 1064

51_128_e5_3e-5/checkpoint-1576/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
+  }
+}

51_128_e5_3e-5/checkpoint-1576/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
+}

51_128_e5_3e-5/checkpoint-1576/trainer_state.json

@@ -0,0 +1,2239 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 1576,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.012706480304955527,
+      "grad_norm": 1.0249743461608887,
+      "learning_rate": 1.2121212121212122e-06,
+      "loss": 1.2769,
+      "step": 5
+    },
+    {
+      "epoch": 0.025412960609911054,
+      "grad_norm": 0.894174337387085,
+      "learning_rate": 2.7272727272727272e-06,
+      "loss": 1.3117,
+      "step": 10
+    },
+    {
+      "epoch": 0.03811944091486658,
+      "grad_norm": 0.6720679998397827,
+      "learning_rate": 4.242424242424242e-06,
+      "loss": 1.33,
+      "step": 15
+    },
+    {
+      "epoch": 0.05082592121982211,
+      "grad_norm": 0.7012405395507812,
+      "learning_rate": 5.757575757575758e-06,
+      "loss": 1.3123,
+      "step": 20
+    },
+    {
+      "epoch": 0.06353240152477764,
+      "grad_norm": 0.8811719417572021,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 1.2455,
+      "step": 25
+    },
+    {
+      "epoch": 0.07623888182973317,
+      "grad_norm": 0.5775105357170105,
+      "learning_rate": 8.787878787878788e-06,
+      "loss": 1.2503,
+      "step": 30
+    },
+    {
+      "epoch": 0.08894536213468869,
+      "grad_norm": 0.6601129770278931,
+      "learning_rate": 1.0303030303030302e-05,
+      "loss": 1.2622,
+      "step": 35
+    },
+    {
+      "epoch": 0.10165184243964422,
+      "grad_norm": 0.563920795917511,
+      "learning_rate": 1.1818181818181819e-05,
+      "loss": 1.2293,
+      "step": 40
+    },
+    {
+      "epoch": 0.11435832274459974,
+      "grad_norm": 0.6447675824165344,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.1266,
+      "step": 45
+    },
+    {
+      "epoch": 0.12706480304955528,
+      "grad_norm": 0.6008123755455017,
+      "learning_rate": 1.484848484848485e-05,
+      "loss": 1.1518,
+      "step": 50
+    },
+    {
+      "epoch": 0.1397712833545108,
+      "grad_norm": 0.5885509848594666,
+      "learning_rate": 1.6363636363636363e-05,
+      "loss": 1.2858,
+      "step": 55
+    },
+    {
+      "epoch": 0.15247776365946633,
+      "grad_norm": 0.47118306159973145,
+      "learning_rate": 1.7878787878787877e-05,
+      "loss": 1.1583,
+      "step": 60
+    },
+    {
+      "epoch": 0.16518424396442186,
+      "grad_norm": 0.5283555388450623,
+      "learning_rate": 1.9393939393939395e-05,
+      "loss": 1.2376,
+      "step": 65
+    },
+    {
+      "epoch": 0.17789072426937738,
+      "grad_norm": 0.5695042014122009,
+      "learning_rate": 2.090909090909091e-05,
+      "loss": 1.2086,
+      "step": 70
+    },
+    {
+      "epoch": 0.1905972045743329,
+      "grad_norm": 0.7871887683868408,
+      "learning_rate": 2.2424242424242424e-05,
+      "loss": 1.178,
+      "step": 75
+    },
+    {
+      "epoch": 0.20330368487928843,
+      "grad_norm": 0.5051509141921997,
+      "learning_rate": 2.3939393939393942e-05,
+      "loss": 1.1244,
+      "step": 80
+    },
+    {
+      "epoch": 0.21601016518424396,
+      "grad_norm": 0.6592188477516174,
+      "learning_rate": 2.5454545454545457e-05,
+      "loss": 1.1719,
+      "step": 85
+    },
+    {
+      "epoch": 0.22871664548919948,
+      "grad_norm": 0.5462698340415955,
+      "learning_rate": 2.696969696969697e-05,
+      "loss": 1.1111,
+      "step": 90
+    },
+    {
+      "epoch": 0.241423125794155,
+      "grad_norm": 0.5268920063972473,
+      "learning_rate": 2.8484848484848486e-05,
+      "loss": 1.1048,
+      "step": 95
+    },
+    {
+      "epoch": 0.25412960609911056,
+      "grad_norm": 0.5009660124778748,
+      "learning_rate": 3e-05,
+      "loss": 1.1586,
+      "step": 100
+    },
+    {
+      "epoch": 0.2668360864040661,
+      "grad_norm": 0.6100103259086609,
+      "learning_rate": 2.999947137111298e-05,
+      "loss": 1.1005,
+      "step": 105
+    },
+    {
+      "epoch": 0.2795425667090216,
+      "grad_norm": 0.5694701075553894,
+      "learning_rate": 2.9997885521711738e-05,
+      "loss": 1.0613,
+      "step": 110
+    },
+    {
+      "epoch": 0.29224904701397714,
+      "grad_norm": 0.5084468722343445,
+      "learning_rate": 2.9995242563573035e-05,
+      "loss": 1.0681,
+      "step": 115
+    },
+    {
+      "epoch": 0.30495552731893266,
+      "grad_norm": 0.5901311635971069,
+      "learning_rate": 2.9991542682982747e-05,
+      "loss": 1.0505,
+      "step": 120
+    },
+    {
+      "epoch": 0.3176620076238882,
+      "grad_norm": 0.5251779556274414,
+      "learning_rate": 2.99867861407227e-05,
+      "loss": 1.0543,
+      "step": 125
+    },
+    {
+      "epoch": 0.3303684879288437,
+      "grad_norm": 0.5326008796691895,
+      "learning_rate": 2.9980973272052328e-05,
+      "loss": 1.0744,
+      "step": 130
+    },
+    {
+      "epoch": 0.34307496823379924,
+      "grad_norm": 0.5696294903755188,
+      "learning_rate": 2.9974104486684988e-05,
+      "loss": 1.0599,
+      "step": 135
+    },
+    {
+      "epoch": 0.35578144853875476,
+      "grad_norm": 0.5977292656898499,
+      "learning_rate": 2.996618026875914e-05,
+      "loss": 1.071,
+      "step": 140
+    },
+    {
+      "epoch": 0.3684879288437103,
+      "grad_norm": 0.5755591988563538,
+      "learning_rate": 2.995720117680417e-05,
+      "loss": 1.0214,
+      "step": 145
+    },
+    {
+      "epoch": 0.3811944091486658,
+      "grad_norm": 0.8020675778388977,
+      "learning_rate": 2.994716784370108e-05,
+      "loss": 1.0587,
+      "step": 150
+    },
+    {
+      "epoch": 0.39390088945362134,
+      "grad_norm": 0.6691463589668274,
+      "learning_rate": 2.9936080976637823e-05,
+      "loss": 0.9774,
+      "step": 155
+    },
+    {
+      "epoch": 0.40660736975857686,
+      "grad_norm": 0.6647286415100098,
+      "learning_rate": 2.992394135705949e-05,
+      "loss": 1.0251,
+      "step": 160
+    },
+    {
+      "epoch": 0.4193138500635324,
+      "grad_norm": 0.7553662061691284,
+      "learning_rate": 2.9910749840613233e-05,
+      "loss": 0.9626,
+      "step": 165
+    },
+    {
+      "epoch": 0.4320203303684879,
+      "grad_norm": 0.6578956842422485,
+      "learning_rate": 2.9896507357087928e-05,
+      "loss": 0.9649,
+      "step": 170
+    },
+    {
+      "epoch": 0.44472681067344344,
+      "grad_norm": 0.6698135137557983,
+      "learning_rate": 2.988121491034868e-05,
+      "loss": 0.9372,
+      "step": 175
+    },
+    {
+      "epoch": 0.45743329097839897,
+      "grad_norm": 0.7247338891029358,
+      "learning_rate": 2.9864873578266034e-05,
+      "loss": 0.9995,
+      "step": 180
+    },
+    {
+      "epoch": 0.4701397712833545,
+      "grad_norm": 0.7628229856491089,
+      "learning_rate": 2.9847484512640018e-05,
+      "loss": 0.9421,
+      "step": 185
+    },
+    {
+      "epoch": 0.48284625158831,
+      "grad_norm": 0.7909829020500183,
+      "learning_rate": 2.9829048939118944e-05,
+      "loss": 0.9142,
+      "step": 190
+    },
+    {
+      "epoch": 0.49555273189326554,
+      "grad_norm": 0.7012385725975037,
+      "learning_rate": 2.9809568157113047e-05,
+      "loss": 0.9399,
