Upload folder using huggingface_hub

23_128_e5_3e-5/checkpoint-1198/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

23_128_e5_3e-5/checkpoint-1198/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",
+    "gate_proj",
+    "v_proj",
+    "k_proj",
+    "up_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

23_128_e5_3e-5/checkpoint-1198/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3dd4ea345d6e8fd6d89b64de1782f49a3823ca63b15972dafe3008c268c0121f
+size 791751704

23_128_e5_3e-5/checkpoint-1198/latest

@@ -0,0 +1 @@
+global_step1198

23_128_e5_3e-5/checkpoint-1198/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:949a258c7d2d0c4d9b209312740319b0b55ac6f1f9e4ab0b37f460abc7dddfce
+size 15920

23_128_e5_3e-5/checkpoint-1198/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c31646c1443aa1b1ae2dc69602eba155c3af9ffb8fa60bfe89a12fe06d9b3406
+size 15920

23_128_e5_3e-5/checkpoint-1198/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e50ff21633a9976904f216b0c442e6fa71f7b90603c443572d6816a4fd908cee
+size 15920

23_128_e5_3e-5/checkpoint-1198/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7e52be2905b96d7686bdfdef29374e302a928850a8ae458b733803fc6e2d3bf2
+size 15920

23_128_e5_3e-5/checkpoint-1198/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:080919beed1f5f930781cbcc16fb489a1e216817ca4a648ff83629d0e7a5b4e0
+size 15920

23_128_e5_3e-5/checkpoint-1198/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7ea9935785ed55fa296b951c8ad04f1deb6338042e6905328562ec1b2ea1a54c
+size 15920

23_128_e5_3e-5/checkpoint-1198/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e091582a98d6cbabdd006ccedc298a9c9126507fff2db918f0bcbdf1b8bb5f76
+size 15920

23_128_e5_3e-5/checkpoint-1198/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a5d06667e8e65d245ea234b6046364f4797b68e0dd8ce9f46bf28be1edca261d
+size 15920

23_128_e5_3e-5/checkpoint-1198/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7225fdea565f5eefbf13c4b5b5b32bf836437b017390f96d9137a22570261f75
+size 1064

23_128_e5_3e-5/checkpoint-1198/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
+  }
+}

23_128_e5_3e-5/checkpoint-1198/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
+}

