Upload folder using huggingface_hub

33_128_e5_3e-5/checkpoint-1284/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

33_128_e5_3e-5/checkpoint-1284/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": [
+    "o_proj",
+    "k_proj",
+    "gate_proj",
+    "q_proj",
+    "v_proj",
+    "down_proj",
+    "up_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

33_128_e5_3e-5/checkpoint-1284/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0cf5fecd99f7789f42103f65765f0bd6f37dcda6285423d1314643834a815c7f
+size 791751704

33_128_e5_3e-5/checkpoint-1284/latest

@@ -0,0 +1 @@
+global_step1284

33_128_e5_3e-5/checkpoint-1284/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5a828d150426a9911f7d011b9e012b90c480df063578dc53882bb6f0692edd83
+size 15920

33_128_e5_3e-5/checkpoint-1284/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f5877b6155475a8100b3bf053000da113c0068ca84ff2c5beb007b89ac5708a5
+size 15920

33_128_e5_3e-5/checkpoint-1284/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:657059677ac3b3c6720b7322827f39ba627e0e58892687db7cb3a8a40385f858
+size 15920

33_128_e5_3e-5/checkpoint-1284/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61594247c54b94014f8c0b714b176da169d7859a4a1c172f8f7a465fdc4d0a6c
+size 15920

33_128_e5_3e-5/checkpoint-1284/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b969d160bb28cf22cd404f310694bd95af1f28844ce5dc816aff81eccee9a1b9
+size 15920

33_128_e5_3e-5/checkpoint-1284/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0d5ab7eb2c90933fc72686b58290d5cb534696fff79f871011a5c54a423beab5
+size 15920

33_128_e5_3e-5/checkpoint-1284/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5eaec5c0a7df109bb729f38c90af941d695753653605bc8259d9b99d39dc1f79
+size 15920

33_128_e5_3e-5/checkpoint-1284/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f105bf30fc92ba9173715bbf3f95338f1580d1eb4aa25d921ea0fb6adf3608c7
+size 15920

33_128_e5_3e-5/checkpoint-1284/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b327dbd9009f70f8a679595228b64cc0962553bcd50a75bd18e4ad6315872886
+size 1064

33_128_e5_3e-5/checkpoint-1284/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
+  }
+}

33_128_e5_3e-5/checkpoint-1284/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
+}