+      "step": 195
+    },
+    {
+      "epoch": 0.5082592121982211,
+      "grad_norm": 0.7712706327438354,
+      "learning_rate": 2.9789043539702875e-05,
+      "loss": 0.9837,
+      "step": 200
+    },
+    {
+      "epoch": 0.5209656925031766,
+      "grad_norm": 0.7406087517738342,
+      "learning_rate": 2.9767476533542513e-05,
+      "loss": 0.9507,
+      "step": 205
+    },
+    {
+      "epoch": 0.5336721728081322,
+      "grad_norm": 0.7858076095581055,
+      "learning_rate": 2.9744868658757628e-05,
+      "loss": 0.9529,
+      "step": 210
+    },
+    {
+      "epoch": 0.5463786531130876,
+      "grad_norm": 0.7638741731643677,
+      "learning_rate": 2.9721221508838302e-05,
+      "loss": 0.9502,
+      "step": 215
+    },
+    {
+      "epoch": 0.5590851334180432,
+      "grad_norm": 0.9117738604545593,
+      "learning_rate": 2.9696536750526748e-05,
+      "loss": 0.8913,
+      "step": 220
+    },
+    {
+      "epoch": 0.5717916137229987,
+      "grad_norm": 0.8238304853439331,
+      "learning_rate": 2.9670816123699812e-05,
+      "loss": 0.8704,
+      "step": 225
+    },
+    {
+      "epoch": 0.5844980940279543,
+      "grad_norm": 0.7925108075141907,
+      "learning_rate": 2.9644061441246323e-05,
+      "loss": 0.8565,
+      "step": 230
+    },
+    {
+      "epoch": 0.5972045743329097,
+      "grad_norm": 0.7725275754928589,
+      "learning_rate": 2.9616274588939364e-05,
+      "loss": 0.8292,
+      "step": 235
+    },
+    {
+      "epoch": 0.6099110546378653,
+      "grad_norm": 0.8474175333976746,
+      "learning_rate": 2.9587457525303305e-05,
+      "loss": 0.8822,
+      "step": 240
+    },
+    {
+      "epoch": 0.6226175349428208,
+      "grad_norm": 1.0394459962844849,
+      "learning_rate": 2.9557612281475776e-05,
+      "loss": 0.8602,
+      "step": 245
+    },
+    {
+      "epoch": 0.6353240152477764,
+      "grad_norm": 0.9201862812042236,
+      "learning_rate": 2.9526740961064516e-05,
+      "loss": 0.8374,
+      "step": 250
+    },
+    {
+      "epoch": 0.6480304955527318,
+      "grad_norm": 0.9244139790534973,
+      "learning_rate": 2.9494845739999103e-05,
+      "loss": 0.8361,
+      "step": 255
+    },
+    {
+      "epoch": 0.6607369758576874,
+      "grad_norm": 0.9039394855499268,
+      "learning_rate": 2.9461928866377553e-05,
+      "loss": 0.8261,
+      "step": 260
+    },
+    {
+      "epoch": 0.6734434561626429,
+      "grad_norm": 0.7772687673568726,
+      "learning_rate": 2.942799266030791e-05,
+      "loss": 0.8687,
+      "step": 265
+    },
+    {
+      "epoch": 0.6861499364675985,
+      "grad_norm": 1.1102648973464966,
+      "learning_rate": 2.9393039513744684e-05,
+      "loss": 0.8003,
+      "step": 270
+    },
+    {
+      "epoch": 0.6988564167725541,
+      "grad_norm": 0.8548868298530579,
+      "learning_rate": 2.9357071890320262e-05,
+      "loss": 0.8169,
+      "step": 275
+    },
+    {
+      "epoch": 0.7115628970775095,
+      "grad_norm": 0.8160343766212463,
+      "learning_rate": 2.9320092325171292e-05,
+      "loss": 0.7822,
+      "step": 280
+    },
+    {
+      "epoch": 0.7242693773824651,
+      "grad_norm": 0.825340986251831,
+      "learning_rate": 2.9282103424759935e-05,
+      "loss": 0.7904,
+      "step": 285
+    },
+    {
+      "epoch": 0.7369758576874206,
+      "grad_norm": 0.7895886898040771,
+      "learning_rate": 2.924310786669023e-05,
+      "loss": 0.8006,
+      "step": 290
+    },
+    {
+      "epoch": 0.7496823379923762,
+      "grad_norm": 1.0612972974777222,
+      "learning_rate": 2.9203108399519295e-05,
+      "loss": 0.7724,
+      "step": 295
+    },
+    {
+      "epoch": 0.7623888182973316,
+      "grad_norm": 0.937221884727478,
+      "learning_rate": 2.9162107842563645e-05,
+      "loss": 0.7597,
+      "step": 300
+    },
+    {
+      "epoch": 0.7750952986022872,
+      "grad_norm": 0.9658941626548767,
+      "learning_rate": 2.9120109085700443e-05,
+      "loss": 0.7786,
+      "step": 305
+    },
+    {
+      "epoch": 0.7878017789072427,
+      "grad_norm": 1.1432039737701416,
+      "learning_rate": 2.9077115089163842e-05,
+      "loss": 0.8014,
+      "step": 310
+    },
+    {
+      "epoch": 0.8005082592121983,
+      "grad_norm": 0.9009749293327332,
+      "learning_rate": 2.903312888333631e-05,
+      "loss": 0.8,
+      "step": 315
+    },
+    {
+      "epoch": 0.8132147395171537,
+      "grad_norm": 1.0406774282455444,
+      "learning_rate": 2.8988153568535053e-05,
+      "loss": 0.743,
+      "step": 320
+    },
+    {
+      "epoch": 0.8259212198221093,
+      "grad_norm": 1.0573517084121704,
+      "learning_rate": 2.8942192314793486e-05,
+      "loss": 0.746,
+      "step": 325
+    },
+    {
+      "epoch": 0.8386277001270648,
+      "grad_norm": 0.9490292072296143,
+      "learning_rate": 2.8895248361637795e-05,
+      "loss": 0.7417,
+      "step": 330
+    },
+    {
+      "epoch": 0.8513341804320204,
+      "grad_norm": 1.0302095413208008,
+      "learning_rate": 2.8847325017858608e-05,
+      "loss": 0.7573,
+      "step": 335
+    },
+    {
+      "epoch": 0.8640406607369758,
+      "grad_norm": 0.8958389759063721,
+      "learning_rate": 2.879842566127778e-05,
+      "loss": 0.7489,
+      "step": 340
+    },
+    {
+      "epoch": 0.8767471410419314,
+      "grad_norm": 0.9970207214355469,
+      "learning_rate": 2.8748553738510296e-05,
+      "loss": 0.6952,
+      "step": 345
+    },
+    {
+      "epoch": 0.8894536213468869,
+      "grad_norm": 1.0113465785980225,
+      "learning_rate": 2.869771276472137e-05,
+      "loss": 0.6985,
+      "step": 350
+    },
+    {
+      "epoch": 0.9021601016518425,
+      "grad_norm": 0.9318073391914368,
+      "learning_rate": 2.8645906323378642e-05,
+      "loss": 0.7184,
+      "step": 355
+    },
+    {
+      "epoch": 0.9148665819567979,
+      "grad_norm": 0.9506980180740356,
+      "learning_rate": 2.8593138065999648e-05,
+      "loss": 0.6907,
+      "step": 360
+    },
+    {
+      "epoch": 0.9275730622617535,
+      "grad_norm": 0.9164155721664429,
+      "learning_rate": 2.8539411711894397e-05,
+      "loss": 0.7064,
+      "step": 365
+    },
+    {
+      "epoch": 0.940279542566709,
+      "grad_norm": 1.1175841093063354,
+      "learning_rate": 2.8484731047903274e-05,
+      "loss": 0.6761,
+      "step": 370
+    },
+    {
+      "epoch": 0.9529860228716646,
+      "grad_norm": 0.9533513188362122,
+      "learning_rate": 2.842909992813007e-05,
+      "loss": 0.6527,
+      "step": 375
+    },
+    {
+      "epoch": 0.96569250317662,
+      "grad_norm": 1.088073968887329,
+      "learning_rate": 2.8372522273670386e-05,
+      "loss": 0.7256,
+      "step": 380
+    },
+    {
+      "epoch": 0.9783989834815756,
+      "grad_norm": 0.9813941121101379,
+      "learning_rate": 2.8315002072335216e-05,
+      "loss": 0.6904,
+      "step": 385
+    },
+    {
+      "epoch": 0.9911054637865311,
+      "grad_norm": 1.0815173387527466,
+      "learning_rate": 2.8256543378369906e-05,
+      "loss": 0.7024,
+      "step": 390
+    },
+    {
+      "epoch": 1.0025412960609912,
+      "grad_norm": 0.9916452169418335,
+      "learning_rate": 2.8197150312168353e-05,
+      "loss": 0.648,
+      "step": 395
+    },
+    {
+      "epoch": 1.0152477763659467,
+      "grad_norm": 0.9849523901939392,
+      "learning_rate": 2.813682705998263e-05,
+      "loss": 0.5599,
+      "step": 400
+    },
+    {
+      "epoch": 1.0279542566709021,
+      "grad_norm": 1.178002119064331,
+      "learning_rate": 2.8075577873627888e-05,
+      "loss": 0.6203,
+      "step": 405
+    },
+    {
+      "epoch": 1.0406607369758576,
+      "grad_norm": 1.0375463962554932,
+      "learning_rate": 2.8013407070182688e-05,
+      "loss": 0.593,
+      "step": 410
+    },
+    {
+      "epoch": 1.0533672172808133,
+      "grad_norm": 1.0772303342819214,
+      "learning_rate": 2.795031903168472e-05,
+      "loss": 0.5981,
+      "step": 415
+    },
+    {
+      "epoch": 1.0660736975857688,
+      "grad_norm": 0.9816545248031616,
+      "learning_rate": 2.7886318204821925e-05,
+      "loss": 0.5964,
+      "step": 420
+    },
+    {
+      "epoch": 1.0787801778907242,
+      "grad_norm": 1.0742219686508179,
+      "learning_rate": 2.7821409100619083e-05,
+      "loss": 0.5081,
+      "step": 425
+    },
+    {
+      "epoch": 1.0914866581956797,
+      "grad_norm": 0.9489808082580566,
+      "learning_rate": 2.7755596294119863e-05,
+      "loss": 0.5634,
+      "step": 430
+    },
+    {
+      "epoch": 1.1041931385006354,
+      "grad_norm": 0.8840591907501221,
+      "learning_rate": 2.768888442406435e-05,
+      "loss": 0.5896,
+      "step": 435
+    },
+    {
+      "epoch": 1.1168996188055909,
+      "grad_norm": 1.1017783880233765,
+      "learning_rate": 2.76212781925621e-05,
+      "loss": 0.5543,
+      "step": 440
+    },
+    {
+      "epoch": 1.1296060991105463,
+      "grad_norm": 0.9155241847038269,
+      "learning_rate": 2.7552782364760694e-05,
+      "loss": 0.5682,
+      "step": 445
+    },
+    {
+      "epoch": 1.1423125794155018,
+      "grad_norm": 1.0706830024719238,
+      "learning_rate": 2.74834017685099e-05,
+      "loss": 0.5214,
+      "step": 450
+    },
+    {
+      "epoch": 1.1550190597204575,
+      "grad_norm": 1.0417948961257935,
+      "learning_rate": 2.7413141294021368e-05,
+      "loss": 0.5803,
+      "step": 455
+    },
+    {
+      "epoch": 1.167725540025413,
+      "grad_norm": 1.1154874563217163,
+      "learning_rate": 2.734200589352395e-05,
+      "loss": 0.5837,
+      "step": 460
+    },
+    {
+      "epoch": 1.1804320203303684,
+      "grad_norm": 0.9265525937080383,
+      "learning_rate": 2.727000058091466e-05,
+      "loss": 0.509,
+      "step": 465
+    },
+    {
+      "epoch": 1.1931385006353241,
+      "grad_norm": 1.1180102825164795,
+      "learning_rate": 2.7197130431405274e-05,
+      "loss": 0.5551,
+      "step": 470
+    },
+    {
+      "epoch": 1.2058449809402796,
+      "grad_norm": 1.0591485500335693,
+      "learning_rate": 2.7123400581164584e-05,
+      "loss": 0.5185,
+      "step": 475
+    },
+    {
+      "epoch": 1.218551461245235,
+      "grad_norm": 1.030959129333496,
+      "learning_rate": 2.7048816226956418e-05,
+      "loss": 0.5607,
+      "step": 480
+    },
+    {
+      "epoch": 1.2312579415501905,
+      "grad_norm": 1.2631689310073853,
+      "learning_rate": 2.6973382625773334e-05,
+      "loss": 0.5098,
+      "step": 485
+    },
+    {
+      "epoch": 1.2439644218551462,
+      "grad_norm": 1.231269121170044,
+      "learning_rate": 2.689710509446608e-05,
+      "loss": 0.5228,
+      "step": 490
+    },
+    {
+      "epoch": 1.2566709021601017,
+      "grad_norm": 1.1229807138442993,
+      "learning_rate": 2.6819989009368856e-05,
+      "loss": 0.5284,
+      "step": 495
+    },
+    {
+      "epoch": 1.2693773824650572,
+      "grad_norm": 1.0254383087158203,
+      "learning_rate": 2.6742039805920352e-05,
+      "loss": 0.5224,
+      "step": 500
+    },
+    {
+      "epoch": 1.2820838627700126,
+      "grad_norm": 1.0726102590560913,
+      "learning_rate": 2.6663262978280667e-05,
+      "loss": 0.4937,
+      "step": 505
+    },
+    {
+      "epoch": 1.2947903430749683,
+      "grad_norm": 1.1195117235183716,
+      "learning_rate": 2.6583664078944025e-05,
+      "loss": 0.4933,
+      "step": 510
+    },
+    {
+      "epoch": 1.3074968233799238,
+      "grad_norm": 1.0920395851135254,
+      "learning_rate": 2.6503248718347438e-05,
+      "loss": 0.468,
+      "step": 515
+    },
+    {
+      "epoch": 1.3202033036848793,
+      "grad_norm": 1.142539381980896,
+      "learning_rate": 2.6422022564475244e-05,
+      "loss": 0.506,
+      "step": 520
+    },
+    {
+      "epoch": 1.3329097839898347,
+      "grad_norm": 1.1334837675094604,
+      "learning_rate": 2.6339991342459612e-05,
+      "loss": 0.4724,
+      "step": 525
+    },
+    {
+      "epoch": 1.3456162642947904,
+      "grad_norm": 1.0070760250091553,
+      "learning_rate": 2.625716083417704e-05,
+      "loss": 0.4973,
+      "step": 530
+    },
+    {
+      "epoch": 1.358322744599746,
+      "grad_norm": 1.0452746152877808,
+      "learning_rate": 2.6173536877840767e-05,
+      "loss": 0.4788,
+      "step": 535
+    },
+    {
+      "epoch": 1.3710292249047014,
+      "grad_norm": 1.0710910558700562,
+      "learning_rate": 2.6089125367589327e-05,
+      "loss": 0.4638,
+      "step": 540
+    },
+    {
+      "epoch": 1.383735705209657,
+      "grad_norm": 1.1262043714523315,
+      "learning_rate": 2.600393225307108e-05,
+      "loss": 0.454,
+      "step": 545
+    },
+    {
+      "epoch": 1.3964421855146125,
+      "grad_norm": 1.2925057411193848,
+      "learning_rate": 2.5917963539024873e-05,
+      "loss": 0.5065,
+      "step": 550
+    },
+    {
+      "epoch": 1.409148665819568,
+      "grad_norm": 1.1768943071365356,
+      "learning_rate": 2.583122528485678e-05,
+      "loss": 0.4812,
+      "step": 555
+    },
+    {
+      "epoch": 1.4218551461245235,
+      "grad_norm": 1.0473438501358032,
+      "learning_rate": 2.574372360421304e-05,
+      "loss": 0.531,
+      "step": 560
+    },
+    {
+      "epoch": 1.434561626429479,
+      "grad_norm": 1.1397316455841064,
+      "learning_rate": 2.565546466454913e-05,
+      "loss": 0.4855,
+      "step": 565
+    },
+    {
+      "epoch": 1.4472681067344346,
+      "grad_norm": 1.1195589303970337,
+      "learning_rate": 2.556645468669506e-05,
+      "loss": 0.4697,
+      "step": 570
+    },
+    {
+      "epoch": 1.45997458703939,
+      "grad_norm": 1.0572574138641357,
+      "learning_rate": 2.547669994441688e-05,
+      "loss": 0.5084,
+      "step": 575
+    },
+    {
+      "epoch": 1.4726810673443456,
+      "grad_norm": 1.121638298034668,
+      "learning_rate": 2.538620676397455e-05,
+      "loss": 0.4843,
+      "step": 580
+    },
+    {
+      "epoch": 1.4853875476493013,
+      "grad_norm": 1.0831444263458252,
+      "learning_rate": 2.5294981523675954e-05,
+      "loss": 0.4599,
+      "step": 585
+    },
+    {
+      "epoch": 1.4980940279542567,
+      "grad_norm": 1.2113792896270752,
+      "learning_rate": 2.52030306534274e-05,
+      "loss": 0.4739,
+      "step": 590
+    },
+    {
+      "epoch": 1.5108005082592122,
+      "grad_norm": 1.015008568763733,
+      "learning_rate": 2.5110360634280375e-05,
+      "loss": 0.4583,
+      "step": 595
+    },
+    {
+      "epoch": 1.5235069885641677,
+      "grad_norm": 1.171127200126648,
+      "learning_rate": 2.5016977997974762e-05,
+      "loss": 0.4776,
+      "step": 600
+    },
+    {
+      "epoch": 1.5362134688691231,