33_128_e5_3e-5/checkpoint-1284/trainer_state.json

@@ -0,0 +1,1826 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 1284,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01557632398753894,
+      "grad_norm": 1.1759116649627686,
+      "learning_rate": 1.4814814814814815e-06,
+      "loss": 1.3361,
+      "step": 5
+    },
+    {
+      "epoch": 0.03115264797507788,
+      "grad_norm": 0.9162538647651672,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.3261,
+      "step": 10
+    },
+    {
+      "epoch": 0.04672897196261682,
+      "grad_norm": 0.7471997141838074,
+      "learning_rate": 5.185185185185185e-06,
+      "loss": 1.2639,
+      "step": 15
+    },
+    {
+      "epoch": 0.06230529595015576,
+      "grad_norm": 0.5297987461090088,
+      "learning_rate": 7.037037037037037e-06,
+      "loss": 1.2929,
+      "step": 20
+    },
+    {
+      "epoch": 0.0778816199376947,
+      "grad_norm": 0.6010679006576538,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 1.2993,
+      "step": 25
+    },
+    {
+      "epoch": 0.09345794392523364,
+      "grad_norm": 0.5925828218460083,
+      "learning_rate": 1.074074074074074e-05,
+      "loss": 1.286,
+      "step": 30
+    },
+    {
+      "epoch": 0.10903426791277258,
+      "grad_norm": 0.7390991449356079,
+      "learning_rate": 1.2592592592592592e-05,
+      "loss": 1.2887,
+      "step": 35
+    },
+    {
+      "epoch": 0.12461059190031153,
+      "grad_norm": 0.5752444863319397,
+      "learning_rate": 1.4444444444444444e-05,
+      "loss": 1.2614,
+      "step": 40
+    },
+    {
+      "epoch": 0.14018691588785046,
+      "grad_norm": 0.7002797722816467,
+      "learning_rate": 1.6296296296296297e-05,
+      "loss": 1.1979,
+      "step": 45
+    },
+    {
+      "epoch": 0.1557632398753894,
+      "grad_norm": 0.5052980780601501,
+      "learning_rate": 1.814814814814815e-05,
+      "loss": 1.2206,
+      "step": 50
+    },
+    {
+      "epoch": 0.17133956386292834,
+      "grad_norm": 0.45726311206817627,
+      "learning_rate": 1.9999999999999998e-05,
+      "loss": 1.2247,
+      "step": 55
+    },
+    {
+      "epoch": 0.18691588785046728,
+      "grad_norm": 0.575664222240448,
+      "learning_rate": 2.1851851851851852e-05,
+      "loss": 1.2307,
+      "step": 60
+    },
+    {
+      "epoch": 0.20249221183800623,
+      "grad_norm": 0.5114677548408508,
+      "learning_rate": 2.3703703703703703e-05,
+      "loss": 1.1682,
+      "step": 65
+    },
+    {
+      "epoch": 0.21806853582554517,
+      "grad_norm": 0.4684683084487915,
+      "learning_rate": 2.5555555555555557e-05,
+      "loss": 1.1644,
+      "step": 70
+    },
+    {
+      "epoch": 0.2336448598130841,
+      "grad_norm": 0.6372343897819519,
+      "learning_rate": 2.7407407407407408e-05,
+      "loss": 1.1121,
+      "step": 75
+    },
+    {
+      "epoch": 0.24922118380062305,
+      "grad_norm": 0.4992181360721588,
+      "learning_rate": 2.925925925925926e-05,
+      "loss": 1.1629,
+      "step": 80
+    },
+    {
+      "epoch": 0.26479750778816197,
+      "grad_norm": 0.4986189603805542,
+      "learning_rate": 2.9999713164962586e-05,
+      "loss": 1.142,
+      "step": 85
+    },
+    {
+      "epoch": 0.2803738317757009,
+      "grad_norm": 0.5351710915565491,
+      "learning_rate": 2.9997960323904657e-05,
+      "loss": 1.1477,
+      "step": 90
+    },
+    {
+      "epoch": 0.29595015576323985,
+      "grad_norm": 0.5250363945960999,
+      "learning_rate": 2.999461418057401e-05,
+      "loss": 1.0438,
+      "step": 95
+    },
+    {
+      "epoch": 0.3115264797507788,
+      "grad_norm": 0.5040462017059326,
+      "learning_rate": 2.998967509044683e-05,
+      "loss": 1.1118,
+      "step": 100
+    },
+    {
+      "epoch": 0.32710280373831774,
+      "grad_norm": 0.54499351978302,
+      "learning_rate": 2.998314357822537e-05,
+      "loss": 1.0831,
+      "step": 105
+    },
+    {
+      "epoch": 0.3426791277258567,
+      "grad_norm": 0.563107430934906,
+      "learning_rate": 2.9975020337782196e-05,
+      "loss": 1.0647,
+      "step": 110
+    },
+    {
+      "epoch": 0.3582554517133956,
+      "grad_norm": 0.5709556937217712,
+      "learning_rate": 2.9965306232086484e-05,
+      "loss": 1.0379,
+      "step": 115
+    },
+    {
+      "epoch": 0.37383177570093457,
+      "grad_norm": 0.5909741520881653,
+      "learning_rate": 2.9954002293112343e-05,
+      "loss": 1.046,
+      "step": 120
+    },
+    {
+      "epoch": 0.3894080996884735,
+      "grad_norm": 0.7227066159248352,
+      "learning_rate": 2.994110972172917e-05,
+      "loss": 1.0072,
+      "step": 125
+    },
+    {
+      "epoch": 0.40498442367601245,
+      "grad_norm": 0.7385948300361633,
+      "learning_rate": 2.99266298875741e-05,
+      "loss": 0.9954,
+      "step": 130
+    },
+    {
+      "epoch": 0.4205607476635514,
+      "grad_norm": 0.6370318531990051,
+      "learning_rate": 2.9910564328906474e-05,
+      "loss": 0.9612,
+      "step": 135
+    },
+    {
+      "epoch": 0.43613707165109034,
+      "grad_norm": 0.6569796204566956,
+      "learning_rate": 2.9892914752444444e-05,
+      "loss": 0.9786,
+      "step": 140
+    },
+    {
+      "epoch": 0.4517133956386293,
+      "grad_norm": 0.7288140058517456,
+      "learning_rate": 2.987368303318366e-05,
+      "loss": 0.9255,
+      "step": 145
+    },
+    {
+      "epoch": 0.4672897196261682,
+      "grad_norm": 0.6503431797027588,
+      "learning_rate": 2.9852871214198072e-05,
+      "loss": 1.0384,
+      "step": 150
+    },
+    {
+      "epoch": 0.48286604361370716,
+      "grad_norm": 0.6864191889762878,
+      "learning_rate": 2.9830481506422872e-05,
+      "loss": 0.955,
+      "step": 155
+    },
+    {
+      "epoch": 0.4984423676012461,
+      "grad_norm": 0.7586852312088013,
+      "learning_rate": 2.9806516288419662e-05,
+      "loss": 0.977,
+      "step": 160
+    },
+    {
+      "epoch": 0.514018691588785,
+      "grad_norm": 0.8165059685707092,
+      "learning_rate": 2.9780978106123713e-05,
+      "loss": 0.9314,
+      "step": 165
+    },
+    {
+      "epoch": 0.5295950155763239,
+      "grad_norm": 0.7308902144432068,
+      "learning_rate": 2.975386967257353e-05,
+      "loss": 0.9298,
+      "step": 170
+    },
+    {
+      "epoch": 0.5451713395638629,
+      "grad_norm": 0.7369241714477539,
+      "learning_rate": 2.9725193867622633e-05,
+      "loss": 0.8837,
+      "step": 175
+    },
+    {
+      "epoch": 0.5607476635514018,
+      "grad_norm": 0.8425676822662354,
+      "learning_rate": 2.969495373763361e-05,
+      "loss": 0.9184,
+      "step": 180
+    },
+    {
+      "epoch": 0.5763239875389408,
+      "grad_norm": 0.7827080488204956,
+      "learning_rate": 2.966315249515448e-05,
+      "loss": 0.8533,
+      "step": 185
+    },
+    {
+      "epoch": 0.5919003115264797,
+      "grad_norm": 0.8050850033760071,
+      "learning_rate": 2.9629793518577443e-05,
+      "loss": 0.9124,
+      "step": 190
+    },
+    {
+      "epoch": 0.6074766355140186,
+      "grad_norm": 0.8091520667076111,
+      "learning_rate": 2.9594880351779914e-05,
+      "loss": 0.8296,
+      "step": 195
+    },
+    {
+      "epoch": 0.6230529595015576,
+      "grad_norm": 1.0030003786087036,
+      "learning_rate": 2.9558416703748117e-05,
+      "loss": 0.9065,
+      "step": 200
+    },
+    {
+      "epoch": 0.6386292834890965,
+      "grad_norm": 0.8687117099761963,
+      "learning_rate": 2.952040644818301e-05,
+      "loss": 0.8802,
+      "step": 205
+    },
+    {
+      "epoch": 0.6542056074766355,
+      "grad_norm": 0.9049736261367798,
+      "learning_rate": 2.948085362308877e-05,
+      "loss": 0.8904,
+      "step": 210
+    },
+    {
+      "epoch": 0.6697819314641744,
+      "grad_norm": 1.2105063199996948,
+      "learning_rate": 2.943976243034385e-05,
+      "loss": 0.8275,
+      "step": 215
+    },
+    {
+      "epoch": 0.6853582554517134,
+      "grad_norm": 0.8405113816261292,
+      "learning_rate": 2.9397137235254543e-05,
+      "loss": 0.8687,
+      "step": 220
+    },
+    {
+      "epoch": 0.7009345794392523,
+      "grad_norm": 0.9490503072738647,
+      "learning_rate": 2.935298256609128e-05,
+      "loss": 0.8583,
+      "step": 225
+    },
+    {
+      "epoch": 0.7165109034267912,
+      "grad_norm": 0.877875030040741,
+      "learning_rate": 2.9307303113607557e-05,
+      "loss": 0.8134,
+      "step": 230
+    },
+    {
+      "epoch": 0.7320872274143302,
+      "grad_norm": 0.9228745698928833,
+      "learning_rate": 2.92601037305416e-05,
+      "loss": 0.8101,
+      "step": 235
+    },
+    {
+      "epoch": 0.7476635514018691,
+      "grad_norm": 0.8575282096862793,
+      "learning_rate": 2.9211389431100854e-05,
+      "loss": 0.7929,
+      "step": 240
+    },
+    {
+      "epoch": 0.7632398753894081,
+      "grad_norm": 1.0170135498046875,
+      "learning_rate": 2.91611653904293e-05,
+      "loss": 0.7931,
+      "step": 245
+    },
+    {
+      "epoch": 0.778816199376947,
+      "grad_norm": 0.9553667902946472,
+      "learning_rate": 2.9109436944057662e-05,
+      "loss": 0.8035,
+      "step": 250
+    },
+    {
+      "epoch": 0.794392523364486,
+      "grad_norm": 0.919255793094635,
+      "learning_rate": 2.9056209587336592e-05,
+      "loss": 0.7739,
+      "step": 255
+    },
+    {
+      "epoch": 0.8099688473520249,
+      "grad_norm": 0.9529440999031067,
+      "learning_rate": 2.90014889748529e-05,
+      "loss": 0.7342,
+      "step": 260
+    },
+    {
+      "epoch": 0.8255451713395638,
+      "grad_norm": 0.8980151414871216,
+      "learning_rate": 2.8945280919828796e-05,
+      "loss": 0.7176,
+      "step": 265
+    },
+    {
+      "epoch": 0.8411214953271028,
+      "grad_norm": 0.8686856627464294,
+      "learning_rate": 2.888759139350437e-05,
+      "loss": 0.7349,
+      "step": 270
+    },
+    {
+      "epoch": 0.8566978193146417,
+      "grad_norm": 0.9111968278884888,
+      "learning_rate": 2.8828426524503205e-05,
+      "loss": 0.754,
+      "step": 275
+    },
+    {
+      "epoch": 0.8722741433021807,
+      "grad_norm": 0.9784020185470581,
+      "learning_rate": 2.8767792598181326e-05,
+      "loss": 0.7536,
+      "step": 280
+    },
+    {
+      "epoch": 0.8878504672897196,
+      "grad_norm": 0.9565854072570801,
+      "learning_rate": 2.8705696055959466e-05,
+      "loss": 0.7186,
+      "step": 285
+    },
+    {
+      "epoch": 0.9034267912772586,
+      "grad_norm": 1.0711281299591064,
+      "learning_rate": 2.864214349463877e-05,
+      "loss": 0.6841,
+      "step": 290
+    },
+    {
+      "epoch": 0.9190031152647975,
+      "grad_norm": 1.0697325468063354,
+      "learning_rate": 2.8577141665699988e-05,
+      "loss": 0.7694,
+      "step": 295
+    },
+    {
+      "epoch": 0.9345794392523364,
+      "grad_norm": 0.947496771812439,
+      "learning_rate": 2.8510697474586228e-05,
+      "loss": 0.7155,
+      "step": 300
+    },
+    {
+      "epoch": 0.9501557632398754,
+      "grad_norm": 0.9437328577041626,
+      "learning_rate": 2.8442817979969366e-05,
+      "loss": 0.7146,
+      "step": 305
+    },
+    {
+      "epoch": 0.9657320872274143,
+      "grad_norm": 1.0520305633544922,
+      "learning_rate": 2.8373510393000163e-05,
+      "loss": 0.6837,
+      "step": 310
+    },
+    {
+      "epoch": 0.9813084112149533,
+      "grad_norm": 1.0140153169631958,
+      "learning_rate": 2.8302782076542207e-05,
+      "loss": 0.6774,
+      "step": 315
+    },
+    {
+      "epoch": 0.9968847352024922,
+      "grad_norm": 1.0676052570343018,
+      "learning_rate": 2.823064054438969e-05,
+      "loss": 0.6743,
+      "step": 320
+    },
+    {
+      "epoch": 1.0124610591900312,
+      "grad_norm": 1.0298436880111694,
+      "learning_rate": 2.8157093460469227e-05,
+      "loss": 0.6426,
+      "step": 325
+    },
+    {
+      "epoch": 1.02803738317757,
+      "grad_norm": 0.9810764193534851,
+      "learning_rate": 2.808214863802564e-05,
+      "loss": 0.6041,
+      "step": 330
+    },
+    {
+      "epoch": 1.043613707165109,
+      "grad_norm": 1.102921485900879,
+      "learning_rate": 2.8005814038791948e-05,
+      "loss": 0.5911,
+      "step": 335
+    },
+    {
+      "epoch": 1.0591900311526479,
+      "grad_norm": 1.093255877494812,
+      "learning_rate": 2.7928097772143555e-05,
+      "loss": 0.541,
+      "step": 340
+    },
+    {
+      "epoch": 1.074766355140187,
+      "grad_norm": 1.167300820350647,
+      "learning_rate": 2.7849008094236734e-05,
+      "loss": 0.5812,
+      "step": 345
+    },
+    {
+      "epoch": 1.0903426791277258,
+      "grad_norm": 1.170893907546997,
+      "learning_rate": 2.7768553407131555e-05,
+      "loss": 0.5476,
+      "step": 350
+    },
+    {
+      "epoch": 1.1059190031152648,
+      "grad_norm": 1.0814523696899414,
+      "learning_rate": 2.7686742257899288e-05,
+      "loss": 0.5671,
+      "step": 355
+    },
+    {
+      "epoch": 1.1214953271028036,
+      "grad_norm": 1.0484672784805298,
+      "learning_rate": 2.7603583337714412e-05,
+      "loss": 0.5707,
+      "step": 360
+    },
+    {
+      "epoch": 1.1370716510903427,
+      "grad_norm": 1.2223975658416748,
+      "learning_rate": 2.7519085480931312e-05,
+      "loss": 0.5276,
+      "step": 365
+    },
+    {
+      "epoch": 1.1526479750778815,
+      "grad_norm": 1.2355427742004395,
+      "learning_rate": 2.7433257664145757e-05,
+      "loss": 0.5492,
+      "step": 370
+    },
+    {
+      "epoch": 1.1682242990654206,
+      "grad_norm": 1.2064166069030762,
+      "learning_rate": 2.7346109005241294e-05,
+      "loss": 0.5337,
+      "step": 375
+    },
+    {
+      "epoch": 1.1838006230529594,
+      "grad_norm": 1.1383205652236938,
+      "learning_rate": 2.7257648762420586e-05,
+      "loss": 0.5487,
+      "step": 380
+    },
+    {
+      "epoch": 1.1993769470404985,
+      "grad_norm": 1.1303715705871582,
+      "learning_rate": 2.716788633322189e-05,
+      "loss": 0.589,
+      "step": 385
+    },
+    {
+      "epoch": 1.2149532710280373,
+      "grad_norm": 1.2811906337738037,
+      "learning_rate": 2.707683125352071e-05,
+      "loss": 0.5475,
+      "step": 390
+    },
+    {
+      "epoch": 1.2305295950155763,
+      "grad_norm": 1.3321250677108765,
+      "learning_rate": 2.6984493196516756e-05,
+      "loss": 0.5283,
+      "step": 395
+    },
+    {
+      "epoch": 1.2461059190031152,
+      "grad_norm": 1.132002830505371,
+      "learning_rate": 2.6890881971706312e-05,
+      "loss": 0.5733,
+      "step": 400
+    },
+    {
+      "epoch": 1.2616822429906542,
+      "grad_norm": 1.1940255165100098,
+      "learning_rate": 2.679600752384013e-05,
+      "loss": 0.578,
+      "step": 405
+    },
+    {
+      "epoch": 1.277258566978193,
+      "grad_norm": 1.219218134880066,
+      "learning_rate": 2.669987993186696e-05,
+      "loss": 0.5359,
+      "step": 410
+    },
+    {
+      "epoch": 1.2928348909657321,
+      "grad_norm": 1.1772730350494385,
+      "learning_rate": 2.6602509407862806e-05,
+      "loss": 0.561,
+      "step": 415
+    },
+    {
+      "epoch": 1.308411214953271,
+      "grad_norm": 1.1644192934036255,
+      "learning_rate": 2.6503906295946063e-05,
+      "loss": 0.5552,
+      "step": 420
+    },
+    {
+      "epoch": 1.32398753894081,
+      "grad_norm": 1.1815481185913086,
+      "learning_rate": 2.640408107117859e-05,
+      "loss": 0.4946,
+      "step": 425
+    },
+    {
+      "epoch": 1.3395638629283488,
+      "grad_norm": 1.0733100175857544,
+      "learning_rate": 2.630304433845293e-05,
+      "loss": 0.4903,
+      "step": 430
+    },
+    {
+      "epoch": 1.355140186915888,
+      "grad_norm": 1.1842522621154785,
+      "learning_rate": 2.6200806831365685e-05,
+      "loss": 0.493,
+      "step": 435
+    },
+    {
+      "epoch": 1.3707165109034267,
+      "grad_norm": 1.120402216911316,
+      "learning_rate": 2.609737941107722e-05,
+      "loss": 0.5127,
+      "step": 440
+    },
+    {
+      "epoch": 1.3862928348909658,
+      "grad_norm": 1.1652181148529053,
+      "learning_rate": 2.5992773065157863e-05,
+      "loss": 0.4745,
+      "step": 445
+    },
+    {
+      "epoch": 1.4018691588785046,
+      "grad_norm": 1.093859076499939,
+      "learning_rate": 2.5886998906420635e-05,
+      "loss": 0.4846,
+      "step": 450
+    },
+    {
+      "epoch": 1.4174454828660437,
+      "grad_norm": 1.2011666297912598,
+      "learning_rate": 2.5780068171740677e-05,
+      "loss": 0.471,
+      "step": 455
+    },
+    {
+      "epoch": 1.4330218068535825,
+      "grad_norm": 1.2274277210235596,
+      "learning_rate": 2.56719922208615e-05,
+      "loss": 0.4768,
+      "step": 460
+    },
+    {
+      "epoch": 1.4485981308411215,
+      "grad_norm": 1.198301076889038,
+      "learning_rate": 2.5562782535188213e-05,
+      "loss": 0.4921,
+      "step": 465
+    },
+    {
+      "epoch": 1.4641744548286604,
+      "grad_norm": 1.0955687761306763,
+      "learning_rate": 2.5452450716567763e-05,
+      "loss": 0.477,
+      "step": 470
+    },
+    {
+      "epoch": 1.4797507788161994,
+      "grad_norm": 1.079829216003418,
+      "learning_rate": 2.534100848605647e-05,
+      "loss": 0.4636,
+      "step": 475
+    },
+    {
+      "epoch": 1.4953271028037383,
+      "grad_norm": 1.049848198890686,
+      "learning_rate": 2.522846768267478e-05,
+      "loss": 0.4796,
+      "step": 480
+    },
+    {
+      "epoch": 1.5109034267912773,
+      "grad_norm": 1.2734771966934204,
+      "learning_rate": 2.5114840262149604e-05,
+      "loss": 0.4543,
+      "step": 485
+    },
+    {
+      "epoch": 1.5264797507788161,
+      "grad_norm": 1.1971790790557861,
+      "learning_rate": 2.500013829564419e-05,
+      "loss": 0.4807,
+      "step": 490
+    },
+    {
+      "epoch": 1.542056074766355,
+      "grad_norm": 1.1881734132766724,
+      "learning_rate": 2.4884373968475735e-05,
+      "loss": 0.4029,
+      "step": 495
+    },
+    {
+      "epoch": 1.557632398753894,
+      "grad_norm": 1.0760289430618286,
+      "learning_rate": 2.4767559578820897e-05,
+      "loss": 0.4933,
+      "step": 500
+    },
+    {
+      "epoch": 1.573208722741433,
+      "grad_norm": 1.2057150602340698,
+      "learning_rate": 2.4649707536409297e-05,
+      "loss": 0.4271,
+      "step": 505
+    },
+    {
+      "epoch": 1.588785046728972,
+      "grad_norm": 1.0976535081863403,
+      "learning_rate": 2.4530830361205184e-05,
+      "loss": 0.4645,
+      "step": 510
+    },
+    {
+      "epoch": 1.6043613707165107,
+      "grad_norm": 1.481553077697754,
+      "learning_rate": 2.4410940682077377e-05,
+      "loss": 0.4233,
+      "step": 515
+    },
+    {
+      "epoch": 1.6199376947040498,
+      "grad_norm": 1.1634652614593506,
+      "learning_rate": 2.429005123545763e-05,
+      "loss": 0.4394,
+      "step": 520
+    },
+    {
+      "epoch": 1.6355140186915889,
+      "grad_norm": 1.1656765937805176,
+      "learning_rate": 2.4168174863987605e-05,
+      "loss": 0.4434,
+      "step": 525
+    },
+    {
+      "epoch": 1.6510903426791277,
+      "grad_norm": 1.0849956274032593,
+      "learning_rate": 2.404532451515453e-05,
+      "loss": 0.4255,
+      "step": 530
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 1.0943026542663574,
+      "learning_rate": 2.392151323991571e-05,
+      "loss": 0.4099,
+      "step": 535
+    },
+    {
+      "epoch": 1.6822429906542056,
+      "grad_norm": 1.2393852472305298,
+      "learning_rate": 2.3796754191312094e-05,
+      "loss": 0.3917,
+      "step": 540
+    },
+    {
+      "epoch": 1.6978193146417446,
+      "grad_norm": 1.2227272987365723,
+      "learning_rate": 2.367106062307095e-05,
+      "loss": 0.4232,
+      "step": 545
+    },
+    {
+      "epoch": 1.7133956386292835,
+      "grad_norm": 1.3530696630477905,
+      "learning_rate": 2.3544445888197855e-05,
+      "loss": 0.3996,
+      "step": 550
+    },
+    {
+      "epoch": 1.7289719626168223,
+      "grad_norm": 1.1738826036453247,
+      "learning_rate": 2.3416923437558175e-05,
+      "loss": 0.4333,
+      "step": 555
+    },
+    {
+      "epoch": 1.7445482866043613,
+      "grad_norm": 1.2589629888534546,
+      "learning_rate": 2.3288506818448057e-05,
+      "loss": 0.4021,
+      "step": 560
+    },
+    {
+      "epoch": 1.7601246105919004,
+      "grad_norm": 1.1301474571228027,
+      "learning_rate": 2.3159209673155306e-05,
+      "loss": 0.4056,
+      "step": 565
+    },
+    {
+      "epoch": 1.7757009345794392,
+      "grad_norm": 1.2090823650360107,
+      "learning_rate": 2.302904573751006e-05,
+      "loss": 0.4173,
+      "step": 570
+    },
+    {
+      "epoch": 1.791277258566978,
+      "grad_norm": 1.1565548181533813,
+      "learning_rate": 2.2898028839425592e-05,
+      "loss": 0.3904,
+      "step": 575
+    },
+    {
+      "epoch": 1.8068535825545171,
+      "grad_norm": 1.1722818613052368,
+      "learning_rate": 2.2766172897429272e-05,
+      "loss": 0.4146,
+      "step": 580
+    },
+    {
+      "epoch": 1.8224299065420562,
+      "grad_norm": 1.0461018085479736,
+      "learning_rate": 2.2633491919183988e-05,
+      "loss": 0.4062,
+      "step": 585
+    },
+    {
+      "epoch": 1.838006230529595,
+      "grad_norm": 1.0770604610443115,
+      "learning_rate": 2.25e-05,
+      "loss": 0.4073,
+      "step": 590
+    },
+    {
+      "epoch": 1.8535825545171338,
+      "grad_norm": 1.1962300539016724,
+      "learning_rate": 2.2365711321337573e-05,
+      "loss": 0.3932,
+      "step": 595
+    },
+    {
+      "epoch": 1.8691588785046729,
+      "grad_norm": 1.264811396598816,
+      "learning_rate": 2.2230640149300375e-05,
+      "loss": 0.3782,
+      "step": 600
+    },
+    {
+      "epoch": 1.884735202492212,
+      "grad_norm": 1.1047016382217407,
+      "learning_rate": 2.2094800833119954e-05,
+      "loss": 0.3725,
+      "step": 605
+    },
+    {
+      "epoch": 1.9003115264797508,
+      "grad_norm": 1.2060890197753906,
+      "learning_rate": 2.1958207803631343e-05,
+      "loss": 0.3467,
+      "step": 610
+    },
+    {
+      "epoch": 1.9158878504672896,
+      "grad_norm": 1.3942707777023315,
+      "learning_rate": 2.1820875571739987e-05,
+      "loss": 0.3567,
+      "step": 615
+    },
+    {
+      "epoch": 1.9314641744548287,
+      "grad_norm": 1.4150056838989258,
+      "learning_rate": 2.1682818726880227e-05,
+      "loss": 0.3699,
+      "step": 620
+    },
+    {
+      "epoch": 1.9470404984423677,
+      "grad_norm": 1.1999322175979614,
+      "learning_rate": 2.154405193546535e-05,
+      "loss": 0.37,
+      "step": 625
+    },
+    {
+      "epoch": 1.9626168224299065,
+      "grad_norm": 1.3075921535491943,
+      "learning_rate": 2.1404589939329538e-05,
+      "loss": 0.3518,
+      "step": 630
+    },
+    {
+      "epoch": 1.9781931464174454,
+      "grad_norm": 1.3312402963638306,
+      "learning_rate": 2.1264447554161766e-05,
+      "loss": 0.3646,
+      "step": 635
+    },
+    {
+      "epoch": 1.9937694704049844,
+      "grad_norm": 1.1783993244171143,
+      "learning_rate": 2.1123639667931854e-05,
+      "loss": 0.3462,
+      "step": 640
+    },
+    {