+      "grad_norm": 1.1218819618225098,
+      "learning_rate": 2.492288932647844e-05,
+      "loss": 0.4267,
+      "step": 605
+    },
+    {
+      "epoch": 1.5489199491740788,
+      "grad_norm": 1.0391136407852173,
+      "learning_rate": 2.4828101251523367e-05,
+      "loss": 0.4188,
+      "step": 610
+    },
+    {
+      "epoch": 1.5616264294790343,
+      "grad_norm": 0.9929266571998596,
+      "learning_rate": 2.4732620454138147e-05,
+      "loss": 0.4219,
+      "step": 615
+    },
+    {
+      "epoch": 1.57433290978399,
+      "grad_norm": 1.246825098991394,
+      "learning_rate": 2.4636453664177147e-05,
+      "loss": 0.4374,
+      "step": 620
+    },
+    {
+      "epoch": 1.5870393900889455,
+      "grad_norm": 1.1008881330490112,
+      "learning_rate": 2.4539607659846103e-05,
+      "loss": 0.458,
+      "step": 625
+    },
+    {
+      "epoch": 1.599745870393901,
+      "grad_norm": 1.341241717338562,
+      "learning_rate": 2.444208926722443e-05,
+      "loss": 0.4315,
+      "step": 630
+    },
+    {
+      "epoch": 1.6124523506988564,
+      "grad_norm": 1.101243019104004,
+      "learning_rate": 2.4343905359784027e-05,
+      "loss": 0.4466,
+      "step": 635
+    },
+    {
+      "epoch": 1.6251588310038119,
+      "grad_norm": 1.1305433511734009,
+      "learning_rate": 2.424506285790486e-05,
+      "loss": 0.4303,
+      "step": 640
+    },
+    {
+      "epoch": 1.6378653113087673,
+      "grad_norm": 1.2521634101867676,
+      "learning_rate": 2.4145568728387162e-05,
+      "loss": 0.4058,
+      "step": 645
+    },
+    {
+      "epoch": 1.650571791613723,
+      "grad_norm": 1.2680494785308838,
+      "learning_rate": 2.404542998396041e-05,
+      "loss": 0.4248,
+      "step": 650
+    },
+    {
+      "epoch": 1.6632782719186785,
+      "grad_norm": 1.1652700901031494,
+      "learning_rate": 2.394465368278898e-05,
+      "loss": 0.3996,
+      "step": 655
+    },
+    {
+      "epoch": 1.6759847522236342,
+      "grad_norm": 1.0949989557266235,
+      "learning_rate": 2.384324692797474e-05,
+      "loss": 0.4128,
+      "step": 660
+    },
+    {
+      "epoch": 1.6886912325285897,
+      "grad_norm": 1.1956212520599365,
+      "learning_rate": 2.374121686705635e-05,
+      "loss": 0.4542,
+      "step": 665
+    },
+    {
+      "epoch": 1.7013977128335451,
+      "grad_norm": 1.0765902996063232,
+      "learning_rate": 2.3638570691505486e-05,
+      "loss": 0.3978,
+      "step": 670
+    },
+    {
+      "epoch": 1.7141041931385006,
+      "grad_norm": 1.2470359802246094,
+      "learning_rate": 2.3535315636219948e-05,
+      "loss": 0.4004,
+      "step": 675
+    },
+    {
+      "epoch": 1.726810673443456,
+      "grad_norm": 1.1302058696746826,
+      "learning_rate": 2.3431458979013727e-05,
+      "loss": 0.4148,
+      "step": 680
+    },
+    {
+      "epoch": 1.7395171537484115,
+      "grad_norm": 1.3136945962905884,
+      "learning_rate": 2.332700804010404e-05,
+      "loss": 0.4071,
+      "step": 685
+    },
+    {
+      "epoch": 1.7522236340533672,
+      "grad_norm": 1.0668866634368896,
+      "learning_rate": 2.3221970181595367e-05,
+      "loss": 0.3577,
+      "step": 690
+    },
+    {
+      "epoch": 1.7649301143583227,
+      "grad_norm": 1.1561533212661743,
+      "learning_rate": 2.3116352806960533e-05,
+      "loss": 0.3903,
+      "step": 695
+    },
+    {
+      "epoch": 1.7776365946632784,
+      "grad_norm": 1.2595140933990479,
+      "learning_rate": 2.301016336051891e-05,
+      "loss": 0.3801,
+      "step": 700
+    },
+    {
+      "epoch": 1.7903430749682339,
+      "grad_norm": 1.0591890811920166,
+      "learning_rate": 2.2903409326911674e-05,
+      "loss": 0.37,
+      "step": 705
+    },
+    {
+      "epoch": 1.8030495552731893,
+      "grad_norm": 1.1846657991409302,
+      "learning_rate": 2.279609823057429e-05,
+      "loss": 0.4001,
+      "step": 710
+    },
+    {
+      "epoch": 1.8157560355781448,
+      "grad_norm": 0.9889497756958008,
+      "learning_rate": 2.2688237635206147e-05,
+      "loss": 0.3738,
+      "step": 715
+    },
+    {
+      "epoch": 1.8284625158831003,
+      "grad_norm": 1.1138595342636108,
+      "learning_rate": 2.257983514323744e-05,
+      "loss": 0.4148,
+      "step": 720
+    },
+    {
+      "epoch": 1.841168996188056,
+      "grad_norm": 1.2167162895202637,
+      "learning_rate": 2.2470898395293332e-05,
+      "loss": 0.4046,
+      "step": 725
+    },
+    {
+      "epoch": 1.8538754764930114,
+      "grad_norm": 1.0153459310531616,
+      "learning_rate": 2.236143506965539e-05,
+      "loss": 0.4015,
+      "step": 730
+    },
+    {
+      "epoch": 1.8665819567979671,
+      "grad_norm": 1.159151554107666,
+      "learning_rate": 2.2251452881720435e-05,
+      "loss": 0.3683,
+      "step": 735
+    },
+    {
+      "epoch": 1.8792884371029226,
+      "grad_norm": 1.088222861289978,
+      "learning_rate": 2.2140959583456655e-05,
+      "loss": 0.356,
+      "step": 740
+    },
+    {
+      "epoch": 1.891994917407878,
+      "grad_norm": 1.2370007038116455,
+      "learning_rate": 2.20299629628573e-05,
+      "loss": 0.3809,
+      "step": 745
+    },
+    {
+      "epoch": 1.9047013977128335,
+      "grad_norm": 1.3320659399032593,
+      "learning_rate": 2.1918470843391703e-05,
+      "loss": 0.3802,
+      "step": 750
+    },
+    {
+      "epoch": 1.917407878017789,
+      "grad_norm": 1.0436190366744995,
+      "learning_rate": 2.1806491083453868e-05,
+      "loss": 0.3626,
+      "step": 755
+    },
+    {
+      "epoch": 1.9301143583227445,
+      "grad_norm": 1.369911789894104,
+      "learning_rate": 2.1694031575808577e-05,
+      "loss": 0.3418,
+      "step": 760
+    },
+    {
+      "epoch": 1.9428208386277002,
+      "grad_norm": 1.1903291940689087,
+      "learning_rate": 2.1581100247035074e-05,
+      "loss": 0.3619,
+      "step": 765
+    },
+    {
+      "epoch": 1.9555273189326556,
+      "grad_norm": 1.119796872138977,
+      "learning_rate": 2.146770505696838e-05,
+      "loss": 0.3274,
+      "step": 770
+    },
+    {
+      "epoch": 1.9682337992376113,
+      "grad_norm": 1.1661344766616821,
+      "learning_rate": 2.1353853998138252e-05,
+      "loss": 0.3933,
+      "step": 775
+    },
+    {
+      "epoch": 1.9809402795425668,
+      "grad_norm": 1.1047519445419312,
+      "learning_rate": 2.1239555095205815e-05,
+      "loss": 0.3464,
+      "step": 780
+    },
+    {
+      "epoch": 1.9936467598475223,
+      "grad_norm": 1.3075625896453857,
+      "learning_rate": 2.1124816404397983e-05,
+      "loss": 0.365,
+      "step": 785
+    },
+    {
+      "epoch": 2.0050825921219824,
+      "grad_norm": 1.1474101543426514,
+      "learning_rate": 2.1009646012939613e-05,
+      "loss": 0.3349,
+      "step": 790
+    },
+    {
+      "epoch": 2.017789072426938,
+      "grad_norm": 1.1730916500091553,
+      "learning_rate": 2.0894052038483494e-05,
+      "loss": 0.2965,
+      "step": 795
+    },
+    {
+      "epoch": 2.0304955527318933,
+      "grad_norm": 1.0433058738708496,
+      "learning_rate": 2.077804262853815e-05,
+      "loss": 0.2934,
+      "step": 800
+    },
+    {
+      "epoch": 2.0432020330368488,
+      "grad_norm": 1.1516586542129517,
+      "learning_rate": 2.066162595989363e-05,
+      "loss": 0.2743,
+      "step": 805
+    },
+    {
+      "epoch": 2.0559085133418042,
+      "grad_norm": 1.1744905710220337,
+      "learning_rate": 2.0544810238045123e-05,
+      "loss": 0.2614,
+      "step": 810
+    },
+    {
+      "epoch": 2.0686149936467597,
+      "grad_norm": 1.276524543762207,
+      "learning_rate": 2.0427603696614643e-05,
+      "loss": 0.2489,
+      "step": 815
+    },
+    {
+      "epoch": 2.081321473951715,
+      "grad_norm": 1.121077060699463,
+      "learning_rate": 2.031001459677065e-05,
+      "loss": 0.3004,
+      "step": 820
+    },
+    {
+      "epoch": 2.094027954256671,
+      "grad_norm": 1.1395784616470337,
+      "learning_rate": 2.019205122664582e-05,
+      "loss": 0.2813,
+      "step": 825
+    },
+    {
+      "epoch": 2.1067344345616266,
+      "grad_norm": 1.3531075716018677,
+      "learning_rate": 2.007372190075281e-05,
+      "loss": 0.2607,
+      "step": 830
+    },
+    {
+      "epoch": 2.119440914866582,
+      "grad_norm": 1.1465169191360474,
+      "learning_rate": 1.995503495939829e-05,
+      "loss": 0.2719,
+      "step": 835
+    },
+    {
+      "epoch": 2.1321473951715375,
+      "grad_norm": 1.0055954456329346,
+      "learning_rate": 1.9835998768095e-05,
+      "loss": 0.2677,
+      "step": 840
+    },
+    {
+      "epoch": 2.144853875476493,
+      "grad_norm": 1.3007574081420898,
+      "learning_rate": 1.9716621716972198e-05,
+      "loss": 0.2536,
+      "step": 845
+    },
+    {
+      "epoch": 2.1575603557814484,
+      "grad_norm": 1.222320795059204,
+      "learning_rate": 1.959691222018423e-05,
+      "loss": 0.252,
+      "step": 850
+    },
+    {
+      "epoch": 2.170266836086404,
+      "grad_norm": 1.1874042749404907,
+      "learning_rate": 1.947687871531751e-05,
+      "loss": 0.2424,
+      "step": 855
+    },
+    {
+      "epoch": 2.1829733163913594,
+      "grad_norm": 1.0278013944625854,
+      "learning_rate": 1.9356529662795776e-05,
+      "loss": 0.2641,
+      "step": 860
+    },
+    {
+      "epoch": 2.1956797966963153,
+      "grad_norm": 1.111448049545288,
+      "learning_rate": 1.9235873545283788e-05,
+      "loss": 0.2312,
+      "step": 865
+    },
+    {
+      "epoch": 2.2083862770012708,
+      "grad_norm": 1.1844691038131714,
+      "learning_rate": 1.9114918867089435e-05,
+      "loss": 0.2715,
+      "step": 870
+    },
+    {
+      "epoch": 2.2210927573062262,
+      "grad_norm": 1.1054425239562988,
+      "learning_rate": 1.8993674153564295e-05,
+      "loss": 0.2275,
+      "step": 875
+    },
+    {
+      "epoch": 2.2337992376111817,
+      "grad_norm": 1.1780076026916504,
+      "learning_rate": 1.887214795050277e-05,
+      "loss": 0.2619,
+      "step": 880
+    },
+    {
+      "epoch": 2.246505717916137,
+      "grad_norm": 1.0730093717575073,
+      "learning_rate": 1.875034882353972e-05,
+      "loss": 0.2644,
+      "step": 885
+    },
+    {
+      "epoch": 2.2592121982210926,
+      "grad_norm": 1.374300479888916,
+      "learning_rate": 1.8628285357546737e-05,
+      "loss": 0.2707,
+      "step": 890
+    },
+    {
+      "epoch": 2.271918678526048,
+      "grad_norm": 1.0528799295425415,
+      "learning_rate": 1.8505966156027042e-05,
+      "loss": 0.2281,
+      "step": 895
+    },
+    {
+      "epoch": 2.2846251588310036,
+      "grad_norm": 1.0651166439056396,
+      "learning_rate": 1.8383399840509084e-05,
+      "loss": 0.2373,
+      "step": 900
+    },
+    {
+      "epoch": 2.2973316391359595,
+      "grad_norm": 1.1552715301513672,
+      "learning_rate": 1.8260595049938856e-05,
+      "loss": 0.2387,
+      "step": 905
+    },
+    {
+      "epoch": 2.310038119440915,
+      "grad_norm": 1.1984320878982544,
+      "learning_rate": 1.8137560440070992e-05,
+      "loss": 0.2518,
+      "step": 910
+    },
+    {
+      "epoch": 2.3227445997458704,
+      "grad_norm": 1.0798629522323608,
+      "learning_rate": 1.8014304682858686e-05,
+      "loss": 0.2187,
+      "step": 915
+    },
+    {
+      "epoch": 2.335451080050826,
+      "grad_norm": 1.1161442995071411,
+      "learning_rate": 1.789083646584242e-05,
+      "loss": 0.2546,
+      "step": 920
+    },
+    {
+      "epoch": 2.3481575603557814,
+      "grad_norm": 1.2970261573791504,
+      "learning_rate": 1.7767164491537698e-05,
+      "loss": 0.2515,
+      "step": 925
+    },
+    {
+      "epoch": 2.360864040660737,
+      "grad_norm": 1.1087594032287598,
+      "learning_rate": 1.7643297476821588e-05,
+      "loss": 0.25,
+      "step": 930
+    },
+    {
+      "epoch": 2.3735705209656923,
+      "grad_norm": 1.2234079837799072,
+      "learning_rate": 1.7519244152318388e-05,
+      "loss": 0.2095,
+      "step": 935
+    },
+    {
+      "epoch": 2.3862770012706482,
+      "grad_norm": 1.1008849143981934,
+      "learning_rate": 1.7395013261784207e-05,
+      "loss": 0.2202,
+      "step": 940
+    },
+    {
+      "epoch": 2.3989834815756037,
+      "grad_norm": 1.0901923179626465,
+      "learning_rate": 1.7270613561490696e-05,
+      "loss": 0.2265,
+      "step": 945
+    },
+    {
+      "epoch": 2.411689961880559,
+      "grad_norm": 1.0975196361541748,
+      "learning_rate": 1.7146053819607867e-05,
+      "loss": 0.2256,
+      "step": 950
+    },
+    {
+      "epoch": 2.4243964421855146,
+      "grad_norm": 1.086605191230774,
+      "learning_rate": 1.7021342815586094e-05,
+      "loss": 0.2058,
+      "step": 955
+    },
+    {
+      "epoch": 2.43710292249047,
+      "grad_norm": 1.2081586122512817,
+      "learning_rate": 1.6896489339537267e-05,
+      "loss": 0.2541,
+      "step": 960
+    },
+    {
+      "epoch": 2.4498094027954256,
+      "grad_norm": 1.2829140424728394,
+      "learning_rate": 1.6771502191615264e-05,
+      "loss": 0.2367,
+      "step": 965
+    },
+    {
+      "epoch": 2.462515883100381,
+      "grad_norm": 0.9739468693733215,
+      "learning_rate": 1.6646390181395676e-05,
+      "loss": 0.2596,
+      "step": 970
+    },
+    {
+      "epoch": 2.4752223634053365,
+      "grad_norm": 1.1967625617980957,
+      "learning_rate": 1.6521162127254865e-05,
+      "loss": 0.2666,
+      "step": 975
+    },
+    {
+      "epoch": 2.4879288437102924,
+      "grad_norm": 1.1611639261245728,
+      "learning_rate": 1.6395826855748412e-05,
+      "loss": 0.1999,
+      "step": 980
+    },
+    {
+      "epoch": 2.500635324015248,
+      "grad_norm": 1.1321150064468384,
+      "learning_rate": 1.6270393200989e-05,
+      "loss": 0.2429,
+      "step": 985
+    },
+    {
+      "epoch": 2.5133418043202034,
+      "grad_norm": 0.9815230965614319,
+      "learning_rate": 1.614487000402373e-05,
+      "loss": 0.2242,
+      "step": 990
+    },
+    {
+      "epoch": 2.526048284625159,
+      "grad_norm": 1.096049427986145,
+      "learning_rate": 1.6019266112210996e-05,
+      "loss": 0.2343,
+      "step": 995
+    },
+    {
+      "epoch": 2.5387547649301143,
+      "grad_norm": 1.083471417427063,
+      "learning_rate": 1.5893590378596862e-05,
+      "loss": 0.2297,
+      "step": 1000
+    },
+    {
+      "epoch": 2.55146124523507,
+      "grad_norm": 1.1834745407104492,
+      "learning_rate": 1.5767851661291095e-05,
+      "loss": 0.1976,
+      "step": 1005
+    },
+    {
+      "epoch": 2.5641677255400253,
+      "grad_norm": 1.2436003684997559,
+      "learning_rate": 1.5642058822842773e-05,