+      "epoch": 2.0093457943925235,
+      "grad_norm": 1.1549129486083984,
+      "learning_rate": 2.0982181239308887e-05,
+      "loss": 0.2886,
+      "step": 645
+    },
+    {
+      "epoch": 2.0249221183800623,
+      "grad_norm": 1.127125859260559,
+      "learning_rate": 2.0840087296072017e-05,
+      "loss": 0.3207,
+      "step": 650
+    },
+    {
+      "epoch": 2.040498442367601,
+      "grad_norm": 1.2090187072753906,
+      "learning_rate": 2.0697372933514076e-05,
+      "loss": 0.284,
+      "step": 655
+    },
+    {
+      "epoch": 2.05607476635514,
+      "grad_norm": 1.1644618511199951,
+      "learning_rate": 2.0554053312837865e-05,
+      "loss": 0.2961,
+      "step": 660
+    },
+    {
+      "epoch": 2.0716510903426792,
+      "grad_norm": 1.1487720012664795,
+      "learning_rate": 2.0410143659545546e-05,
+      "loss": 0.2789,
+      "step": 665
+    },
+    {
+      "epoch": 2.087227414330218,
+      "grad_norm": 1.2465693950653076,
+      "learning_rate": 2.0265659261821163e-05,
+      "loss": 0.2802,
+      "step": 670
+    },
+    {
+      "epoch": 2.102803738317757,
+      "grad_norm": 1.16876220703125,
+      "learning_rate": 2.0120615468906502e-05,
+      "loss": 0.2761,
+      "step": 675
+    },
+    {
+      "epoch": 2.1183800623052957,
+      "grad_norm": 1.255817174911499,
+      "learning_rate": 1.9975027689470462e-05,
+      "loss": 0.276,
+      "step": 680
+    },
+    {
+      "epoch": 2.133956386292835,
+      "grad_norm": 1.2631534337997437,
+      "learning_rate": 1.9828911389972142e-05,
+      "loss": 0.2738,
+      "step": 685
+    },
+    {
+      "epoch": 2.149532710280374,
+      "grad_norm": 1.2952145338058472,
+      "learning_rate": 1.968228209301775e-05,
+      "loss": 0.2859,
+      "step": 690
+    },
+    {
+      "epoch": 2.1651090342679127,
+      "grad_norm": 1.2698311805725098,
+      "learning_rate": 1.9535155375711576e-05,
+      "loss": 0.2457,
+      "step": 695
+    },
+    {
+      "epoch": 2.1806853582554515,
+      "grad_norm": 1.2509089708328247,
+      "learning_rate": 1.9387546868001166e-05,
+      "loss": 0.3084,
+      "step": 700
+    },
+    {
+      "epoch": 2.196261682242991,
+      "grad_norm": 1.184861183166504,
+      "learning_rate": 1.9239472251016856e-05,
+      "loss": 0.2587,
+      "step": 705
+    },
+    {
+      "epoch": 2.2118380062305296,
+      "grad_norm": 1.4472482204437256,
+      "learning_rate": 1.9090947255405935e-05,
+      "loss": 0.2836,
+      "step": 710
+    },
+    {
+      "epoch": 2.2274143302180685,
+      "grad_norm": 1.2142419815063477,
+      "learning_rate": 1.8941987659661474e-05,
+      "loss": 0.2719,
+      "step": 715
+    },
+    {
+      "epoch": 2.2429906542056073,
+      "grad_norm": 1.0755081176757812,
+      "learning_rate": 1.879260928844611e-05,
+      "loss": 0.2573,
+      "step": 720
+    },
+    {
+      "epoch": 2.2585669781931466,
+      "grad_norm": 1.1565829515457153,
+      "learning_rate": 1.8642828010910937e-05,
+      "loss": 0.2675,
+      "step": 725
+    },
+    {
+      "epoch": 2.2741433021806854,
+      "grad_norm": 1.233801007270813,
+      "learning_rate": 1.849265973900962e-05,
+      "loss": 0.2826,
+      "step": 730
+    },
+    {
+      "epoch": 2.289719626168224,
+      "grad_norm": 1.4652220010757446,
+      "learning_rate": 1.8342120425808036e-05,
+      "loss": 0.2218,
+      "step": 735
+    },
+    {
+      "epoch": 2.305295950155763,
+      "grad_norm": 1.1132934093475342,
+      "learning_rate": 1.819122606378947e-05,
+      "loss": 0.2359,
+      "step": 740
+    },
+    {
+      "epoch": 2.3208722741433023,
+      "grad_norm": 1.2268480062484741,
+      "learning_rate": 1.803999268315568e-05,
+      "loss": 0.2319,
+      "step": 745
+    },
+    {
+      "epoch": 2.336448598130841,
+      "grad_norm": 1.1891158819198608,
+      "learning_rate": 1.788843635012393e-05,
+      "loss": 0.244,
+      "step": 750
+    },
+    {
+      "epoch": 2.35202492211838,
+      "grad_norm": 1.0149863958358765,
+      "learning_rate": 1.77365731652202e-05,
+      "loss": 0.2484,
+      "step": 755
+    },
+    {
+      "epoch": 2.367601246105919,
+      "grad_norm": 1.2054303884506226,
+      "learning_rate": 1.758441926156875e-05,
+      "loss": 0.2764,
+      "step": 760
+    },
+    {
+      "epoch": 2.383177570093458,
+      "grad_norm": 1.3622403144836426,
+      "learning_rate": 1.7431990803178248e-05,
+      "loss": 0.2449,
+      "step": 765
+    },
+    {
+      "epoch": 2.398753894080997,
+      "grad_norm": 1.3636577129364014,
+      "learning_rate": 1.727930398322456e-05,
+      "loss": 0.2437,
+      "step": 770
+    },
+    {
+      "epoch": 2.4143302180685358,
+      "grad_norm": 1.1471670866012573,
+      "learning_rate": 1.7126375022330483e-05,
+      "loss": 0.2521,
+      "step": 775
+    },
+    {
+      "epoch": 2.4299065420560746,
+      "grad_norm": 1.2446048259735107,
+      "learning_rate": 1.697322016684257e-05,
+      "loss": 0.2313,
+      "step": 780
+    },
+    {
+      "epoch": 2.445482866043614,
+      "grad_norm": 1.3057276010513306,
+      "learning_rate": 1.6819855687105197e-05,
+      "loss": 0.2643,
+      "step": 785
+    },
+    {
+      "epoch": 2.4610591900311527,
+      "grad_norm": 1.2132842540740967,
+      "learning_rate": 1.6666297875732072e-05,
+      "loss": 0.2481,
+      "step": 790
+    },
+    {
+      "epoch": 2.4766355140186915,
+      "grad_norm": 1.6572365760803223,
+      "learning_rate": 1.6512563045875424e-05,
+      "loss": 0.2505,
+      "step": 795
+    },
+    {
+      "epoch": 2.4922118380062304,
+      "grad_norm": 1.0174639225006104,
+      "learning_rate": 1.6358667529492977e-05,
+      "loss": 0.2086,
+      "step": 800
+    },
+    {
+      "epoch": 2.507788161993769,
+      "grad_norm": 1.2086824178695679,
+      "learning_rate": 1.6204627675612904e-05,
+      "loss": 0.2166,
+      "step": 805
+    },
+    {
+      "epoch": 2.5233644859813085,
+      "grad_norm": 1.0739854574203491,
+      "learning_rate": 1.6050459848597023e-05,
+      "loss": 0.211,
+      "step": 810
+    },
+    {
+      "epoch": 2.5389408099688473,
+      "grad_norm": 1.2083333730697632,
+      "learning_rate": 1.589618042640233e-05,
+      "loss": 0.2279,
+      "step": 815
+    },
+    {
+      "epoch": 2.554517133956386,
+      "grad_norm": 1.1843072175979614,
+      "learning_rate": 1.574180579884109e-05,
+      "loss": 0.2055,
+      "step": 820
+    },
+    {
+      "epoch": 2.5700934579439254,
+      "grad_norm": 1.1035974025726318,
+      "learning_rate": 1.558735236583966e-05,
+      "loss": 0.2195,
+      "step": 825
+    },
+    {
+      "epoch": 2.5856697819314642,
+      "grad_norm": 1.1672109365463257,
+      "learning_rate": 1.5432836535696275e-05,
+      "loss": 0.2191,
+      "step": 830
+    },
+    {
+      "epoch": 2.601246105919003,
+      "grad_norm": 1.1042742729187012,
+      "learning_rate": 1.5278274723337895e-05,
+      "loss": 0.216,
+      "step": 835
+    },
+    {
+      "epoch": 2.616822429906542,
+      "grad_norm": 1.2599068880081177,
+      "learning_rate": 1.5123683348576398e-05,
+      "loss": 0.2432,
+      "step": 840
+    },
+    {
+      "epoch": 2.6323987538940807,
+      "grad_norm": 1.0473904609680176,
+      "learning_rate": 1.496907883436419e-05,
+      "loss": 0.2048,
+      "step": 845
+    },
+    {
+      "epoch": 2.64797507788162,
+      "grad_norm": 1.183077096939087,
+      "learning_rate": 1.4814477605049557e-05,
+      "loss": 0.2252,
+      "step": 850
+    },
+    {
+      "epoch": 2.663551401869159,
+      "grad_norm": 1.1935157775878906,
+      "learning_rate": 1.465989608463181e-05,
+      "loss": 0.221,
+      "step": 855
+    },
+    {
+      "epoch": 2.6791277258566977,
+      "grad_norm": 1.2083299160003662,
+      "learning_rate": 1.450535069501649e-05,
+      "loss": 0.2047,
+      "step": 860
+    },
+    {
+      "epoch": 2.694704049844237,
+      "grad_norm": 1.1584832668304443,
+      "learning_rate": 1.4350857854270807e-05,
+      "loss": 0.2155,
+      "step": 865
+    },
+    {
+      "epoch": 2.710280373831776,
+      "grad_norm": 1.0836254358291626,
+      "learning_rate": 1.4196433974879452e-05,
+      "loss": 0.1927,
+      "step": 870
+    },
+    {
+      "epoch": 2.7258566978193146,
+      "grad_norm": 1.1828125715255737,
+      "learning_rate": 1.404209546200104e-05,
+      "loss": 0.2193,
+      "step": 875
+    },
+    {
+      "epoch": 2.7414330218068534,
+      "grad_norm": 1.276404619216919,
+      "learning_rate": 1.388785871172531e-05,
+      "loss": 0.2185,
+      "step": 880
+    },
+    {
+      "epoch": 2.7570093457943923,
+      "grad_norm": 1.2181018590927124,
+      "learning_rate": 1.3733740109331282e-05,
+      "loss": 0.2102,
+      "step": 885
+    },
+    {
+      "epoch": 2.7725856697819315,
+      "grad_norm": 1.1601808071136475,
+      "learning_rate": 1.3579756027546597e-05,
+      "loss": 0.2082,
+      "step": 890
+    },
+    {
+      "epoch": 2.7881619937694704,
+      "grad_norm": 1.2979238033294678,
+      "learning_rate": 1.3425922824808146e-05,
+      "loss": 0.2099,
+      "step": 895
+    },
+    {
+      "epoch": 2.803738317757009,
+      "grad_norm": 1.068938970565796,
+      "learning_rate": 1.327225684352425e-05,
+      "loss": 0.2054,
+      "step": 900
+    },
+    {
+      "epoch": 2.8193146417445485,
+      "grad_norm": 1.171219825744629,
+      "learning_rate": 1.3118774408338517e-05,
+      "loss": 0.2168,
+      "step": 905
+    },
+    {
+      "epoch": 2.8348909657320873,
+      "grad_norm": 1.1562765836715698,
+      "learning_rate": 1.2965491824395614e-05,
+      "loss": 0.2266,
+      "step": 910
+    },
+    {
+      "epoch": 2.850467289719626,
+      "grad_norm": 1.0936180353164673,
+      "learning_rate": 1.2812425375609101e-05,
+      "loss": 0.1971,
+      "step": 915
+    },
+    {
+      "epoch": 2.866043613707165,
+      "grad_norm": 1.166494607925415,
+      "learning_rate": 1.2659591322931491e-05,
+      "loss": 0.1849,
+      "step": 920
+    },
+    {
+      "epoch": 2.881619937694704,
+      "grad_norm": 1.102161169052124,
+      "learning_rate": 1.2507005902626797e-05,
+      "loss": 0.1951,
+      "step": 925
+    },
+    {
+      "epoch": 2.897196261682243,
+      "grad_norm": 1.0930756330490112,
+      "learning_rate": 1.2354685324545676e-05,
+      "loss": 0.2151,
+      "step": 930
+    },
+    {
+      "epoch": 2.912772585669782,
+      "grad_norm": 1.1219929456710815,
+      "learning_rate": 1.2202645770403376e-05,
+      "loss": 0.1761,
+      "step": 935
+    },
+    {
+      "epoch": 2.9283489096573208,
+      "grad_norm": 1.1747044324874878,
+      "learning_rate": 1.2050903392060674e-05,
+      "loss": 0.1746,
+      "step": 940
+    },
+    {
+      "epoch": 2.94392523364486,
+      "grad_norm": 1.1695518493652344,
+      "learning_rate": 1.1899474309807998e-05,
+      "loss": 0.1868,
+      "step": 945
+    },
+    {
+      "epoch": 2.959501557632399,
+      "grad_norm": 1.3246632814407349,
+      "learning_rate": 1.1748374610652906e-05,
+      "loss": 0.1942,
+      "step": 950
+    },
+    {
+      "epoch": 2.9750778816199377,
+      "grad_norm": 1.2505390644073486,
+      "learning_rate": 1.159762034661105e-05,
+      "loss": 0.2147,
+      "step": 955
+    },
+    {
+      "epoch": 2.9906542056074765,
+      "grad_norm": 1.1202441453933716,
+      "learning_rate": 1.1447227533000939e-05,
+      "loss": 0.1741,
+      "step": 960
+    },
+    {
+      "epoch": 3.0062305295950154,
+      "grad_norm": 1.0747812986373901,
+      "learning_rate": 1.1297212146742543e-05,
+      "loss": 0.1704,
+      "step": 965
+    },
+    {
+      "epoch": 3.0218068535825546,
+      "grad_norm": 1.2752288579940796,
+      "learning_rate": 1.1147590124659983e-05,
+      "loss": 0.1438,
+      "step": 970
+    },
+    {
+      "epoch": 3.0373831775700935,
+      "grad_norm": 1.0363479852676392,
+      "learning_rate": 1.0998377361788489e-05,
+      "loss": 0.1591,
+      "step": 975
+    },
+    {
+      "epoch": 3.0529595015576323,
+      "grad_norm": 1.2088156938552856,
+      "learning_rate": 1.0849589709685816e-05,
+      "loss": 0.1496,
+      "step": 980
+    },
+    {
+      "epoch": 3.068535825545171,
+      "grad_norm": 1.0746065378189087,
+      "learning_rate": 1.0701242974748245e-05,
+      "loss": 0.1276,
+      "step": 985
+    },
+    {
+      "epoch": 3.0841121495327104,
+      "grad_norm": 1.1134722232818604,
+      "learning_rate": 1.0553352916531389e-05,
+      "loss": 0.1523,
+      "step": 990
+    },
+    {
+      "epoch": 3.0996884735202492,
+      "grad_norm": 1.002192735671997,
+      "learning_rate": 1.0405935246076011e-05,
+      "loss": 0.1379,
+      "step": 995
+    },
+    {
+      "epoch": 3.115264797507788,
+      "grad_norm": 1.0264095067977905,
+      "learning_rate": 1.0259005624238931e-05,
+      "loss": 0.1469,
+      "step": 1000
+    },
+    {
+      "epoch": 3.130841121495327,
+      "grad_norm": 1.0532501935958862,
+      "learning_rate": 1.0112579660029341e-05,
+      "loss": 0.1294,
+      "step": 1005
+    },
+    {
+      "epoch": 3.146417445482866,
+      "grad_norm": 1.0389549732208252,
+      "learning_rate": 9.966672908950552e-06,
+      "loss": 0.1246,
+      "step": 1010
+    },
+    {
+      "epoch": 3.161993769470405,
+      "grad_norm": 1.1664501428604126,
+      "learning_rate": 9.82130087134748e-06,
+      "loss": 0.136,
+      "step": 1015
+    },
+    {
+      "epoch": 3.177570093457944,
+      "grad_norm": 1.0726090669631958,
+      "learning_rate": 9.676478990759979e-06,
+      "loss": 0.1422,
+      "step": 1020
+    },
+    {
+      "epoch": 3.1931464174454827,
+      "grad_norm": 1.0180519819259644,
+      "learning_rate": 9.532222652282187e-06,
+      "loss": 0.1381,
+      "step": 1025
+    },
+    {
+      "epoch": 3.208722741433022,
+      "grad_norm": 1.2948464155197144,
+      "learning_rate": 9.388547180928103e-06,
+      "loss": 0.1211,
+      "step": 1030
+    },
+    {
+      "epoch": 3.2242990654205608,
+      "grad_norm": 1.1216830015182495,
+      "learning_rate": 9.245467840003544e-06,
+      "loss": 0.1428,
+      "step": 1035
+    },
+    {
+      "epoch": 3.2398753894080996,
+      "grad_norm": 1.1088138818740845,
+      "learning_rate": 9.102999829484652e-06,
+      "loss": 0.1389,
+      "step": 1040
+    },
+    {
+      "epoch": 3.2554517133956384,
+      "grad_norm": 1.2717032432556152,
+      "learning_rate": 8.961158284403123e-06,
+      "loss": 0.1304,
+      "step": 1045
+    },
+    {
+      "epoch": 3.2710280373831777,
+      "grad_norm": 0.9676588177680969,
+      "learning_rate": 8.819958273238358e-06,
+      "loss": 0.1289,
+      "step": 1050
+    },
+    {
+      "epoch": 3.2866043613707165,
+      "grad_norm": 1.1699531078338623,
+      "learning_rate": 8.679414796316658e-06,
+      "loss": 0.1347,
+      "step": 1055
+    },
+    {
+      "epoch": 3.3021806853582554,
+      "grad_norm": 1.1606409549713135,
+      "learning_rate": 8.539542784217675e-06,
+      "loss": 0.134,
+      "step": 1060
+    },
+    {
+      "epoch": 3.317757009345794,
+      "grad_norm": 1.0339468717575073,
+      "learning_rate": 8.400357096188256e-06,
+      "loss": 0.14,
+      "step": 1065
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 1.2066926956176758,
+      "learning_rate": 8.26187251856389e-06,
+      "loss": 0.1361,
+      "step": 1070
+    },
+    {
+      "epoch": 3.3489096573208723,
+      "grad_norm": 0.9468661546707153,
+      "learning_rate": 8.124103763197911e-06,
+      "loss": 0.1384,
+      "step": 1075
+    },
+    {
+      "epoch": 3.364485981308411,
+      "grad_norm": 1.0264408588409424,
+      "learning_rate": 7.987065465898522e-06,
+      "loss": 0.1282,
+      "step": 1080
+    },
+    {
+      "epoch": 3.38006230529595,
+      "grad_norm": 0.9723944664001465,
+      "learning_rate": 7.85077218487403e-06,
+      "loss": 0.1298,
+      "step": 1085
+    },
+    {
+      "epoch": 3.3956386292834893,
+      "grad_norm": 1.0408103466033936,
+      "learning_rate": 7.715238399186246e-06,
+      "loss": 0.1321,
+      "step": 1090
+    },
+    {
+      "epoch": 3.411214953271028,
+      "grad_norm": 1.1424450874328613,
+      "learning_rate": 7.580478507212283e-06,
+      "loss": 0.1215,
+      "step": 1095
+    },
+    {
+      "epoch": 3.426791277258567,
+      "grad_norm": 1.001061201095581,
+      "learning_rate": 7.446506825114969e-06,
+      "loss": 0.1211,
+      "step": 1100
+    },
+    {
+      "epoch": 3.4423676012461057,
+      "grad_norm": 1.1220428943634033,
+      "learning_rate": 7.313337585321978e-06,
+      "loss": 0.1158,
+      "step": 1105
+    },
+    {
+      "epoch": 3.457943925233645,
+      "grad_norm": 1.1033799648284912,
+      "learning_rate": 7.180984935013862e-06,
+      "loss": 0.1442,
+      "step": 1110
+    },
+    {
+      "epoch": 3.473520249221184,
+      "grad_norm": 1.141599416732788,
+      "learning_rate": 7.0494629346210924e-06,
+      "loss": 0.1363,
+      "step": 1115
+    },
+    {
+      "epoch": 3.4890965732087227,
+      "grad_norm": 0.9771517515182495,
+      "learning_rate": 6.918785556330396e-06,
+      "loss": 0.127,
+      "step": 1120
+    },
+    {
+      "epoch": 3.5046728971962615,
+      "grad_norm": 0.999470591545105,
+      "learning_rate": 6.788966682600412e-06,
+      "loss": 0.1086,
+      "step": 1125
+    },
+    {
+      "epoch": 3.520249221183801,
+      "grad_norm": 1.0332609415054321,
+      "learning_rate": 6.6600201046868944e-06,
+      "loss": 0.1187,
+      "step": 1130
+    },
+    {
+      "epoch": 3.5358255451713396,
+      "grad_norm": 1.0396648645401,
+      "learning_rate": 6.5319595211775876e-06,
+      "loss": 0.1383,
+      "step": 1135
+    },
+    {
+      "epoch": 3.5514018691588785,
+      "grad_norm": 0.9917812347412109,
+      "learning_rate": 6.404798536536991e-06,
+      "loss": 0.1187,
+      "step": 1140
+    },
+    {
+      "epoch": 3.5669781931464173,
+      "grad_norm": 0.9046400785446167,
+      "learning_rate": 6.278550659661091e-06,
+      "loss": 0.1222,
+      "step": 1145
+    },
+    {
+      "epoch": 3.582554517133956,
+      "grad_norm": 0.8576000332832336,
+      "learning_rate": 6.153229302442209e-06,
+      "loss": 0.1227,
+      "step": 1150
+    },
+    {
+      "epoch": 3.5981308411214954,
+      "grad_norm": 0.9951794147491455,
+      "learning_rate": 6.028847778344249e-06,
+      "loss": 0.1283,
+      "step": 1155
+    },
+    {
+      "epoch": 3.6137071651090342,
+      "grad_norm": 1.1326957941055298,
+      "learning_rate": 5.905419300988311e-06,
+      "loss": 0.1162,
+      "step": 1160
+    },
+    {
+      "epoch": 3.629283489096573,
+      "grad_norm": 0.9401041865348816,
+      "learning_rate": 5.782956982748966e-06,
+      "loss": 0.1177,
+      "step": 1165
+    },
+    {
+      "epoch": 3.6448598130841123,
+      "grad_norm": 1.0154671669006348,
+      "learning_rate": 5.661473833361247e-06,
+      "loss": 0.1236,
+      "step": 1170
+    },
+    {
+      "epoch": 3.660436137071651,
+      "grad_norm": 0.9394180178642273,
+      "learning_rate": 5.540982758538586e-06,
+      "loss": 0.1123,
+      "step": 1175
+    },
+    {
+      "epoch": 3.67601246105919,
+      "grad_norm": 0.9814959764480591,
+      "learning_rate": 5.4214965586017814e-06,
+      "loss": 0.122,
+      "step": 1180
+    },
+    {
+      "epoch": 3.691588785046729,
+      "grad_norm": 0.935127317905426,
+      "learning_rate": 5.30302792711915e-06,
+      "loss": 0.1144,
+      "step": 1185
+    },
+    {
+      "epoch": 3.7071651090342677,
+      "grad_norm": 1.111989140510559,
+      "learning_rate": 5.185589449558016e-06,
+      "loss": 0.1026,
+      "step": 1190
+    },
+    {
+      "epoch": 3.722741433021807,
+      "grad_norm": 0.7687360048294067,
+      "learning_rate": 5.069193601947736e-06,
+      "loss": 0.1044,
+      "step": 1195
+    },
+    {
+      "epoch": 3.7383177570093458,
+      "grad_norm": 1.0546218156814575,
+      "learning_rate": 4.953852749554294e-06,
+      "loss": 0.1184,
+      "step": 1200
+    },
+    {
+      "epoch": 3.7538940809968846,
+      "grad_norm": 1.0001137256622314,
+      "learning_rate": 4.839579145566665e-06,
+      "loss": 0.1181,
+      "step": 1205
+    },
+    {
+      "epoch": 3.769470404984424,
+      "grad_norm": 0.9862017035484314,
+      "learning_rate": 4.726384929795133e-06,
+      "loss": 0.1225,
+      "step": 1210
+    },
+    {
+      "epoch": 3.7850467289719627,
+      "grad_norm": 1.0694918632507324,
+      "learning_rate": 4.614282127381613e-06,
+      "loss": 0.1059,
+      "step": 1215
+    },
+    {
+      "epoch": 3.8006230529595015,
+      "grad_norm": 0.9774491190910339,
+      "learning_rate": 4.503282647522161e-06,
+      "loss": 0.116,
+      "step": 1220
+    },
+    {
+      "epoch": 3.8161993769470404,
+      "grad_norm": 1.0102533102035522,
+      "learning_rate": 4.393398282201788e-06,
+      "loss": 0.1106,
+      "step": 1225
+    },
+    {
+      "epoch": 3.831775700934579,
+      "grad_norm": 1.047069787979126,
+      "learning_rate": 4.284640704941781e-06,
+      "loss": 0.1162,
+      "step": 1230
+    },
+    {
+      "epoch": 3.8473520249221185,
+      "grad_norm": 0.905773401260376,
+      "learning_rate": 4.177021469559548e-06,
+      "loss": 0.0953,
+      "step": 1235
+    },
+    {
+      "epoch": 3.8629283489096573,
+      "grad_norm": 0.944668710231781,
+      "learning_rate": 4.070552008941193e-06,
+      "loss": 0.1059,
+      "step": 1240
+    },
+    {
+      "epoch": 3.878504672897196,
+      "grad_norm": 0.9875009059906006,
+      "learning_rate": 3.965243633826974e-06,
+      "loss": 0.1088,
+      "step": 1245
+    },
+    {
+      "epoch": 3.8940809968847354,
+      "grad_norm": 0.8600956201553345,
+      "learning_rate": 3.861107531609701e-06,
+      "loss": 0.1009,
+      "step": 1250
+    },
+    {
+      "epoch": 3.9096573208722742,
+      "grad_norm": 0.9825694561004639,
+      "learning_rate": 3.758154765146241e-06,
+      "loss": 0.1221,
+      "step": 1255
+    },
+    {
+      "epoch": 3.925233644859813,
+      "grad_norm": 0.885953962802887,
+      "learning_rate": 3.6563962715822534e-06,
+      "loss": 0.1149,
+      "step": 1260
+    },
+    {
+      "epoch": 3.940809968847352,
+      "grad_norm": 0.919476330280304,
+      "learning_rate": 3.5558428611903087e-06,
+      "loss": 0.1125,
+      "step": 1265
+    },
+    {
+      "epoch": 3.9563862928348907,
+      "grad_norm": 1.0793930292129517,
+      "learning_rate": 3.4565052162214573e-06,
+      "loss": 0.0992,
+      "step": 1270
+    },
+    {
+      "epoch": 3.97196261682243,
+      "grad_norm": 1.032047986984253,
+      "learning_rate": 3.3583938897703787e-06,
+      "loss": 0.1138,
+      "step": 1275
+    },
+    {
+      "epoch": 3.987538940809969,
+      "grad_norm": 0.9010823965072632,
+      "learning_rate": 3.2615193046543175e-06,
+      "loss": 0.1011,
+      "step": 1280
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1605,
+  "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": 1.8396761894850396e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

33_128_e5_3e-5/checkpoint-1284/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:da61cbe608501d2834de43cd31aa18db3ae688ca81d76fe4c7a0a77149e96662
+size 7736

33_128_e5_3e-5/checkpoint-1284/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)

33_128_e5_3e-5/checkpoint-1605/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

33_128_e5_3e-5/checkpoint-1605/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": [
+    "o_proj",
+    "k_proj",
+    "gate_proj",
+    "q_proj",
+    "v_proj",
+    "down_proj",
+    "up_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

33_128_e5_3e-5/checkpoint-1605/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fa3e844aef137805b660d89555aad0771ba67bceed7a9697539ddd382e5fc951
+size 791751704

33_128_e5_3e-5/checkpoint-1605/latest

@@ -0,0 +1 @@
+global_step1605

33_128_e5_3e-5/checkpoint-1605/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7b1801382cbef583da46ed5196409dac846314af86568a755a320dc1c9ac23fa
+size 15920

33_128_e5_3e-5/checkpoint-1605/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e6c45354e926f72a3a4bc63d8d0c1315d46318ff2f416bf487e65fe5e91ac89
+size 15920

33_128_e5_3e-5/checkpoint-1605/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:49e106b1cec1725e67319fda30e1ea315e3942c4a5d2c5dd05f7d6f6fe426166
+size 15920

33_128_e5_3e-5/checkpoint-1605/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:16ea1c829b7599c7a4bcb886961c79c0c08fddeccddabdd33421850b53d925c7
+size 15920

33_128_e5_3e-5/checkpoint-1605/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5adec85aef18707e776aac2b97eccf6d5a506399d9df9aeb31d1381966142e94
+size 15920

33_128_e5_3e-5/checkpoint-1605/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:343ab02dc190765ece461966c5beb5a8a898865f49e1d369db45f56ab5189b82
+size 15920

33_128_e5_3e-5/checkpoint-1605/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:37dc8b63f7b08f41ebc7d2f2a8f3db5ac84cae2f7e2238f80e224e9b9d494fe3
+size 15920

33_128_e5_3e-5/checkpoint-1605/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:41516ce4a73d044d9ba8439090c9d5e6056c871f75a65ba4958097567eb3e300
+size 15920

33_128_e5_3e-5/checkpoint-1605/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:348178de7dde625902bc8d6915d0deaf86bece07a1e3a05b9d5670ae91f178c8
+size 1064

33_128_e5_3e-5/checkpoint-1605/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
+  }
+}

33_128_e5_3e-5/checkpoint-1605/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
+}