+      "loss": 0.2048,
+      "step": 1010
+    },
+    {
+      "epoch": 2.576874205844981,
+      "grad_norm": 1.1318174600601196,
+      "learning_rate": 1.5516220729615672e-05,
+      "loss": 0.2309,
+      "step": 1015
+    },
+    {
+      "epoch": 2.5895806861499366,
+      "grad_norm": 1.167190670967102,
+      "learning_rate": 1.5390346251163263e-05,
+      "loss": 0.2133,
+      "step": 1020
+    },
+    {
+      "epoch": 2.602287166454892,
+      "grad_norm": 1.095115303993225,
+      "learning_rate": 1.526444425960362e-05,
+      "loss": 0.2285,
+      "step": 1025
+    },
+    {
+      "epoch": 2.6149936467598476,
+      "grad_norm": 1.158190369606018,
+      "learning_rate": 1.5138523628994016e-05,
+      "loss": 0.2152,
+      "step": 1030
+    },
+    {
+      "epoch": 2.627700127064803,
+      "grad_norm": 1.2040417194366455,
+      "learning_rate": 1.5012593234705515e-05,
+      "loss": 0.2215,
+      "step": 1035
+    },
+    {
+      "epoch": 2.6404066073697585,
+      "grad_norm": 1.102165937423706,
+      "learning_rate": 1.488666195279732e-05,
+      "loss": 0.1956,
+      "step": 1040
+    },
+    {
+      "epoch": 2.653113087674714,
+      "grad_norm": 1.3835619688034058,
+      "learning_rate": 1.4760738659391227e-05,
+      "loss": 0.2218,
+      "step": 1045
+    },
+    {
+      "epoch": 2.6658195679796695,
+      "grad_norm": 1.0833338499069214,
+      "learning_rate": 1.463483223004596e-05,
+      "loss": 0.2127,
+      "step": 1050
+    },
+    {
+      "epoch": 2.678526048284625,
+      "grad_norm": 0.9585152268409729,
+      "learning_rate": 1.4508951539131596e-05,
+      "loss": 0.2299,
+      "step": 1055
+    },
+    {
+      "epoch": 2.691232528589581,
+      "grad_norm": 1.0633128881454468,
+      "learning_rate": 1.4383105459204087e-05,
+      "loss": 0.2013,
+      "step": 1060
+    },
+    {
+      "epoch": 2.7039390088945363,
+      "grad_norm": 1.189265489578247,
+      "learning_rate": 1.4257302860379837e-05,
+      "loss": 0.2009,
+      "step": 1065
+    },
+    {
+      "epoch": 2.716645489199492,
+      "grad_norm": 1.2080296277999878,
+      "learning_rate": 1.4131552609710564e-05,
+      "loss": 0.2142,
+      "step": 1070
+    },
+    {
+      "epoch": 2.7293519695044473,
+      "grad_norm": 1.0356550216674805,
+      "learning_rate": 1.400586357055827e-05,
+      "loss": 0.1941,
+      "step": 1075
+    },
+    {
+      "epoch": 2.7420584498094027,
+      "grad_norm": 0.926860511302948,
+      "learning_rate": 1.3880244601970549e-05,
+      "loss": 0.2016,
+      "step": 1080
+    },
+    {
+      "epoch": 2.754764930114358,
+      "grad_norm": 1.039290189743042,
+      "learning_rate": 1.3754704558056124e-05,
+      "loss": 0.1964,
+      "step": 1085
+    },
+    {
+      "epoch": 2.767471410419314,
+      "grad_norm": 1.125374674797058,
+      "learning_rate": 1.3629252287360832e-05,
+      "loss": 0.2288,
+      "step": 1090
+    },
+    {
+      "epoch": 2.7801778907242696,
+      "grad_norm": 1.1088508367538452,
+      "learning_rate": 1.3503896632243902e-05,
+      "loss": 0.2034,
+      "step": 1095
+    },
+    {
+      "epoch": 2.792884371029225,
+      "grad_norm": 0.9535992741584778,
+      "learning_rate": 1.337864642825473e-05,
+      "loss": 0.2041,
+      "step": 1100
+    },
+    {
+      "epoch": 2.8055908513341805,
+      "grad_norm": 1.1275798082351685,
+      "learning_rate": 1.32535105035101e-05,
+      "loss": 0.2139,
+      "step": 1105
+    },
+    {
+      "epoch": 2.818297331639136,
+      "grad_norm": 1.090336561203003,
+      "learning_rate": 1.3128497678071961e-05,
+      "loss": 0.189,
+      "step": 1110
+    },
+    {
+      "epoch": 2.8310038119440915,
+      "grad_norm": 1.122253179550171,
+      "learning_rate": 1.3003616763325759e-05,
+      "loss": 0.2302,
+      "step": 1115
+    },
+    {
+      "epoch": 2.843710292249047,
+      "grad_norm": 1.1841496229171753,
+      "learning_rate": 1.2878876561359353e-05,
+      "loss": 0.1944,
+      "step": 1120
+    },
+    {
+      "epoch": 2.8564167725540024,
+      "grad_norm": 1.0980600118637085,
+      "learning_rate": 1.2754285864342624e-05,
+      "loss": 0.1969,
+      "step": 1125
+    },
+    {
+      "epoch": 2.869123252858958,
+      "grad_norm": 1.0629725456237793,
+      "learning_rate": 1.2629853453907769e-05,
+      "loss": 0.1788,
+      "step": 1130
+    },
+    {
+      "epoch": 2.8818297331639133,
+      "grad_norm": 1.1231316328048706,
+      "learning_rate": 1.2505588100530346e-05,
+      "loss": 0.2011,
+      "step": 1135
+    },
+    {
+      "epoch": 2.8945362134688692,
+      "grad_norm": 1.427032470703125,
+      "learning_rate": 1.2381498562911084e-05,
+      "loss": 0.1809,
+      "step": 1140
+    },
+    {
+      "epoch": 2.9072426937738247,
+      "grad_norm": 1.0919214487075806,
+      "learning_rate": 1.2257593587358531e-05,
+      "loss": 0.1898,
+      "step": 1145
+    },
+    {
+      "epoch": 2.91994917407878,
+      "grad_norm": 1.0702979564666748,
+      "learning_rate": 1.2133881907172599e-05,
+      "loss": 0.2107,
+      "step": 1150
+    },
+    {
+      "epoch": 2.9326556543837357,
+      "grad_norm": 1.2244259119033813,
+      "learning_rate": 1.2010372242028992e-05,
+      "loss": 0.1744,
+      "step": 1155
+    },
+    {
+      "epoch": 2.945362134688691,
+      "grad_norm": 1.0888618230819702,
+      "learning_rate": 1.188707329736463e-05,
+      "loss": 0.2126,
+      "step": 1160
+    },
+    {
+      "epoch": 2.9580686149936466,
+      "grad_norm": 1.0948479175567627,
+      "learning_rate": 1.176399376376402e-05,
+      "loss": 0.1866,
+      "step": 1165
+    },
+    {
+      "epoch": 2.9707750952986025,
+      "grad_norm": 1.1298916339874268,
+      "learning_rate": 1.1641142316346747e-05,
+      "loss": 0.1745,
+      "step": 1170
+    },
+    {
+      "epoch": 2.983481575603558,
+      "grad_norm": 1.0410146713256836,
+      "learning_rate": 1.1518527614156001e-05,
+      "loss": 0.1866,
+      "step": 1175
+    },
+    {
+      "epoch": 2.9961880559085134,
+      "grad_norm": 1.108626365661621,
+      "learning_rate": 1.1396158299548263e-05,
+      "loss": 0.1883,
+      "step": 1180
+    },
+    {
+      "epoch": 3.007623888182973,
+      "grad_norm": 1.0174769163131714,
+      "learning_rate": 1.127404299758413e-05,
+      "loss": 0.1291,
+      "step": 1185
+    },
+    {
+      "epoch": 3.020330368487929,
+      "grad_norm": 0.8531823754310608,
+      "learning_rate": 1.115219031542043e-05,
+      "loss": 0.1485,
+      "step": 1190
+    },
+    {
+      "epoch": 3.0330368487928845,
+      "grad_norm": 1.010972023010254,
+      "learning_rate": 1.1030608841703534e-05,
+      "loss": 0.1512,
+      "step": 1195
+    },
+    {
+      "epoch": 3.04574332909784,
+      "grad_norm": 1.135859727859497,
+      "learning_rate": 1.0909307145963994e-05,
+      "loss": 0.1391,
+      "step": 1200
+    },
+    {
+      "epoch": 3.0584498094027954,
+      "grad_norm": 0.9453634023666382,
+      "learning_rate": 1.078829377801252e-05,
+      "loss": 0.1314,
+      "step": 1205
+    },
+    {
+      "epoch": 3.071156289707751,
+      "grad_norm": 1.096675992012024,
+      "learning_rate": 1.0667577267337388e-05,
+      "loss": 0.1254,
+      "step": 1210
+    },
+    {
+      "epoch": 3.0838627700127064,