33_128_e5_3e-5/checkpoint-1605/trainer_state.json

@@ -0,0 +1,2281 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 1605,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01557632398753894,
+      "grad_norm": 1.1759116649627686,
+      "learning_rate": 1.4814814814814815e-06,
+      "loss": 1.3361,
+      "step": 5
+    },
+    {
+      "epoch": 0.03115264797507788,
+      "grad_norm": 0.9162538647651672,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.3261,
+      "step": 10
+    },
+    {
+      "epoch": 0.04672897196261682,
+      "grad_norm": 0.7471997141838074,
+      "learning_rate": 5.185185185185185e-06,
+      "loss": 1.2639,
+      "step": 15
+    },
+    {
+      "epoch": 0.06230529595015576,
+      "grad_norm": 0.5297987461090088,
+      "learning_rate": 7.037037037037037e-06,
+      "loss": 1.2929,
+      "step": 20
+    },
+    {
+      "epoch": 0.0778816199376947,
+      "grad_norm": 0.6010679006576538,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 1.2993,
+      "step": 25
+    },
+    {
+      "epoch": 0.09345794392523364,
+      "grad_norm": 0.5925828218460083,
+      "learning_rate": 1.074074074074074e-05,
+      "loss": 1.286,
+      "step": 30
+    },
+    {
+      "epoch": 0.10903426791277258,
+      "grad_norm": 0.7390991449356079,
+      "learning_rate": 1.2592592592592592e-05,
+      "loss": 1.2887,
+      "step": 35
+    },
+    {
+      "epoch": 0.12461059190031153,
+      "grad_norm": 0.5752444863319397,
+      "learning_rate": 1.4444444444444444e-05,
+      "loss": 1.2614,
+      "step": 40
+    },
+    {
+      "epoch": 0.14018691588785046,
+      "grad_norm": 0.7002797722816467,
+      "learning_rate": 1.6296296296296297e-05,
+      "loss": 1.1979,
+      "step": 45
+    },
+    {
+      "epoch": 0.1557632398753894,
+      "grad_norm": 0.5052980780601501,
+      "learning_rate": 1.814814814814815e-05,
+      "loss": 1.2206,
+      "step": 50
+    },
+    {
+      "epoch": 0.17133956386292834,
+      "grad_norm": 0.45726311206817627,
+      "learning_rate": 1.9999999999999998e-05,
+      "loss": 1.2247,
+      "step": 55
+    },
+    {
+      "epoch": 0.18691588785046728,
+      "grad_norm": 0.575664222240448,
+      "learning_rate": 2.1851851851851852e-05,
+      "loss": 1.2307,
+      "step": 60
+    },
+    {
+      "epoch": 0.20249221183800623,
+      "grad_norm": 0.5114677548408508,
+      "learning_rate": 2.3703703703703703e-05,
+      "loss": 1.1682,
+      "step": 65
+    },
+    {
+      "epoch": 0.21806853582554517,
+      "grad_norm": 0.4684683084487915,
+      "learning_rate": 2.5555555555555557e-05,
+      "loss": 1.1644,
+      "step": 70
+    },
+    {
+      "epoch": 0.2336448598130841,
+      "grad_norm": 0.6372343897819519,
+      "learning_rate": 2.7407407407407408e-05,
+      "loss": 1.1121,
+      "step": 75
+    },
+    {
+      "epoch": 0.24922118380062305,
+      "grad_norm": 0.4992181360721588,
+      "learning_rate": 2.925925925925926e-05,
+      "loss": 1.1629,
+      "step": 80
+    },
+    {
+      "epoch": 0.26479750778816197,
+      "grad_norm": 0.4986189603805542,
+      "learning_rate": 2.9999713164962586e-05,
+      "loss": 1.142,
+      "step": 85
+    },
+    {
+      "epoch": 0.2803738317757009,
+      "grad_norm": 0.5351710915565491,
+      "learning_rate": 2.9997960323904657e-05,
+      "loss": 1.1477,
+      "step": 90
+    },
+    {
+      "epoch": 0.29595015576323985,
+      "grad_norm": 0.5250363945960999,
+      "learning_rate": 2.999461418057401e-05,
+      "loss": 1.0438,
+      "step": 95
+    },
+    {
+      "epoch": 0.3115264797507788,
+      "grad_norm": 0.5040462017059326,
+      "learning_rate": 2.998967509044683e-05,
+      "loss": 1.1118,
+      "step": 100
+    },
+    {
+      "epoch": 0.32710280373831774,
+      "grad_norm": 0.54499351978302,
+      "learning_rate": 2.998314357822537e-05,
+      "loss": 1.0831,
+      "step": 105
+    },
+    {
+      "epoch": 0.3426791277258567,
+      "grad_norm": 0.563107430934906,
+      "learning_rate": 2.9975020337782196e-05,
+      "loss": 1.0647,
+      "step": 110
+    },
+    {
+      "epoch": 0.3582554517133956,
+      "grad_norm": 0.5709556937217712,
+      "learning_rate": 2.9965306232086484e-05,
+      "loss": 1.0379,
+      "step": 115
+    },
+    {
+      "epoch": 0.37383177570093457,
+      "grad_norm": 0.5909741520881653,
+      "learning_rate": 2.9954002293112343e-05,
+      "loss": 1.046,
+      "step": 120
+    },
+    {
+      "epoch": 0.3894080996884735,
+      "grad_norm": 0.7227066159248352,
+      "learning_rate": 2.994110972172917e-05,
+      "loss": 1.0072,
+      "step": 125
+    },
+    {
+      "epoch": 0.40498442367601245,
+      "grad_norm": 0.7385948300361633,
+      "learning_rate": 2.99266298875741e-05,
+      "loss": 0.9954,
+      "step": 130
+    },
+    {
+      "epoch": 0.4205607476635514,
+      "grad_norm": 0.6370318531990051,
+      "learning_rate": 2.9910564328906474e-05,
+      "loss": 0.9612,
+      "step": 135
+    },
+    {
+      "epoch": 0.43613707165109034,
+      "grad_norm": 0.6569796204566956,
+      "learning_rate": 2.9892914752444444e-05,
+      "loss": 0.9786,
+      "step": 140
+    },
+    {
+      "epoch": 0.4517133956386293,
+      "grad_norm": 0.7288140058517456,
+      "learning_rate": 2.987368303318366e-05,
+      "loss": 0.9255,
+      "step": 145
+    },
+    {
+      "epoch": 0.4672897196261682,
+      "grad_norm": 0.6503431797027588,
+      "learning_rate": 2.9852871214198072e-05,
+      "loss": 1.0384,
+      "step": 150
+    },
+    {
+      "epoch": 0.48286604361370716,
+      "grad_norm": 0.6864191889762878,
+      "learning_rate": 2.9830481506422872e-05,
+      "loss": 0.955,
+      "step": 155
+    },
+    {
+      "epoch": 0.4984423676012461,
+      "grad_norm": 0.7586852312088013,
+      "learning_rate": 2.9806516288419662e-05,
+      "loss": 0.977,
+      "step": 160
+    },
+    {
+      "epoch": 0.514018691588785,
+      "grad_norm": 0.8165059685707092,
+      "learning_rate": 2.9780978106123713e-05,
+      "loss": 0.9314,
+      "step": 165
+    },
+    {
+      "epoch": 0.5295950155763239,
+      "grad_norm": 0.7308902144432068,
+      "learning_rate": 2.975386967257353e-05,
+      "loss": 0.9298,
+      "step": 170
+    },
+    {
+      "epoch": 0.5451713395638629,
+      "grad_norm": 0.7369241714477539,
+      "learning_rate": 2.9725193867622633e-05,
+      "loss": 0.8837,
+      "step": 175
+    },
+    {
+      "epoch": 0.5607476635514018,
+      "grad_norm": 0.8425676822662354,
+      "learning_rate": 2.969495373763361e-05,
+      "loss": 0.9184,
+      "step": 180
+    },
+    {
+      "epoch": 0.5763239875389408,
+      "grad_norm": 0.7827080488204956,
+      "learning_rate": 2.966315249515448e-05,
+      "loss": 0.8533,
+      "step": 185
+    },
+    {
+      "epoch": 0.5919003115264797,
+      "grad_norm": 0.8050850033760071,
+      "learning_rate": 2.9629793518577443e-05,
+      "loss": 0.9124,
+      "step": 190
+    },
+    {
+      "epoch": 0.6074766355140186,
+      "grad_norm": 0.8091520667076111,
+      "learning_rate": 2.9594880351779914e-05,
+      "loss": 0.8296,
+      "step": 195
+    },
+    {
+      "epoch": 0.6230529595015576,
+      "grad_norm": 1.0030003786087036,
+      "learning_rate": 2.9558416703748117e-05,
+      "loss": 0.9065,
+      "step": 200
+    },
+    {
+      "epoch": 0.6386292834890965,
+      "grad_norm": 0.8687117099761963,
+      "learning_rate": 2.952040644818301e-05,
+      "loss": 0.8802,
+      "step": 205
+    },
+    {
+      "epoch": 0.6542056074766355,
+      "grad_norm": 0.9049736261367798,
+      "learning_rate": 2.948085362308877e-05,
+      "loss": 0.8904,
+      "step": 210
+    },
+    {
+      "epoch": 0.6697819314641744,
+      "grad_norm": 1.2105063199996948,
+      "learning_rate": 2.943976243034385e-05,
+      "loss": 0.8275,
+      "step": 215
+    },
+    {
+      "epoch": 0.6853582554517134,
+      "grad_norm": 0.8405113816261292,
+      "learning_rate": 2.9397137235254543e-05,
+      "loss": 0.8687,
+      "step": 220
+    },
+    {
+      "epoch": 0.7009345794392523,
+      "grad_norm": 0.9490503072738647,
+      "learning_rate": 2.935298256609128e-05,
+      "loss": 0.8583,
+      "step": 225
+    },
+    {
+      "epoch": 0.7165109034267912,
+      "grad_norm": 0.877875030040741,
+      "learning_rate": 2.9307303113607557e-05,
+      "loss": 0.8134,
+      "step": 230
+    },
+    {
+      "epoch": 0.7320872274143302,
+      "grad_norm": 0.9228745698928833,
+      "learning_rate": 2.92601037305416e-05,
+      "loss": 0.8101,
+      "step": 235
+    },
+    {
+      "epoch": 0.7476635514018691,
+      "grad_norm": 0.8575282096862793,
+      "learning_rate": 2.9211389431100854e-05,
+      "loss": 0.7929,
+      "step": 240
+    },
+    {
+      "epoch": 0.7632398753894081,
+      "grad_norm": 1.0170135498046875,
+      "learning_rate": 2.91611653904293e-05,
+      "loss": 0.7931,
+      "step": 245
+    },
+    {
+      "epoch": 0.778816199376947,
+      "grad_norm": 0.9553667902946472,
+      "learning_rate": 2.9109436944057662e-05,
+      "loss": 0.8035,
+      "step": 250
+    },
+    {
+      "epoch": 0.794392523364486,
+      "grad_norm": 0.919255793094635,
+      "learning_rate": 2.9056209587336592e-05,
+      "loss": 0.7739,
+      "step": 255
+    },
+    {
+      "epoch": 0.8099688473520249,
+      "grad_norm": 0.9529440999031067,
+      "learning_rate": 2.90014889748529e-05,
+      "loss": 0.7342,
+      "step": 260
+    },
+    {
+      "epoch": 0.8255451713395638,
+      "grad_norm": 0.8980151414871216,
+      "learning_rate": 2.8945280919828796e-05,
+      "loss": 0.7176,
+      "step": 265
+    },
+    {
+      "epoch": 0.8411214953271028,
+      "grad_norm": 0.8686856627464294,
+      "learning_rate": 2.888759139350437e-05,
+      "loss": 0.7349,
+      "step": 270
+    },
+    {
+      "epoch": 0.8566978193146417,
+      "grad_norm": 0.9111968278884888,
+      "learning_rate": 2.8828426524503205e-05,
+      "loss": 0.754,
+      "step": 275
+    },
+    {
+      "epoch": 0.8722741433021807,
+      "grad_norm": 0.9784020185470581,
+      "learning_rate": 2.8767792598181326e-05,
+      "loss": 0.7536,
+      "step": 280
+    },
+    {
+      "epoch": 0.8878504672897196,
+      "grad_norm": 0.9565854072570801,
+      "learning_rate": 2.8705696055959466e-05,
+      "loss": 0.7186,
+      "step": 285
+    },
+    {
+      "epoch": 0.9034267912772586,
+      "grad_norm": 1.0711281299591064,
+      "learning_rate": 2.864214349463877e-05,
+      "loss": 0.6841,
+      "step": 290
+    },
+    {
+      "epoch": 0.9190031152647975,
+      "grad_norm": 1.0697325468063354,
+      "learning_rate": 2.8577141665699988e-05,
+      "loss": 0.7694,
+      "step": 295
+    },
+    {
+      "epoch": 0.9345794392523364,
+      "grad_norm": 0.947496771812439,
+      "learning_rate": 2.8510697474586228e-05,
+      "loss": 0.7155,
+      "step": 300
+    },
+    {
+      "epoch": 0.9501557632398754,
+      "grad_norm": 0.9437328577041626,
+      "learning_rate": 2.8442817979969366e-05,
+      "loss": 0.7146,
+      "step": 305
+    },
+    {
+      "epoch": 0.9657320872274143,
+      "grad_norm": 1.0520305633544922,
+      "learning_rate": 2.8373510393000163e-05,
+      "loss": 0.6837,
+      "step": 310
+    },
+    {
+      "epoch": 0.9813084112149533,
+      "grad_norm": 1.0140153169631958,
+      "learning_rate": 2.8302782076542207e-05,
+      "loss": 0.6774,
+      "step": 315
+    },
+    {
+      "epoch": 0.9968847352024922,
+      "grad_norm": 1.0676052570343018,
+      "learning_rate": 2.823064054438969e-05,
+      "loss": 0.6743,
+      "step": 320
+    },
+    {
+      "epoch": 1.0124610591900312,
+      "grad_norm": 1.0298436880111694,
+      "learning_rate": 2.8157093460469227e-05,
+      "loss": 0.6426,
+      "step": 325
+    },
+    {
+      "epoch": 1.02803738317757,
+      "grad_norm": 0.9810764193534851,
+      "learning_rate": 2.808214863802564e-05,
+      "loss": 0.6041,
+      "step": 330
+    },
+    {
+      "epoch": 1.043613707165109,
+      "grad_norm": 1.102921485900879,
+      "learning_rate": 2.8005814038791948e-05,
+      "loss": 0.5911,
+      "step": 335
+    },
+    {
+      "epoch": 1.0591900311526479,
+      "grad_norm": 1.093255877494812,
+      "learning_rate": 2.7928097772143555e-05,
+      "loss": 0.541,
+      "step": 340
+    },
+    {
+      "epoch": 1.074766355140187,
+      "grad_norm": 1.167300820350647,
+      "learning_rate": 2.7849008094236734e-05,
+      "loss": 0.5812,
+      "step": 345
+    },
+    {
+      "epoch": 1.0903426791277258,
+      "grad_norm": 1.170893907546997,
+      "learning_rate": 2.7768553407131555e-05,
+      "loss": 0.5476,
+      "step": 350
+    },
+    {
+      "epoch": 1.1059190031152648,
+      "grad_norm": 1.0814523696899414,
+      "learning_rate": 2.7686742257899288e-05,
+      "loss": 0.5671,
+      "step": 355
+    },
+    {
+      "epoch": 1.1214953271028036,
+      "grad_norm": 1.0484672784805298,
+      "learning_rate": 2.7603583337714412e-05,
+      "loss": 0.5707,
+      "step": 360
+    },
+    {
+      "epoch": 1.1370716510903427,
+      "grad_norm": 1.2223975658416748,
+      "learning_rate": 2.7519085480931312e-05,
+      "loss": 0.5276,
+      "step": 365
+    },
+    {
+      "epoch": 1.1526479750778815,
+      "grad_norm": 1.2355427742004395,
+      "learning_rate": 2.7433257664145757e-05,
+      "loss": 0.5492,
+      "step": 370
+    },
+    {
+      "epoch": 1.1682242990654206,
+      "grad_norm": 1.2064166069030762,
+      "learning_rate": 2.7346109005241294e-05,
+      "loss": 0.5337,
+      "step": 375
+    },
+    {
+      "epoch": 1.1838006230529594,
+      "grad_norm": 1.1383205652236938,
+      "learning_rate": 2.7257648762420586e-05,
+      "loss": 0.5487,
+      "step": 380
+    },
+    {
+      "epoch": 1.1993769470404985,
+      "grad_norm": 1.1303715705871582,
+      "learning_rate": 2.716788633322189e-05,
+      "loss": 0.589,
+      "step": 385
+    },
+    {
+      "epoch": 1.2149532710280373,
+      "grad_norm": 1.2811906337738037,
+      "learning_rate": 2.707683125352071e-05,
+      "loss": 0.5475,
+      "step": 390
+    },
+    {
+      "epoch": 1.2305295950155763,
+      "grad_norm": 1.3321250677108765,
+      "learning_rate": 2.6984493196516756e-05,
+      "loss": 0.5283,
+      "step": 395
+    },
+    {
+      "epoch": 1.2461059190031152,
+      "grad_norm": 1.132002830505371,
+      "learning_rate": 2.6890881971706312e-05,
+      "loss": 0.5733,
+      "step": 400
+    },
+    {
+      "epoch": 1.2616822429906542,
+      "grad_norm": 1.1940255165100098,
+      "learning_rate": 2.679600752384013e-05,
+      "loss": 0.578,
+      "step": 405
+    },
+    {
+      "epoch": 1.277258566978193,
+      "grad_norm": 1.219218134880066,
+      "learning_rate": 2.669987993186696e-05,
+      "loss": 0.5359,
+      "step": 410
+    },
+    {
+      "epoch": 1.2928348909657321,
+      "grad_norm": 1.1772730350494385,
+      "learning_rate": 2.6602509407862806e-05,
+      "loss": 0.561,
+      "step": 415
+    },
+    {
+      "epoch": 1.308411214953271,
+      "grad_norm": 1.1644192934036255,
+      "learning_rate": 2.6503906295946063e-05,
+      "loss": 0.5552,
+      "step": 420
+    },
+    {
+      "epoch": 1.32398753894081,
+      "grad_norm": 1.1815481185913086,
+      "learning_rate": 2.640408107117859e-05,
+      "loss": 0.4946,
+      "step": 425
+    },
+    {
+      "epoch": 1.3395638629283488,
+      "grad_norm": 1.0733100175857544,
+      "learning_rate": 2.630304433845293e-05,
+      "loss": 0.4903,
+      "step": 430
+    },
+    {
+      "epoch": 1.355140186915888,
+      "grad_norm": 1.1842522621154785,
+      "learning_rate": 2.6200806831365685e-05,
+      "loss": 0.493,
+      "step": 435
+    },
+    {
+      "epoch": 1.3707165109034267,
+      "grad_norm": 1.120402216911316,
+      "learning_rate": 2.609737941107722e-05,
+      "loss": 0.5127,
+      "step": 440
+    },
+    {
+      "epoch": 1.3862928348909658,
+      "grad_norm": 1.1652181148529053,
+      "learning_rate": 2.5992773065157863e-05,
+      "loss": 0.4745,
+      "step": 445
+    },
+    {
+      "epoch": 1.4018691588785046,
+      "grad_norm": 1.093859076499939,
+      "learning_rate": 2.5886998906420635e-05,
+      "loss": 0.4846,
+      "step": 450
+    },
+    {
+      "epoch": 1.4174454828660437,
+      "grad_norm": 1.2011666297912598,
+      "learning_rate": 2.5780068171740677e-05,
+      "loss": 0.471,
+      "step": 455
+    },
+    {
+      "epoch": 1.4330218068535825,
+      "grad_norm": 1.2274277210235596,
+      "learning_rate": 2.56719922208615e-05,
+      "loss": 0.4768,
+      "step": 460
+    },
+    {
+      "epoch": 1.4485981308411215,
+      "grad_norm": 1.198301076889038,
+      "learning_rate": 2.5562782535188213e-05,
+      "loss": 0.4921,
+      "step": 465
+    },
+    {
+      "epoch": 1.4641744548286604,
+      "grad_norm": 1.0955687761306763,
+      "learning_rate": 2.5452450716567763e-05,
+      "loss": 0.477,
+      "step": 470
+    },
+    {
+      "epoch": 1.4797507788161994,
+      "grad_norm": 1.079829216003418,
+      "learning_rate": 2.534100848605647e-05,
+      "loss": 0.4636,
+      "step": 475
+    },
+    {
+      "epoch": 1.4953271028037383,
+      "grad_norm": 1.049848198890686,
+      "learning_rate": 2.522846768267478e-05,
+      "loss": 0.4796,
+      "step": 480
+    },
+    {
+      "epoch": 1.5109034267912773,
+      "grad_norm": 1.2734771966934204,
+      "learning_rate": 2.5114840262149604e-05,
+      "loss": 0.4543,
+      "step": 485
+    },
+    {
+      "epoch": 1.5264797507788161,
+      "grad_norm": 1.1971790790557861,
+      "learning_rate": 2.500013829564419e-05,
+      "loss": 0.4807,
+      "step": 490
+    },
+    {
+      "epoch": 1.542056074766355,
+      "grad_norm": 1.1881734132766724,
+      "learning_rate": 2.4884373968475735e-05,
+      "loss": 0.4029,
+      "step": 495
+    },
+    {
+      "epoch": 1.557632398753894,
+      "grad_norm": 1.0760289430618286,
+      "learning_rate": 2.4767559578820897e-05,
+      "loss": 0.4933,
+      "step": 500
+    },
+    {
+      "epoch": 1.573208722741433,
+      "grad_norm": 1.2057150602340698,
+      "learning_rate": 2.4649707536409297e-05,
+      "loss": 0.4271,
+      "step": 505
+    },
+    {
+      "epoch": 1.588785046728972,
+      "grad_norm": 1.0976535081863403,
+      "learning_rate": 2.4530830361205184e-05,
+      "loss": 0.4645,
+      "step": 510
+    },
+    {
+      "epoch": 1.6043613707165107,
+      "grad_norm": 1.481553077697754,
+      "learning_rate": 2.4410940682077377e-05,
+      "loss": 0.4233,
+      "step": 515
+    },
+    {
+      "epoch": 1.6199376947040498,
+      "grad_norm": 1.1634652614593506,
+      "learning_rate": 2.429005123545763e-05,
+      "loss": 0.4394,
+      "step": 520
+    },
+    {
+      "epoch": 1.6355140186915889,
+      "grad_norm": 1.1656765937805176,
+      "learning_rate": 2.4168174863987605e-05,
+      "loss": 0.4434,
+      "step": 525
+    },
+    {
+      "epoch": 1.6510903426791277,
+      "grad_norm": 1.0849956274032593,
+      "learning_rate": 2.404532451515453e-05,
+      "loss": 0.4255,
+      "step": 530
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 1.0943026542663574,
+      "learning_rate": 2.392151323991571e-05,
+      "loss": 0.4099,
+      "step": 535
+    },
+    {
+      "epoch": 1.6822429906542056,
+      "grad_norm": 1.2393852472305298,
+      "learning_rate": 2.3796754191312094e-05,
+      "loss": 0.3917,
+      "step": 540
+    },
+    {
+      "epoch": 1.6978193146417446,
+      "grad_norm": 1.2227272987365723,
+      "learning_rate": 2.367106062307095e-05,
+      "loss": 0.4232,
+      "step": 545
+    },
+    {
+      "epoch": 1.7133956386292835,
+      "grad_norm": 1.3530696630477905,
+      "learning_rate": 2.3544445888197855e-05,
+      "loss": 0.3996,
+      "step": 550
+    },
+    {
+      "epoch": 1.7289719626168223,
+      "grad_norm": 1.1738826036453247,
+      "learning_rate": 2.3416923437558175e-05,
+      "loss": 0.4333,
+      "step": 555
+    },
+    {
+      "epoch": 1.7445482866043613,
+      "grad_norm": 1.2589629888534546,
+      "learning_rate": 2.3288506818448057e-05,
+      "loss": 0.4021,
+      "step": 560
+    },
+    {
+      "epoch": 1.7601246105919004,
+      "grad_norm": 1.1301474571228027,
+      "learning_rate": 2.3159209673155306e-05,
+      "loss": 0.4056,
+      "step": 565
+    },
+    {
+      "epoch": 1.7757009345794392,
+      "grad_norm": 1.2090823650360107,
+      "learning_rate": 2.302904573751006e-05,
+      "loss": 0.4173,
+      "step": 570
+    },
+    {
+      "epoch": 1.791277258566978,
+      "grad_norm": 1.1565548181533813,
+      "learning_rate": 2.2898028839425592e-05,
+      "loss": 0.3904,
+      "step": 575
+    },
+    {
+      "epoch": 1.8068535825545171,
+      "grad_norm": 1.1722818613052368,
+      "learning_rate": 2.2766172897429272e-05,
+      "loss": 0.4146,
+      "step": 580
+    },
+    {
+      "epoch": 1.8224299065420562,
+      "grad_norm": 1.0461018085479736,
+      "learning_rate": 2.2633491919183988e-05,
+      "loss": 0.4062,
+      "step": 585
+    },
+    {
+      "epoch": 1.838006230529595,
+      "grad_norm": 1.0770604610443115,
+      "learning_rate": 2.25e-05,
+      "loss": 0.4073,
+      "step": 590
+    },
+    {
+      "epoch": 1.8535825545171338,
+      "grad_norm": 1.1962300539016724,
+      "learning_rate": 2.2365711321337573e-05,
+      "loss": 0.3932,
+      "step": 595
+    },
+    {
+      "epoch": 1.8691588785046729,
+      "grad_norm": 1.264811396598816,
+      "learning_rate": 2.2230640149300375e-05,
+      "loss": 0.3782,
+      "step": 600
+    },
+    {
+      "epoch": 1.884735202492212,
+      "grad_norm": 1.1047016382217407,
+      "learning_rate": 2.2094800833119954e-05,
+      "loss": 0.3725,
+      "step": 605
+    },
+    {
+      "epoch": 1.9003115264797508,
+      "grad_norm": 1.2060890197753906,
+      "learning_rate": 2.1958207803631343e-05,
+      "loss": 0.3467,
+      "step": 610
+    },
+    {
+      "epoch": 1.9158878504672896,
+      "grad_norm": 1.3942707777023315,
+      "learning_rate": 2.1820875571739987e-05,
+      "loss": 0.3567,
+      "step": 615
+    },
+    {
+      "epoch": 1.9314641744548287,
+      "grad_norm": 1.4150056838989258,
+      "learning_rate": 2.1682818726880227e-05,
+      "loss": 0.3699,
+      "step": 620
+    },
+    {
+      "epoch": 1.9470404984423677,
+      "grad_norm": 1.1999322175979614,
+      "learning_rate": 2.154405193546535e-05,
+      "loss": 0.37,
+      "step": 625
+    },
+    {
+      "epoch": 1.9626168224299065,
+      "grad_norm": 1.3075921535491943,
+      "learning_rate": 2.1404589939329538e-05,
+      "loss": 0.3518,
+      "step": 630
+    },
+    {
+      "epoch": 1.9781931464174454,
+      "grad_norm": 1.3312402963638306,
+      "learning_rate": 2.1264447554161766e-05,
+      "loss": 0.3646,
+      "step": 635
+    },
+    {
+      "epoch": 1.9937694704049844,
+      "grad_norm": 1.1783993244171143,
+      "learning_rate": 2.1123639667931854e-05,
+      "loss": 0.3462,
+      "step": 640
+    },
+    {
+      "epoch": 2.0093457943925235,
+      "grad_norm": 1.1549129486083984,
+      "learning_rate": 2.0982181239308887e-05,
+      "loss": 0.2886,
+      "step": 645
+    },
+    {
+      "epoch": 2.0249221183800623,
+      "grad_norm": 1.127125859260559,
+      "learning_rate": 2.0840087296072017e-05,
+      "loss": 0.3207,
+      "step": 650
+    },
+    {
+      "epoch": 2.040498442367601,
+      "grad_norm": 1.2090187072753906,
+      "learning_rate": 2.0697372933514076e-05,
+      "loss": 0.284,
+      "step": 655
+    },
+    {
+      "epoch": 2.05607476635514,
+      "grad_norm": 1.1644618511199951,
+      "learning_rate": 2.0554053312837865e-05,
+      "loss": 0.2961,
+      "step": 660
+    },
+    {
+      "epoch": 2.0716510903426792,
+      "grad_norm": 1.1487720012664795,
+      "learning_rate": 2.0410143659545546e-05,
+      "loss": 0.2789,
+      "step": 665
+    },
+    {
+      "epoch": 2.087227414330218,
+      "grad_norm": 1.2465693950653076,
+      "learning_rate": 2.0265659261821163e-05,
+      "loss": 0.2802,
+      "step": 670
+    },
+    {
+      "epoch": 2.102803738317757,
+      "grad_norm": 1.16876220703125,
+      "learning_rate": 2.0120615468906502e-05,
+      "loss": 0.2761,
+      "step": 675
+    },
+    {
+      "epoch": 2.1183800623052957,
+      "grad_norm": 1.255817174911499,
+      "learning_rate": 1.9975027689470462e-05,
+      "loss": 0.276,
+      "step": 680
+    },
+    {
+      "epoch": 2.133956386292835,
+      "grad_norm": 1.2631534337997437,
+      "learning_rate": 1.9828911389972142e-05,
+      "loss": 0.2738,
+      "step": 685
+    },
+    {
+      "epoch": 2.149532710280374,
+      "grad_norm": 1.2952145338058472,
+      "learning_rate": 1.968228209301775e-05,
+      "loss": 0.2859,
+      "step": 690
+    },
+    {
+      "epoch": 2.1651090342679127,
+      "grad_norm": 1.2698311805725098,
+      "learning_rate": 1.9535155375711576e-05,
+      "loss": 0.2457,
+      "step": 695
+    },
+    {
+      "epoch": 2.1806853582554515,
+      "grad_norm": 1.2509089708328247,
+      "learning_rate": 1.9387546868001166e-05,
+      "loss": 0.3084,
+      "step": 700
+    },
+    {
+      "epoch": 2.196261682242991,
+      "grad_norm": 1.184861183166504,
+      "learning_rate": 1.9239472251016856e-05,
+      "loss": 0.2587,
+      "step": 705
+    },
+    {
+      "epoch": 2.2118380062305296,
+      "grad_norm": 1.4472482204437256,
+      "learning_rate": 1.9090947255405935e-05,
+      "loss": 0.2836,
+      "step": 710
+    },
+    {
+      "epoch": 2.2274143302180685,
+      "grad_norm": 1.2142419815063477,
+      "learning_rate": 1.8941987659661474e-05,
+      "loss": 0.2719,
+      "step": 715
+    },
+    {
+      "epoch": 2.2429906542056073,
+      "grad_norm": 1.0755081176757812,
+      "learning_rate": 1.879260928844611e-05,
+      "loss": 0.2573,
+      "step": 720
+    },
+    {
+      "epoch": 2.2585669781931466,
+      "grad_norm": 1.1565829515457153,
+      "learning_rate": 1.8642828010910937e-05,
+      "loss": 0.2675,
+      "step": 725
+    },
+    {
+      "epoch": 2.2741433021806854,
+      "grad_norm": 1.233801007270813,
+      "learning_rate": 1.849265973900962e-05,
+      "loss": 0.2826,
+      "step": 730
+    },
+    {
+      "epoch": 2.289719626168224,
+      "grad_norm": 1.4652220010757446,
+      "learning_rate": 1.8342120425808036e-05,
+      "loss": 0.2218,
+      "step": 735
+    },
+    {
+      "epoch": 2.305295950155763,
+      "grad_norm": 1.1132934093475342,
+      "learning_rate": 1.819122606378947e-05,
+      "loss": 0.2359,
+      "step": 740
+    },
+    {
+      "epoch": 2.3208722741433023,
+      "grad_norm": 1.2268480062484741,
+      "learning_rate": 1.803999268315568e-05,
+      "loss": 0.2319,
+      "step": 745
+    },
+    {
+      "epoch": 2.336448598130841,
+      "grad_norm": 1.1891158819198608,
+      "learning_rate": 1.788843635012393e-05,
+      "loss": 0.244,
+      "step": 750
+    },
+    {
+      "epoch": 2.35202492211838,
+      "grad_norm": 1.0149863958358765,
+      "learning_rate": 1.77365731652202e-05,
+      "loss": 0.2484,
+      "step": 755
+    },
+    {
+      "epoch": 2.367601246105919,
+      "grad_norm": 1.2054303884506226,
+      "learning_rate": 1.758441926156875e-05,
+      "loss": 0.2764,
+      "step": 760
+    },
+    {
+      "epoch": 2.383177570093458,
+      "grad_norm": 1.3622403144836426,
+      "learning_rate": 1.7431990803178248e-05,
+      "loss": 0.2449,
+      "step": 765
+    },
+    {
+      "epoch": 2.398753894080997,
+      "grad_norm": 1.3636577129364014,
+      "learning_rate": 1.727930398322456e-05,
+      "loss": 0.2437,
+      "step": 770
+    },
+    {
+      "epoch": 2.4143302180685358,
+      "grad_norm": 1.1471670866012573,
+      "learning_rate": 1.7126375022330483e-05,
+      "loss": 0.2521,
+      "step": 775
+    },
+    {
+      "epoch": 2.4299065420560746,
+      "grad_norm": 1.2446048259735107,
+      "learning_rate": 1.697322016684257e-05,
+      "loss": 0.2313,
+      "step": 780
+    },
+    {
+      "epoch": 2.445482866043614,
+      "grad_norm": 1.3057276010513306,
+      "learning_rate": 1.6819855687105197e-05,
+      "loss": 0.2643,
+      "step": 785
+    },
+    {
+      "epoch": 2.4610591900311527,
+      "grad_norm": 1.2132842540740967,
+      "learning_rate": 1.6666297875732072e-05,
+      "loss": 0.2481,
+      "step": 790
+    },
+    {
+      "epoch": 2.4766355140186915,
+      "grad_norm": 1.6572365760803223,
+      "learning_rate": 1.6512563045875424e-05,
+      "loss": 0.2505,
+      "step": 795
+    },
+    {
+      "epoch": 2.4922118380062304,
+      "grad_norm": 1.0174639225006104,
+      "learning_rate": 1.6358667529492977e-05,
+      "loss": 0.2086,
+      "step": 800
+    },
+    {
+      "epoch": 2.507788161993769,
+      "grad_norm": 1.2086824178695679,
+      "learning_rate": 1.6204627675612904e-05,
+      "loss": 0.2166,
+      "step": 805
+    },
+    {
+      "epoch": 2.5233644859813085,
+      "grad_norm": 1.0739854574203491,
+      "learning_rate": 1.6050459848597023e-05,
+      "loss": 0.211,
+      "step": 810
+    },
+    {
+      "epoch": 2.5389408099688473,
+      "grad_norm": 1.2083333730697632,
+      "learning_rate": 1.589618042640233e-05,
+      "loss": 0.2279,
+      "step": 815
+    },
+    {
+      "epoch": 2.554517133956386,
+      "grad_norm": 1.1843072175979614,
+      "learning_rate": 1.574180579884109e-05,
+      "loss": 0.2055,
+      "step": 820
+    },
+    {
+      "epoch": 2.5700934579439254,
+      "grad_norm": 1.1035974025726318,
+      "learning_rate": 1.558735236583966e-05,
+      "loss": 0.2195,
+      "step": 825
+    },
+    {
+      "epoch": 2.5856697819314642,
+      "grad_norm": 1.1672109365463257,
+      "learning_rate": 1.5432836535696275e-05,
+      "loss": 0.2191,
+      "step": 830
+    },
+    {
+      "epoch": 2.601246105919003,
+      "grad_norm": 1.1042742729187012,
+      "learning_rate": 1.5278274723337895e-05,
+      "loss": 0.216,
+      "step": 835
+    },
+    {
+      "epoch": 2.616822429906542,
+      "grad_norm": 1.2599068880081177,
+      "learning_rate": 1.5123683348576398e-05,
+      "loss": 0.2432,
+      "step": 840
+    },
+    {
+      "epoch": 2.6323987538940807,
+      "grad_norm": 1.0473904609680176,
+      "learning_rate": 1.496907883436419e-05,
+      "loss": 0.2048,
+      "step": 845
+    },
+    {
+      "epoch": 2.64797507788162,
+      "grad_norm": 1.183077096939087,
+      "learning_rate": 1.4814477605049557e-05,
+      "loss": 0.2252,
+      "step": 850
+    },
+    {
+      "epoch": 2.663551401869159,
+      "grad_norm": 1.1935157775878906,
+      "learning_rate": 1.465989608463181e-05,
+      "loss": 0.221,
+      "step": 855
+    },
+    {
+      "epoch": 2.6791277258566977,
+      "grad_norm": 1.2083299160003662,
+      "learning_rate": 1.450535069501649e-05,
+      "loss": 0.2047,
+      "step": 860
+    },
+    {
+      "epoch": 2.694704049844237,
+      "grad_norm": 1.1584832668304443,
+      "learning_rate": 1.4350857854270807e-05,
+      "loss": 0.2155,
+      "step": 865
+    },
+    {
+      "epoch": 2.710280373831776,
+      "grad_norm": 1.0836254358291626,
+      "learning_rate": 1.4196433974879452e-05,
+      "loss": 0.1927,
+      "step": 870
+    },
+    {
+      "epoch": 2.7258566978193146,
+      "grad_norm": 1.1828125715255737,
+      "learning_rate": 1.404209546200104e-05,
+      "loss": 0.2193,
+      "step": 875
+    },
+    {
+      "epoch": 2.7414330218068534,
+      "grad_norm": 1.276404619216919,
+      "learning_rate": 1.388785871172531e-05,
+      "loss": 0.2185,
+      "step": 880
+    },
+    {
+      "epoch": 2.7570093457943923,
+      "grad_norm": 1.2181018590927124,
+      "learning_rate": 1.3733740109331282e-05,
+      "loss": 0.2102,
+      "step": 885
+    },
+    {
+      "epoch": 2.7725856697819315,
+      "grad_norm": 1.1601808071136475,
+      "learning_rate": 1.3579756027546597e-05,
+      "loss": 0.2082,
+      "step": 890
+    },
+    {
+      "epoch": 2.7881619937694704,
+      "grad_norm": 1.2979238033294678,
+      "learning_rate": 1.3425922824808146e-05,
+      "loss": 0.2099,
+      "step": 895
+    },
+    {
+      "epoch": 2.803738317757009,
+      "grad_norm": 1.068938970565796,
+      "learning_rate": 1.327225684352425e-05,
+      "loss": 0.2054,
+      "step": 900
+    },
+    {
+      "epoch": 2.8193146417445485,
+      "grad_norm": 1.171219825744629,
+      "learning_rate": 1.3118774408338517e-05,
+      "loss": 0.2168,
+      "step": 905
+    },
+    {
+      "epoch": 2.8348909657320873,
+      "grad_norm": 1.1562765836715698,
+      "learning_rate": 1.2965491824395614e-05,
+      "loss": 0.2266,
+      "step": 910
+    },
+    {
+      "epoch": 2.850467289719626,
+      "grad_norm": 1.0936180353164673,
+      "learning_rate": 1.2812425375609101e-05,
+      "loss": 0.1971,
+      "step": 915
+    },
+    {
+      "epoch": 2.866043613707165,
+      "grad_norm": 1.166494607925415,
+      "learning_rate": 1.2659591322931491e-05,
+      "loss": 0.1849,
+      "step": 920
+    },
+    {
+      "epoch": 2.881619937694704,
+      "grad_norm": 1.102161169052124,
+      "learning_rate": 1.2507005902626797e-05,
+      "loss": 0.1951,
+      "step": 925
+    },
+    {
+      "epoch": 2.897196261682243,
+      "grad_norm": 1.0930756330490112,
+      "learning_rate": 1.2354685324545676e-05,
+      "loss": 0.2151,
+      "step": 930
+    },
+    {
+      "epoch": 2.912772585669782,
+      "grad_norm": 1.1219929456710815,
+      "learning_rate": 1.2202645770403376e-05,
+      "loss": 0.1761,
+      "step": 935
+    },
+    {
+      "epoch": 2.9283489096573208,
+      "grad_norm": 1.1747044324874878,
+      "learning_rate": 1.2050903392060674e-05,
+      "loss": 0.1746,
+      "step": 940
+    },
+    {
+      "epoch": 2.94392523364486,
+      "grad_norm": 1.1695518493652344,
+      "learning_rate": 1.1899474309807998e-05,
+      "loss": 0.1868,
+      "step": 945
+    },
+    {
+      "epoch": 2.959501557632399,
+      "grad_norm": 1.3246632814407349,
+      "learning_rate": 1.1748374610652906e-05,
+      "loss": 0.1942,
+      "step": 950
+    },
+    {
+      "epoch": 2.9750778816199377,
+      "grad_norm": 1.2505390644073486,
+      "learning_rate": 1.159762034661105e-05,
+      "loss": 0.2147,
+      "step": 955
+    },
+    {
+      "epoch": 2.9906542056074765,
+      "grad_norm": 1.1202441453933716,
+      "learning_rate": 1.1447227533000939e-05,
+      "loss": 0.1741,
+      "step": 960
+    },
+    {
+      "epoch": 3.0062305295950154,
+      "grad_norm": 1.0747812986373901,
+      "learning_rate": 1.1297212146742543e-05,
+      "loss": 0.1704,
+      "step": 965
+    },
+    {
+      "epoch": 3.0218068535825546,
+      "grad_norm": 1.2752288579940796,
+      "learning_rate": 1.1147590124659983e-05,
+      "loss": 0.1438,
+      "step": 970
+    },
+    {
+      "epoch": 3.0373831775700935,
+      "grad_norm": 1.0363479852676392,
+      "learning_rate": 1.0998377361788489e-05,
+      "loss": 0.1591,
+      "step": 975
+    },
+    {
+      "epoch": 3.0529595015576323,
+      "grad_norm": 1.2088156938552856,
+      "learning_rate": 1.0849589709685816e-05,
+      "loss": 0.1496,
+      "step": 980
+    },
+    {
+      "epoch": 3.068535825545171,
+      "grad_norm": 1.0746065378189087,
+      "learning_rate": 1.0701242974748245e-05,
+      "loss": 0.1276,
+      "step": 985
+    },
+    {
+      "epoch": 3.0841121495327104,
+      "grad_norm": 1.1134722232818604,
+      "learning_rate": 1.0553352916531389e-05,
+      "loss": 0.1523,
+      "step": 990
+    },
+    {
+      "epoch": 3.0996884735202492,
+      "grad_norm": 1.002192735671997,
+      "learning_rate": 1.0405935246076011e-05,
+      "loss": 0.1379,
+      "step": 995
+    },
+    {
+      "epoch": 3.115264797507788,
+      "grad_norm": 1.0264095067977905,
+      "learning_rate": 1.0259005624238931e-05,
+      "loss": 0.1469,
+      "step": 1000
+    },
+    {
+      "epoch": 3.130841121495327,
+      "grad_norm": 1.0532501935958862,
+      "learning_rate": 1.0112579660029341e-05,
+      "loss": 0.1294,
+      "step": 1005
+    },
+    {
+      "epoch": 3.146417445482866,
+      "grad_norm": 1.0389549732208252,
+      "learning_rate": 9.966672908950552e-06,
+      "loss": 0.1246,
+      "step": 1010
+    },
+    {
+      "epoch": 3.161993769470405,
+      "grad_norm": 1.1664501428604126,
+      "learning_rate": 9.82130087134748e-06,
+      "loss": 0.136,
+      "step": 1015
+    },
+    {
+      "epoch": 3.177570093457944,
+      "grad_norm": 1.0726090669631958,
+      "learning_rate": 9.676478990759979e-06,
+      "loss": 0.1422,
+      "step": 1020
+    },
+    {
+      "epoch": 3.1931464174454827,
+      "grad_norm": 1.0180519819259644,
+      "learning_rate": 9.532222652282187e-06,
+      "loss": 0.1381,
+      "step": 1025
+    },
+    {
+      "epoch": 3.208722741433022,
+      "grad_norm": 1.2948464155197144,
+      "learning_rate": 9.388547180928103e-06,
+      "loss": 0.1211,
+      "step": 1030
+    },
+    {
+      "epoch": 3.2242990654205608,
+      "grad_norm": 1.1216830015182495,
+      "learning_rate": 9.245467840003544e-06,
+      "loss": 0.1428,
+      "step": 1035
+    },
+    {
+      "epoch": 3.2398753894080996,
+      "grad_norm": 1.1088138818740845,
+      "learning_rate": 9.102999829484652e-06,
+      "loss": 0.1389,
+      "step": 1040
+    },
+    {
+      "epoch": 3.2554517133956384,
+      "grad_norm": 1.2717032432556152,
+      "learning_rate": 8.961158284403123e-06,
+      "loss": 0.1304,
+      "step": 1045
+    },
+    {
+      "epoch": 3.2710280373831777,
+      "grad_norm": 0.9676588177680969,
+      "learning_rate": 8.819958273238358e-06,
+      "loss": 0.1289,
+      "step": 1050
+    },
+    {
+      "epoch": 3.2866043613707165,
+      "grad_norm": 1.1699531078338623,
+      "learning_rate": 8.679414796316658e-06,
+      "loss": 0.1347,
+      "step": 1055
+    },
+    {
+      "epoch": 3.3021806853582554,
+      "grad_norm": 1.1606409549713135,
+      "learning_rate": 8.539542784217675e-06,
+      "loss": 0.134,
+      "step": 1060
+    },
+    {
+      "epoch": 3.317757009345794,
+      "grad_norm": 1.0339468717575073,
+      "learning_rate": 8.400357096188256e-06,
+      "loss": 0.14,
+      "step": 1065
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 1.2066926956176758,
+      "learning_rate": 8.26187251856389e-06,
+      "loss": 0.1361,
+      "step": 1070
+    },
+    {
+      "epoch": 3.3489096573208723,
+      "grad_norm": 0.9468661546707153,
+      "learning_rate": 8.124103763197911e-06,