+      "grad_norm": 1.1064974069595337,
+      "learning_rate": 1.0547166122503224e-05,
+      "loss": 0.1473,
+      "step": 1215
+    },
+    {
+      "epoch": 3.096569250317662,
+      "grad_norm": 1.0369409322738647,
+      "learning_rate": 1.0427068830551293e-05,
+      "loss": 0.1328,
+      "step": 1220
+    },
+    {
+      "epoch": 3.1092757306226178,
+      "grad_norm": 1.106688141822815,
+      "learning_rate": 1.0307293856401296e-05,
+      "loss": 0.1473,
+      "step": 1225
+    },
+    {
+      "epoch": 3.121982210927573,
+      "grad_norm": 1.0409021377563477,
+      "learning_rate": 1.018784964225473e-05,
+      "loss": 0.1224,
+      "step": 1230
+    },
+    {
+      "epoch": 3.1346886912325287,
+      "grad_norm": 1.070247769355774,
+      "learning_rate": 1.0068744606999869e-05,
+      "loss": 0.1434,
+      "step": 1235
+    },
+    {
+      "epoch": 3.147395171537484,
+      "grad_norm": 0.8816561102867126,
+      "learning_rate": 9.949987145618344e-06,
+      "loss": 0.1257,
+      "step": 1240
+    },
+    {
+      "epoch": 3.1601016518424396,
+      "grad_norm": 1.1240105628967285,
+      "learning_rate": 9.831585628593433e-06,
+      "loss": 0.1454,
+      "step": 1245
+    },
+    {
+      "epoch": 3.172808132147395,
+      "grad_norm": 1.065769910812378,
+      "learning_rate": 9.713548401320091e-06,
+      "loss": 0.1408,
+      "step": 1250
+    },
+    {
+      "epoch": 3.1855146124523506,
+      "grad_norm": 0.8838290572166443,
+      "learning_rate": 9.595883783516732e-06,
+      "loss": 0.1255,
+      "step": 1255
+    },
+    {
+      "epoch": 3.198221092757306,
+      "grad_norm": 1.0674669742584229,
+      "learning_rate": 9.478600068638825e-06,
+      "loss": 0.1387,
+      "step": 1260
+    },
+    {
+      "epoch": 3.210927573062262,
+      "grad_norm": 1.0269787311553955,
+      "learning_rate": 9.361705523294304e-06,
+      "loss": 0.1425,
+      "step": 1265
+    },
+    {
+      "epoch": 3.2236340533672174,
+      "grad_norm": 0.9741991758346558,
+      "learning_rate": 9.245208386660972e-06,
+      "loss": 0.1254,
+      "step": 1270
+    },
+    {
+      "epoch": 3.236340533672173,
+      "grad_norm": 1.0631365776062012,
+      "learning_rate": 9.129116869905714e-06,
+      "loss": 0.1316,
+      "step": 1275
+    },
+    {
+      "epoch": 3.2490470139771284,
+      "grad_norm": 0.91664719581604,
+      "learning_rate": 9.013439155605777e-06,
+      "loss": 0.1269,
+      "step": 1280
+    },
+    {
+      "epoch": 3.261753494282084,
+      "grad_norm": 0.9208909869194031,
+      "learning_rate": 8.898183397171995e-06,
+      "loss": 0.1353,
+      "step": 1285
+    },
+    {
+      "epoch": 3.2744599745870393,
+      "grad_norm": 0.9317910671234131,
+      "learning_rate": 8.783357718274153e-06,
+      "loss": 0.1183,
+      "step": 1290
+    },
+    {
+      "epoch": 3.2871664548919948,
+      "grad_norm": 1.0099849700927734,
+      "learning_rate": 8.668970212268356e-06,
+      "loss": 0.1549,
+      "step": 1295
+    },
+    {
+      "epoch": 3.2998729351969507,
+      "grad_norm": 1.0487933158874512,
+      "learning_rate": 8.555028941626612e-06,
+      "loss": 0.1224,
+      "step": 1300
+    },
+    {
+      "epoch": 3.312579415501906,
+      "grad_norm": 0.9620557427406311,
+      "learning_rate": 8.44154193736852e-06,
+      "loss": 0.1291,
+      "step": 1305
+    },
+    {
+      "epoch": 3.3252858958068616,
+      "grad_norm": 0.9691675901412964,
+      "learning_rate": 8.328517198495258e-06,
+      "loss": 0.1283,
+      "step": 1310
+    },
+    {
+      "epoch": 3.337992376111817,
+      "grad_norm": 0.9732845425605774,
+      "learning_rate": 8.215962691425736e-06,
+      "loss": 0.1204,
+      "step": 1315
+    },
+    {
+      "epoch": 3.3506988564167726,
+      "grad_norm": 1.1657145023345947,
+      "learning_rate": 8.103886349435142e-06,
+      "loss": 0.1305,
+      "step": 1320
+    },
+    {
+      "epoch": 3.363405336721728,
+      "grad_norm": 1.0779279470443726,
+      "learning_rate": 7.992296072095716e-06,
+      "loss": 0.1234,
+      "step": 1325
+    },
+    {
+      "epoch": 3.3761118170266835,
+      "grad_norm": 1.0233560800552368,
+      "learning_rate": 7.881199724720023e-06,
+      "loss": 0.1269,
+      "step": 1330
+    },
+    {
+      "epoch": 3.388818297331639,
+      "grad_norm": 0.9357303977012634,
+      "learning_rate": 7.770605137806513e-06,
+      "loss": 0.1392,
+      "step": 1335
+    },
+    {
+      "epoch": 3.4015247776365944,
+      "grad_norm": 0.970789909362793,
+      "learning_rate": 7.660520106487648e-06,
+      "loss": 0.1148,
+      "step": 1340
+    },
+    {
+      "epoch": 3.4142312579415504,
+      "grad_norm": 1.0802663564682007,
+      "learning_rate": 7.5509523899804264e-06,
+      "loss": 0.1208,
+      "step": 1345
+    },
+    {
+      "epoch": 3.426937738246506,
+      "grad_norm": 0.9466588497161865,
+      "learning_rate": 7.441909711039532e-06,
+      "loss": 0.1104,
+      "step": 1350
+    },
+    {
+      "epoch": 3.4396442185514613,
+      "grad_norm": 1.036575198173523,
+      "learning_rate": 7.333399755412954e-06,
+      "loss": 0.1076,
+      "step": 1355
+    },
+    {
+      "epoch": 3.4523506988564168,
+      "grad_norm": 0.9527432322502136,
+      "learning_rate": 7.225430171300306e-06,
+      "loss": 0.1209,
+      "step": 1360
+    },
+    {
+      "epoch": 3.4650571791613722,
+      "grad_norm": 0.8460400104522705,
+      "learning_rate": 7.118008568813747e-06,
+      "loss": 0.1148,
+      "step": 1365
+    },
+    {
+      "epoch": 3.4777636594663277,
+      "grad_norm": 0.8945744633674622,
+      "learning_rate": 7.011142519441541e-06,
+      "loss": 0.1136,
+      "step": 1370
+    },
+    {
+      "epoch": 3.490470139771283,
+      "grad_norm": 1.0453909635543823,
+      "learning_rate": 6.904839555514465e-06,
+      "loss": 0.1249,
+      "step": 1375
+    },
+    {
+      "epoch": 3.503176620076239,
+      "grad_norm": 0.9870423674583435,
+      "learning_rate": 6.799107169674857e-06,
+      "loss": 0.1148,
+      "step": 1380
+    },
+    {
+      "epoch": 3.5158831003811946,
+      "grad_norm": 0.8577983379364014,
+      "learning_rate": 6.6939528143485065e-06,
+      "loss": 0.1221,
+      "step": 1385
+    },
+    {
+      "epoch": 3.52858958068615,
+      "grad_norm": 1.0451507568359375,
+      "learning_rate": 6.589383901219383e-06,
+      "loss": 0.1361,
+      "step": 1390
+    },
+    {
+      "epoch": 3.5412960609911055,
+      "grad_norm": 1.0099536180496216,
+      "learning_rate": 6.485407800707248e-06,
+      "loss": 0.1267,
+      "step": 1395
+    },
+    {
+      "epoch": 3.554002541296061,
+      "grad_norm": 1.1504586935043335,
+      "learning_rate": 6.382031841448146e-06,
+      "loss": 0.106,
+      "step": 1400
+    },
+    {
+      "epoch": 3.5667090216010164,
+      "grad_norm": 0.887190580368042,
+      "learning_rate": 6.279263309777841e-06,
+      "loss": 0.1155,
+      "step": 1405
+    },
+    {
+      "epoch": 3.579415501905972,
+      "grad_norm": 1.0317214727401733,
+      "learning_rate": 6.177109449218266e-06,
+      "loss": 0.1109,
+      "step": 1410
+    },
+    {
+      "epoch": 3.5921219822109274,
+      "grad_norm": 0.8396003246307373,
+      "learning_rate": 6.075577459966972e-06,
+      "loss": 0.1148,