+      "loss": 0.1384,
+      "step": 1075
+    },
+    {
+      "epoch": 3.364485981308411,
+      "grad_norm": 1.0264408588409424,
+      "learning_rate": 7.987065465898522e-06,
+      "loss": 0.1282,
+      "step": 1080
+    },
+    {
+      "epoch": 3.38006230529595,
+      "grad_norm": 0.9723944664001465,
+      "learning_rate": 7.85077218487403e-06,
+      "loss": 0.1298,
+      "step": 1085
+    },
+    {
+      "epoch": 3.3956386292834893,
+      "grad_norm": 1.0408103466033936,
+      "learning_rate": 7.715238399186246e-06,
+      "loss": 0.1321,
+      "step": 1090
+    },
+    {
+      "epoch": 3.411214953271028,
+      "grad_norm": 1.1424450874328613,
+      "learning_rate": 7.580478507212283e-06,
+      "loss": 0.1215,
+      "step": 1095
+    },
+    {
+      "epoch": 3.426791277258567,
+      "grad_norm": 1.001061201095581,
+      "learning_rate": 7.446506825114969e-06,
+      "loss": 0.1211,
+      "step": 1100
+    },
+    {
+      "epoch": 3.4423676012461057,
+      "grad_norm": 1.1220428943634033,
+      "learning_rate": 7.313337585321978e-06,
+      "loss": 0.1158,
+      "step": 1105
+    },
+    {
+      "epoch": 3.457943925233645,
+      "grad_norm": 1.1033799648284912,
+      "learning_rate": 7.180984935013862e-06,
+      "loss": 0.1442,
+      "step": 1110
+    },
+    {
+      "epoch": 3.473520249221184,
+      "grad_norm": 1.141599416732788,
+      "learning_rate": 7.0494629346210924e-06,
+      "loss": 0.1363,
+      "step": 1115
+    },
+    {
+      "epoch": 3.4890965732087227,
+      "grad_norm": 0.9771517515182495,
+      "learning_rate": 6.918785556330396e-06,
+      "loss": 0.127,
+      "step": 1120
+    },
+    {
+      "epoch": 3.5046728971962615,
+      "grad_norm": 0.999470591545105,
+      "learning_rate": 6.788966682600412e-06,
+      "loss": 0.1086,
+      "step": 1125
+    },
+    {
+      "epoch": 3.520249221183801,
+      "grad_norm": 1.0332609415054321,
+      "learning_rate": 6.6600201046868944e-06,
+      "loss": 0.1187,
+      "step": 1130
+    },
+    {
+      "epoch": 3.5358255451713396,
+      "grad_norm": 1.0396648645401,
+      "learning_rate": 6.5319595211775876e-06,
+      "loss": 0.1383,
+      "step": 1135
+    },
+    {
+      "epoch": 3.5514018691588785,
+      "grad_norm": 0.9917812347412109,
+      "learning_rate": 6.404798536536991e-06,
+      "loss": 0.1187,
+      "step": 1140
+    },
+    {
+      "epoch": 3.5669781931464173,
+      "grad_norm": 0.9046400785446167,
+      "learning_rate": 6.278550659661091e-06,
+      "loss": 0.1222,
+      "step": 1145
+    },
+    {
+      "epoch": 3.582554517133956,
+      "grad_norm": 0.8576000332832336,
+      "learning_rate": 6.153229302442209e-06,
+      "loss": 0.1227,
+      "step": 1150
+    },
+    {
+      "epoch": 3.5981308411214954,
+      "grad_norm": 0.9951794147491455,
+      "learning_rate": 6.028847778344249e-06,
+      "loss": 0.1283,
+      "step": 1155
+    },
+    {
+      "epoch": 3.6137071651090342,
+      "grad_norm": 1.1326957941055298,
+      "learning_rate": 5.905419300988311e-06,
+      "loss": 0.1162,
+      "step": 1160
+    },
+    {
+      "epoch": 3.629283489096573,
+      "grad_norm": 0.9401041865348816,
+      "learning_rate": 5.782956982748966e-06,
+      "loss": 0.1177,
+      "step": 1165
+    },
+    {
+      "epoch": 3.6448598130841123,
+      "grad_norm": 1.0154671669006348,
+      "learning_rate": 5.661473833361247e-06,
+      "loss": 0.1236,
+      "step": 1170
+    },
+    {
+      "epoch": 3.660436137071651,
+      "grad_norm": 0.9394180178642273,
+      "learning_rate": 5.540982758538586e-06,
+      "loss": 0.1123,
+      "step": 1175
+    },
+    {
+      "epoch": 3.67601246105919,
+      "grad_norm": 0.9814959764480591,
+      "learning_rate": 5.4214965586017814e-06,
+      "loss": 0.122,
+      "step": 1180
+    },
+    {
+      "epoch": 3.691588785046729,
+      "grad_norm": 0.935127317905426,
+      "learning_rate": 5.30302792711915e-06,
+      "loss": 0.1144,
+      "step": 1185
+    },
+    {
+      "epoch": 3.7071651090342677,
+      "grad_norm": 1.111989140510559,
+      "learning_rate": 5.185589449558016e-06,
+      "loss": 0.1026,
+      "step": 1190
+    },
+    {
+      "epoch": 3.722741433021807,
+      "grad_norm": 0.7687360048294067,
+      "learning_rate": 5.069193601947736e-06,
+      "loss": 0.1044,
+      "step": 1195
+    },
+    {
+      "epoch": 3.7383177570093458,
+      "grad_norm": 1.0546218156814575,
+      "learning_rate": 4.953852749554294e-06,
+      "loss": 0.1184,
+      "step": 1200
+    },
+    {
+      "epoch": 3.7538940809968846,
+      "grad_norm": 1.0001137256622314,
+      "learning_rate": 4.839579145566665e-06,
+      "loss": 0.1181,
+      "step": 1205
+    },
+    {
+      "epoch": 3.769470404984424,
+      "grad_norm": 0.9862017035484314,
+      "learning_rate": 4.726384929795133e-06,
+      "loss": 0.1225,
+      "step": 1210
+    },
+    {
+      "epoch": 3.7850467289719627,
+      "grad_norm": 1.0694918632507324,
+      "learning_rate": 4.614282127381613e-06,
+      "loss": 0.1059,
+      "step": 1215
+    },
+    {
+      "epoch": 3.8006230529595015,
+      "grad_norm": 0.9774491190910339,
+      "learning_rate": 4.503282647522161e-06,
+      "loss": 0.116,
+      "step": 1220
+    },
+    {
+      "epoch": 3.8161993769470404,
+      "grad_norm": 1.0102533102035522,
+      "learning_rate": 4.393398282201788e-06,
+      "loss": 0.1106,
+      "step": 1225
+    },
+    {
+      "epoch": 3.831775700934579,
+      "grad_norm": 1.047069787979126,
+      "learning_rate": 4.284640704941781e-06,
+      "loss": 0.1162,
+      "step": 1230
+    },
+    {
+      "epoch": 3.8473520249221185,
+      "grad_norm": 0.905773401260376,
+      "learning_rate": 4.177021469559548e-06,
+      "loss": 0.0953,
+      "step": 1235
+    },
+    {
+      "epoch": 3.8629283489096573,
+      "grad_norm": 0.944668710231781,
+      "learning_rate": 4.070552008941193e-06,
+      "loss": 0.1059,
+      "step": 1240
+    },
+    {
+      "epoch": 3.878504672897196,
+      "grad_norm": 0.9875009059906006,
+      "learning_rate": 3.965243633826974e-06,
+      "loss": 0.1088,
+      "step": 1245
+    },
+    {
+      "epoch": 3.8940809968847354,
+      "grad_norm": 0.8600956201553345,
+      "learning_rate": 3.861107531609701e-06,
+      "loss": 0.1009,
+      "step": 1250
+    },
+    {
+      "epoch": 3.9096573208722742,
+      "grad_norm": 0.9825694561004639,
+      "learning_rate": 3.758154765146241e-06,
+      "loss": 0.1221,
+      "step": 1255
+    },
+    {
+      "epoch": 3.925233644859813,
+      "grad_norm": 0.885953962802887,
+      "learning_rate": 3.6563962715822534e-06,
+      "loss": 0.1149,
+      "step": 1260
+    },
+    {
+      "epoch": 3.940809968847352,
+      "grad_norm": 0.919476330280304,
+      "learning_rate": 3.5558428611903087e-06,
+      "loss": 0.1125,
+      "step": 1265
+    },
+    {
+      "epoch": 3.9563862928348907,
+      "grad_norm": 1.0793930292129517,
+      "learning_rate": 3.4565052162214573e-06,
+      "loss": 0.0992,
+      "step": 1270
+    },
+    {
+      "epoch": 3.97196261682243,
+      "grad_norm": 1.032047986984253,
+      "learning_rate": 3.3583938897703787e-06,
+      "loss": 0.1138,
+      "step": 1275
+    },
+    {
+      "epoch": 3.987538940809969,
+      "grad_norm": 0.9010823965072632,
+      "learning_rate": 3.2615193046543175e-06,
+      "loss": 0.1011,
+      "step": 1280
+    },
+    {
+      "epoch": 4.003115264797508,
+      "grad_norm": 0.7412053346633911,
+      "learning_rate": 3.1658917523057977e-06,
+      "loss": 0.1034,
+      "step": 1285
+    },
+    {
+      "epoch": 4.018691588785047,
+      "grad_norm": 0.7402732372283936,
+      "learning_rate": 3.071521391679323e-06,
+      "loss": 0.0824,
+      "step": 1290
+    },
+    {
+      "epoch": 4.034267912772585,
+      "grad_norm": 0.8276841640472412,
+      "learning_rate": 2.978418248172121e-06,
+      "loss": 0.0855,
+      "step": 1295
+    },
+    {
+      "epoch": 4.049844236760125,
+      "grad_norm": 0.8681264519691467,
+      "learning_rate": 2.8865922125591435e-06,
+      "loss": 0.0926,
+      "step": 1300
+    },
+    {
+      "epoch": 4.065420560747664,
+      "grad_norm": 0.7654516100883484,
+      "learning_rate": 2.796053039942285e-06,
+      "loss": 0.086,
+      "step": 1305
+    },
+    {
+      "epoch": 4.080996884735202,
+      "grad_norm": 0.8676243424415588,
+      "learning_rate": 2.706810348714078e-06,
+      "loss": 0.0869,
+      "step": 1310
+    },
+    {
+      "epoch": 4.0965732087227416,
+      "grad_norm": 0.7806140780448914,
+      "learning_rate": 2.618873619535859e-06,
+      "loss": 0.0848,
+      "step": 1315
+    },
+    {
+      "epoch": 4.11214953271028,
+      "grad_norm": 0.8019266724586487,
+      "learning_rate": 2.5322521943306396e-06,
+      "loss": 0.084,
+      "step": 1320
+    },
+    {
+      "epoch": 4.127725856697819,
+      "grad_norm": 0.9295703768730164,
+      "learning_rate": 2.4469552752906243e-06,
+      "loss": 0.0793,
+      "step": 1325
+    },
+    {
+      "epoch": 4.1433021806853585,
+      "grad_norm": 0.7183976173400879,
+      "learning_rate": 2.3629919238996393e-06,
+      "loss": 0.0954,
+      "step": 1330
+    },
+    {
+      "epoch": 4.158878504672897,
+      "grad_norm": 0.7512747645378113,
+      "learning_rate": 2.280371059970495e-06,
+      "loss": 0.0864,
+      "step": 1335
+    },
+    {
+      "epoch": 4.174454828660436,
+      "grad_norm": 0.7748674154281616,
+      "learning_rate": 2.1991014606973864e-06,
+      "loss": 0.0937,
+      "step": 1340
+    },
+    {
+      "epoch": 4.190031152647975,
+      "grad_norm": 0.9455962181091309,
+      "learning_rate": 2.119191759723453e-06,
+      "loss": 0.0816,
+      "step": 1345
+    },
+    {
+      "epoch": 4.205607476635514,
+      "grad_norm": 0.8093606233596802,
+      "learning_rate": 2.040650446223576e-06,
+      "loss": 0.0833,
+      "step": 1350
+    },
+    {
+      "epoch": 4.221183800623053,
+      "grad_norm": 0.9539604187011719,
+      "learning_rate": 1.963485864002559e-06,
+      "loss": 0.0784,
+      "step": 1355
+    },
+    {
+      "epoch": 4.2367601246105915,
+      "grad_norm": 0.7200955152511597,
+      "learning_rate": 1.8877062106087107e-06,
+      "loss": 0.0856,
+      "step": 1360
+    },
+    {
+      "epoch": 4.252336448598131,
+      "grad_norm": 0.8740413188934326,
+      "learning_rate": 1.8133195364629713e-06,
+      "loss": 0.0904,
+      "step": 1365
+    },
+    {
+      "epoch": 4.26791277258567,
+      "grad_norm": 0.785948634147644,
+      "learning_rate": 1.7403337440037031e-06,
+      "loss": 0.0865,
+      "step": 1370
+    },
+    {
+      "epoch": 4.283489096573208,
+      "grad_norm": 0.8199154734611511,
+      "learning_rate": 1.6687565868471654e-06,
+      "loss": 0.0932,
+      "step": 1375
+    },
+    {
+      "epoch": 4.299065420560748,
+      "grad_norm": 0.7736993432044983,
+      "learning_rate": 1.5985956689638149e-06,
+      "loss": 0.0824,
+      "step": 1380
+    },
+    {
+      "epoch": 4.314641744548287,
+      "grad_norm": 0.7168201804161072,
+      "learning_rate": 1.5298584438704865e-06,
+      "loss": 0.0762,
+      "step": 1385
+    },
+    {
+      "epoch": 4.330218068535825,
+      "grad_norm": 0.806542158126831,
+      "learning_rate": 1.462552213838605e-06,
+      "loss": 0.0886,
+      "step": 1390
+    },
+    {
+      "epoch": 4.345794392523365,
+      "grad_norm": 0.818294882774353,
+      "learning_rate": 1.396684129118402e-06,
+      "loss": 0.0794,
+      "step": 1395
+    },
+    {
+      "epoch": 4.361370716510903,
+      "grad_norm": 0.7591167092323303,
+      "learning_rate": 1.332261187179316e-06,
+      "loss": 0.0738,
+      "step": 1400
+    },
+    {
+      "epoch": 4.376947040498442,
+      "grad_norm": 0.8816602230072021,
+      "learning_rate": 1.2692902319666316e-06,
+      "loss": 0.0901,
+      "step": 1405
+    },
+    {
+      "epoch": 4.392523364485982,
+      "grad_norm": 0.8728139400482178,
+      "learning_rate": 1.2077779531744071e-06,
+      "loss": 0.0782,
+      "step": 1410
+    },
+    {
+      "epoch": 4.40809968847352,
+      "grad_norm": 0.7949065566062927,
+      "learning_rate": 1.1477308855348013e-06,
+      "loss": 0.0813,
+      "step": 1415
+    },
+    {
+      "epoch": 4.423676012461059,
+      "grad_norm": 0.9780331254005432,
+      "learning_rate": 1.0891554081238448e-06,
+      "loss": 0.0932,
+      "step": 1420
+    },
+    {
+      "epoch": 4.4392523364485985,
+      "grad_norm": 0.8240417242050171,
+      "learning_rate": 1.0320577436837908e-06,
+      "loss": 0.0852,
+      "step": 1425
+    },
+    {
+      "epoch": 4.454828660436137,
+      "grad_norm": 0.8341586589813232,
+      "learning_rate": 9.764439579620155e-07,
+      "loss": 0.0874,
+      "step": 1430
+    },
+    {
+      "epoch": 4.470404984423676,
+      "grad_norm": 0.93865966796875,
+      "learning_rate": 9.223199590666364e-07,
+      "loss": 0.0909,
+      "step": 1435
+    },
+    {
+      "epoch": 4.485981308411215,
+      "grad_norm": 0.9722694158554077,
+      "learning_rate": 8.69691496838872e-07,
+      "loss": 0.1025,
+      "step": 1440
+    },
+    {
+      "epoch": 4.501557632398754,
+      "grad_norm": 0.7852746248245239,
+      "learning_rate": 8.185641622422074e-07,
+      "loss": 0.0769,
+      "step": 1445
+    },
+    {
+      "epoch": 4.517133956386293,
+      "grad_norm": 0.8302127122879028,
+      "learning_rate": 7.689433867684359e-07,
+      "loss": 0.0817,
+      "step": 1450
+    },
+    {
+      "epoch": 4.5327102803738315,
+      "grad_norm": 0.8076728582382202,
+      "learning_rate": 7.208344418606411e-07,
+      "loss": 0.082,
+      "step": 1455
+    },
+    {
+      "epoch": 4.548286604361371,
+      "grad_norm": 0.773902952671051,
+      "learning_rate": 6.742424383532119e-07,
+      "loss": 0.0729,
+      "step": 1460
+    },
+    {
+      "epoch": 4.563862928348909,
+      "grad_norm": 0.6929298639297485,
+      "learning_rate": 6.29172325928869e-07,
+      "loss": 0.076,
+      "step": 1465
+    },
+    {
+      "epoch": 4.579439252336448,
+      "grad_norm": 0.699554443359375,
+      "learning_rate": 5.856288925928477e-07,
+      "loss": 0.0756,
+      "step": 1470
+    },
+    {
+      "epoch": 4.595015576323988,
+      "grad_norm": 0.7829858064651489,
+      "learning_rate": 5.436167641642426e-07,
+      "loss": 0.07,
+      "step": 1475
+    },
+    {
+      "epoch": 4.610591900311526,
+      "grad_norm": 0.8051700592041016,
+      "learning_rate": 5.031404037845949e-07,
+      "loss": 0.0827,
+      "step": 1480
+    },
+    {
+      "epoch": 4.626168224299065,
+      "grad_norm": 0.7270315885543823,
+      "learning_rate": 4.642041114437423e-07,
+      "loss": 0.0784,
+      "step": 1485
+    },
+    {
+      "epoch": 4.641744548286605,
+      "grad_norm": 0.7566366791725159,
+      "learning_rate": 4.2681202352301386e-07,
+      "loss": 0.0776,
+      "step": 1490
+    },
+    {
+      "epoch": 4.657320872274143,
+      "grad_norm": 0.7550317645072937,
+      "learning_rate": 3.9096811235580677e-07,
+      "loss": 0.0823,
+      "step": 1495
+    },
+    {
+      "epoch": 4.672897196261682,
+      "grad_norm": 0.8521310687065125,
+      "learning_rate": 3.5667618580558224e-07,
+      "loss": 0.0735,
+      "step": 1500
+    },
+    {
+      "epoch": 4.688473520249222,
+      "grad_norm": 0.7728444933891296,
+      "learning_rate": 3.2393988686134414e-07,
+      "loss": 0.0857,
+      "step": 1505
+    },
+    {
+      "epoch": 4.70404984423676,
+      "grad_norm": 0.7044500112533569,
+      "learning_rate": 2.9276269325062e-07,
+      "loss": 0.0873,
+      "step": 1510
+    },
+    {
+      "epoch": 4.719626168224299,
+      "grad_norm": 0.8075025081634521,
+      "learning_rate": 2.631479170700135e-07,
+      "loss": 0.0815,
+      "step": 1515
+    },
+    {
+      "epoch": 4.735202492211838,
+      "grad_norm": 0.7614153623580933,
+      "learning_rate": 2.3509870443334247e-07,
+      "loss": 0.0682,
+      "step": 1520
+    },
+    {
+      "epoch": 4.750778816199377,
+      "grad_norm": 0.652677595615387,
+      "learning_rate": 2.0861803513740785e-07,
+      "loss": 0.0812,
+      "step": 1525
+    },
+    {
+      "epoch": 4.766355140186916,
+      "grad_norm": 0.9125698208808899,
+      "learning_rate": 1.8370872234544533e-07,
+      "loss": 0.0818,
+      "step": 1530
+    },
+    {
+      "epoch": 4.781931464174455,
+      "grad_norm": 0.7782279849052429,
+      "learning_rate": 1.6037341228826608e-07,
+      "loss": 0.0816,
+      "step": 1535
+    },
+    {
+      "epoch": 4.797507788161994,
+      "grad_norm": 0.7868243455886841,
+      "learning_rate": 1.3861458398312988e-07,
+      "loss": 0.0752,
+      "step": 1540
+    },
+    {
+      "epoch": 4.813084112149532,
+      "grad_norm": 0.8168080449104309,
+      "learning_rate": 1.1843454897039752e-07,
+      "loss": 0.0809,
+      "step": 1545
+    },
+    {
+      "epoch": 4.8286604361370715,
+      "grad_norm": 0.7977619767189026,
+      "learning_rate": 9.983545106795722e-08,
+      "loss": 0.086,
+      "step": 1550
+    },
+    {
+      "epoch": 4.844236760124611,
+      "grad_norm": 0.7246932983398438,
+      "learning_rate": 8.281926614348511e-08,
+      "loss": 0.0789,
+      "step": 1555
+    },
+    {
+      "epoch": 4.859813084112149,
+      "grad_norm": 0.7492623925209045,
+      "learning_rate": 6.738780190452487e-08,
+      "loss": 0.0801,
+      "step": 1560
+    },
+    {
+      "epoch": 4.8753894080996885,
+      "grad_norm": 0.8362622857093811,
+      "learning_rate": 5.3542697706467913e-08,
+      "loss": 0.0815,
+      "step": 1565
+    },
+    {
+      "epoch": 4.890965732087228,
+      "grad_norm": 0.7539082765579224,
+      "learning_rate": 4.1285424378379456e-08,
+      "loss": 0.0992,
+      "step": 1570
+    },
+    {
+      "epoch": 4.906542056074766,
+      "grad_norm": 0.7643052339553833,
+      "learning_rate": 3.061728406675512e-08,
+      "loss": 0.0703,
+      "step": 1575
+    },
+    {
+      "epoch": 4.922118380062305,
+      "grad_norm": 0.770291268825531,
+      "learning_rate": 2.1539410097179968e-08,
+      "loss": 0.0798,
+      "step": 1580
+    },
+    {
+      "epoch": 4.937694704049845,
+      "grad_norm": 0.831664502620697,
+      "learning_rate": 1.4052766853938082e-08,
+      "loss": 0.0878,
+      "step": 1585
+    },
+    {
+      "epoch": 4.953271028037383,
+      "grad_norm": 0.7236193418502808,
+      "learning_rate": 8.158149677561233e-09,
+      "loss": 0.0799,
+      "step": 1590
+    },
+    {
+      "epoch": 4.968847352024922,
+      "grad_norm": 0.8014712333679199,
+      "learning_rate": 3.856184780328121e-09,
+      "loss": 0.0847,
+      "step": 1595
+    },
+    {
+      "epoch": 4.984423676012461,
+      "grad_norm": 0.7500919699668884,
+      "learning_rate": 1.1473291797459152e-09,
+      "loss": 0.078,
+      "step": 1600
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.7876099348068237,
+      "learning_rate": 3.1870650000764834e-11,
+      "loss": 0.0899,
+      "step": 1605
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1605,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 2000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.2992611817491005e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

33_128_e5_3e-5/checkpoint-1605/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:da61cbe608501d2834de43cd31aa18db3ae688ca81d76fe4c7a0a77149e96662
+size 7736

33_128_e5_3e-5/checkpoint-1605/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)

33_128_e5_3e-5/checkpoint-321/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

33_128_e5_3e-5/checkpoint-321/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": [
+    "o_proj",
+    "k_proj",
+    "gate_proj",
+    "q_proj",
+    "v_proj",
+    "down_proj",
+    "up_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

33_128_e5_3e-5/checkpoint-321/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:df754a8f6ba4af2149d6a4e1feeb12785fee7588b95c957a06c0cfeac383be0c
+size 791751704

33_128_e5_3e-5/checkpoint-321/latest

@@ -0,0 +1 @@
+global_step321

33_128_e5_3e-5/checkpoint-321/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7902fbc8c0481e8e42790afc94df0080d19aa7f6758ecfaf99afcbf992876e4e
+size 15920

33_128_e5_3e-5/checkpoint-321/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2da33ac630135ec10081fa1392d19a2b60f5a89d03ea479e773050bc6d8e5211
+size 15920

33_128_e5_3e-5/checkpoint-321/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5b5a55ff06865397ebdc8738e53c0cdd142891033ec79886c285597fccb420a6
+size 15920