+      "step": 1415
+    },
+    {
+      "epoch": 3.604828462515883,
+      "grad_norm": 1.0522810220718384,
+      "learning_rate": 5.974674498389631e-06,
+      "loss": 0.1298,
+      "step": 1420
+    },
+    {
+      "epoch": 3.6175349428208388,
+      "grad_norm": 0.8637679219245911,
+      "learning_rate": 5.874407676515605e-06,
+      "loss": 0.1189,
+      "step": 1425
+    },
+    {
+      "epoch": 3.6302414231257942,
+      "grad_norm": 0.9704442024230957,
+      "learning_rate": 5.774784061536682e-06,
+      "loss": 0.1267,
+      "step": 1430
+    },
+    {
+      "epoch": 3.6429479034307497,
+      "grad_norm": 0.8692399859428406,
+      "learning_rate": 5.675810675308969e-06,
+      "loss": 0.1151,
+      "step": 1435
+    },
+    {
+      "epoch": 3.655654383735705,
+      "grad_norm": 1.042354702949524,
+      "learning_rate": 5.57749449385792e-06,
+      "loss": 0.1086,
+      "step": 1440
+    },
+    {
+      "epoch": 3.6683608640406606,
+      "grad_norm": 1.0230299234390259,
+      "learning_rate": 5.479842446886692e-06,
+      "loss": 0.1173,
+      "step": 1445
+    },
+    {
+      "epoch": 3.681067344345616,
+      "grad_norm": 0.8860698938369751,
+      "learning_rate": 5.38286141728766e-06,
+      "loss": 0.1182,
+      "step": 1450
+    },
+    {
+      "epoch": 3.693773824650572,
+      "grad_norm": 0.87037193775177,
+      "learning_rate": 5.286558240657333e-06,
+      "loss": 0.1159,
+      "step": 1455
+    },
+    {
+      "epoch": 3.7064803049555275,
+      "grad_norm": 0.951565146446228,
+      "learning_rate": 5.19093970481451e-06,
+      "loss": 0.1114,
+      "step": 1460
+    },
+    {
+      "epoch": 3.719186785260483,
+      "grad_norm": 1.0424383878707886,
+      "learning_rate": 5.096012549321895e-06,
+      "loss": 0.1046,
+      "step": 1465
+    },
+    {
+      "epoch": 3.7318932655654384,
+      "grad_norm": 0.9004281759262085,
+      "learning_rate": 5.001783465011018e-06,
+      "loss": 0.1136,
+      "step": 1470
+    },
+    {
+      "epoch": 3.744599745870394,
+      "grad_norm": 0.9678453803062439,
+      "learning_rate": 4.908259093510682e-06,
+      "loss": 0.1086,
+      "step": 1475
+    },
+    {
+      "epoch": 3.7573062261753494,
+      "grad_norm": 0.9389992952346802,
+      "learning_rate": 4.815446026778801e-06,
+      "loss": 0.1088,
+      "step": 1480
+    },
+    {
+      "epoch": 3.770012706480305,
+      "grad_norm": 0.9018406271934509,
+      "learning_rate": 4.723350806637805e-06,
+      "loss": 0.1154,
+      "step": 1485
+    },
+    {
+      "epoch": 3.7827191867852603,
+      "grad_norm": 0.8654550313949585,
+      "learning_rate": 4.631979924313521e-06,
+      "loss": 0.0979,
+      "step": 1490
+    },
+    {
+      "epoch": 3.795425667090216,
+      "grad_norm": 0.8646669387817383,
+      "learning_rate": 4.541339819977654e-06,
+      "loss": 0.1002,
+      "step": 1495
+    },
+    {
+      "epoch": 3.8081321473951717,
+      "grad_norm": 1.138091802597046,
+      "learning_rate": 4.451436882293871e-06,
+      "loss": 0.1067,
+      "step": 1500
+    },
+    {
+      "epoch": 3.820838627700127,
+      "grad_norm": 1.0015368461608887,
+      "learning_rate": 4.362277447967497e-06,
+      "loss": 0.1074,
+      "step": 1505
+    },
+    {
+      "epoch": 3.8335451080050826,
+      "grad_norm": 0.725064754486084,
+      "learning_rate": 4.273867801298863e-06,
+      "loss": 0.1143,
+      "step": 1510
+    },
+    {
+      "epoch": 3.846251588310038,
+      "grad_norm": 0.8730564713478088,
+      "learning_rate": 4.186214173740383e-06,
+      "loss": 0.0995,
+      "step": 1515
+    },
+    {
+      "epoch": 3.8589580686149936,
+      "grad_norm": 0.8183485865592957,
+      "learning_rate": 4.099322743457336e-06,
+      "loss": 0.1052,
+      "step": 1520
+    },
+    {
+      "epoch": 3.871664548919949,
+      "grad_norm": 0.9600716233253479,
+      "learning_rate": 4.013199634892407e-06,
+      "loss": 0.1115,
+      "step": 1525
+    },
+    {
+      "epoch": 3.884371029224905,
+      "grad_norm": 1.213407039642334,
+      "learning_rate": 3.927850918333992e-06,
+      "loss": 0.0962,
+      "step": 1530
+    },
+    {
+      "epoch": 3.8970775095298604,
+      "grad_norm": 1.1696031093597412,
+      "learning_rate": 3.843282609488361e-06,
+      "loss": 0.0981,
+      "step": 1535
+    },
+    {
+      "epoch": 3.909783989834816,
+      "grad_norm": 0.9459490180015564,
+      "learning_rate": 3.7595006690556483e-06,
+      "loss": 0.1029,
+      "step": 1540
+    },
+    {
+      "epoch": 3.9224904701397714,
+      "grad_norm": 0.86700040102005,
+      "learning_rate": 3.676511002309715e-06,
+      "loss": 0.0984,
+      "step": 1545
+    },
+    {
+      "epoch": 3.935196950444727,
+      "grad_norm": 0.9094993472099304,
+      "learning_rate": 3.5943194586819133e-06,
+      "loss": 0.0994,
+      "step": 1550
+    },
+    {
+      "epoch": 3.9479034307496823,
+      "grad_norm": 0.9048479199409485,
+      "learning_rate": 3.5129318313488008e-06,
+      "loss": 0.1054,
+      "step": 1555
+    },
+    {
+      "epoch": 3.9606099110546378,
+      "grad_norm": 0.9437171816825867,
+      "learning_rate": 3.4323538568238296e-06,
+      "loss": 0.1063,
+      "step": 1560
+    },
+    {
+      "epoch": 3.9733163913595932,
+      "grad_norm": 0.7141936421394348,
+      "learning_rate": 3.352591214553008e-06,
+      "loss": 0.1032,
+      "step": 1565
+    },
+    {
+      "epoch": 3.9860228716645487,
+      "grad_norm": 0.9617125391960144,
+      "learning_rate": 3.2736495265145613e-06,
+      "loss": 0.0987,
+      "step": 1570
+    },
+    {
+      "epoch": 3.998729351969504,
+      "grad_norm": 0.9574918746948242,
+      "learning_rate": 3.1955343568227196e-06,
+      "loss": 0.0958,
+      "step": 1575
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1970,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 2000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.1929826594649539e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

51_128_e5_3e-5/checkpoint-1576/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:245714c11904f4ba207f93b00b1888364e827c9293be5f8cd77d776439fa70ff
+size 7736

51_128_e5_3e-5/checkpoint-1576/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)

51_128_e5_3e-5/checkpoint-1970/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

51_128_e5_3e-5/checkpoint-1970/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": [
+    "down_proj",
+    "q_proj",
+    "up_proj",
+    "gate_proj",
+    "k_proj",
+    "o_proj",
+    "v_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

51_128_e5_3e-5/checkpoint-1970/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:422ab10c962b9df8abb8a4b3452a1e42bcc4fc53649a51761a1ac806df256fa8
+size 791751704

51_128_e5_3e-5/checkpoint-1970/latest

@@ -0,0 +1 @@
+global_step1970

51_128_e5_3e-5/checkpoint-1970/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c54f375c6918eee705222076f2ef25622c592fb31be12af460e512e6a67cfc6f
+size 15920

51_128_e5_3e-5/checkpoint-1970/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:24101def2b1356a7e993f7053c5d4f49aa722c0945d7ed533dd2b4564fe51f27
+size 15920

51_128_e5_3e-5/checkpoint-1970/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:14be3f146565033d945ed47cb550b2c4806160aa8dade9dfd7070c94c27085da
+size 15920