23_128_e5_3e-5/checkpoint-1198/trainer_state.json

@@ -0,0 +1,1707 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 1198,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.008347245409015025,
+      "grad_norm": 0.9250032305717468,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 1.3897,
+      "step": 5
+    },
+    {
+      "epoch": 0.01669449081803005,
+      "grad_norm": 1.0577523708343506,
+      "learning_rate": 1.8e-06,
+      "loss": 1.3743,
+      "step": 10
+    },
+    {
+      "epoch": 0.025041736227045076,
+      "grad_norm": 0.7588217258453369,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 1.3647,
+      "step": 15
+    },
+    {
+      "epoch": 0.0333889816360601,
+      "grad_norm": 0.7027095556259155,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 1.3301,
+      "step": 20
+    },
+    {
+      "epoch": 0.041736227045075125,
+      "grad_norm": 0.47005781531333923,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.2668,
+      "step": 25
+    },
+    {
+      "epoch": 0.05008347245409015,
+      "grad_norm": 0.49903222918510437,
+      "learning_rate": 5.8e-06,
+      "loss": 1.3468,
+      "step": 30
+    },
+    {
+      "epoch": 0.05843071786310518,
+      "grad_norm": 0.541542649269104,
+      "learning_rate": 6.8e-06,
+      "loss": 1.2821,
+      "step": 35
+    },
+    {
+      "epoch": 0.0667779632721202,
+      "grad_norm": 0.5814635157585144,
+      "learning_rate": 7.8e-06,
+      "loss": 1.2576,
+      "step": 40
+    },
+    {
+      "epoch": 0.07512520868113523,
+      "grad_norm": 0.4928892254829407,
+      "learning_rate": 8.8e-06,
+      "loss": 1.2328,
+      "step": 45
+    },
+    {
+      "epoch": 0.08347245409015025,
+      "grad_norm": 0.44546306133270264,
+      "learning_rate": 9.8e-06,
+      "loss": 1.2874,
+      "step": 50
+    },
+    {
+      "epoch": 0.09181969949916527,
+      "grad_norm": 0.41543567180633545,
+      "learning_rate": 1.08e-05,
+      "loss": 1.2451,
+      "step": 55
+    },
+    {
+      "epoch": 0.1001669449081803,
+      "grad_norm": 0.4427853524684906,
+      "learning_rate": 1.1799999999999999e-05,
+      "loss": 1.2629,
+      "step": 60
+    },
+    {
+      "epoch": 0.10851419031719532,
+      "grad_norm": 0.36455604434013367,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 1.2237,
+      "step": 65
+    },
+    {
+      "epoch": 0.11686143572621036,
+      "grad_norm": 0.4628913998603821,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 1.2426,
+      "step": 70
+    },
+    {
+      "epoch": 0.12520868113522537,
+      "grad_norm": 0.4718121588230133,
+      "learning_rate": 1.48e-05,
+      "loss": 1.2996,
+      "step": 75
+    },
+    {
+      "epoch": 0.1335559265442404,
+      "grad_norm": 0.4145875573158264,
+      "learning_rate": 1.5799999999999998e-05,
+      "loss": 1.2439,
+      "step": 80
+    },
+    {
+      "epoch": 0.1419031719532554,
+      "grad_norm": 0.46866002678871155,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 1.2307,
+      "step": 85
+    },
+    {
+      "epoch": 0.15025041736227046,
+      "grad_norm": 0.4653532803058624,
+      "learning_rate": 1.7800000000000002e-05,
+      "loss": 1.2022,
+      "step": 90
+    },
+    {
+      "epoch": 0.15859766277128548,
+      "grad_norm": 0.5028699636459351,
+      "learning_rate": 1.8800000000000003e-05,
+      "loss": 1.2042,
+      "step": 95
+    },
+    {
+      "epoch": 0.1669449081803005,
+      "grad_norm": 0.4724132716655731,
+      "learning_rate": 1.98e-05,
+      "loss": 1.1189,
+      "step": 100
+    },
+    {
+      "epoch": 0.17529215358931552,
+      "grad_norm": 0.49735012650489807,
+      "learning_rate": 2.08e-05,
+      "loss": 1.2435,
+      "step": 105
+    },
+    {
+      "epoch": 0.18363939899833054,
+      "grad_norm": 0.5972133874893188,
+      "learning_rate": 2.18e-05,
+      "loss": 1.1465,
+      "step": 110
+    },
+    {
+      "epoch": 0.19198664440734559,
+      "grad_norm": 0.47499340772628784,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 1.207,
+      "step": 115
+    },
+    {
+      "epoch": 0.2003338898163606,
+      "grad_norm": 0.4662191867828369,
+      "learning_rate": 2.38e-05,
+      "loss": 1.18,
+      "step": 120
+    },
+    {
+      "epoch": 0.20868113522537562,
+      "grad_norm": 0.8468295931816101,
+      "learning_rate": 2.48e-05,
+      "loss": 1.1867,
+      "step": 125
+    },
+    {
+      "epoch": 0.21702838063439064,
+      "grad_norm": 0.528782069683075,
+      "learning_rate": 2.58e-05,
+      "loss": 1.1633,
+      "step": 130
+    },
+    {
+      "epoch": 0.22537562604340566,
+      "grad_norm": 0.5306044220924377,
+      "learning_rate": 2.68e-05,
+      "loss": 1.1775,
+      "step": 135
+    },
+    {
+      "epoch": 0.2337228714524207,
+      "grad_norm": 0.5375636219978333,
+      "learning_rate": 2.78e-05,
+      "loss": 1.188,
+      "step": 140
+    },
+    {
+      "epoch": 0.24207011686143573,
+      "grad_norm": 0.5145397782325745,
+      "learning_rate": 2.88e-05,
+      "loss": 1.1268,
+      "step": 145
+    },
+    {
+      "epoch": 0.25041736227045075,
+      "grad_norm": 0.5362802147865295,
+      "learning_rate": 2.98e-05,
+      "loss": 1.1542,
+      "step": 150
+    },
+    {
+      "epoch": 0.2587646076794658,
+      "grad_norm": 0.5439884662628174,
+      "learning_rate": 2.999985367593965e-05,
+      "loss": 1.13,
+      "step": 155
+    },
+    {
+      "epoch": 0.2671118530884808,
+      "grad_norm": 0.487518310546875,
+      "learning_rate": 2.999925923933716e-05,
+      "loss": 1.109,
+      "step": 160
+    },
+    {
+      "epoch": 0.27545909849749584,
+      "grad_norm": 0.5048381686210632,
+      "learning_rate": 2.999820756304571e-05,
+      "loss": 1.1214,
+      "step": 165
+    },
+    {
+      "epoch": 0.2838063439065108,
+      "grad_norm": 0.5778033137321472,
+      "learning_rate": 2.999669867912476e-05,
+      "loss": 1.1179,
+      "step": 170
+    },
+    {
+      "epoch": 0.2921535893155259,
+      "grad_norm": 0.5533081293106079,
+      "learning_rate": 2.999473263357136e-05,
+      "loss": 1.0759,
+      "step": 175
+    },
+    {
+      "epoch": 0.3005008347245409,
+      "grad_norm": 0.6201685070991516,
+      "learning_rate": 2.999230948631874e-05,
+      "loss": 1.058,
+      "step": 180
+    },
+    {
+      "epoch": 0.3088480801335559,
+      "grad_norm": 0.5643416047096252,
+      "learning_rate": 2.9989429311234485e-05,
+      "loss": 1.0972,
+      "step": 185
+    },
+    {
+      "epoch": 0.31719532554257096,
+      "grad_norm": 0.5617562532424927,
+      "learning_rate": 2.9986092196118284e-05,
+      "loss": 1.0984,
+      "step": 190
+    },
+    {
+      "epoch": 0.32554257095158595,
+      "grad_norm": 0.5762996673583984,
+      "learning_rate": 2.998229824269927e-05,
+      "loss": 1.0559,
+      "step": 195
+    },
+    {
+      "epoch": 0.333889816360601,
+      "grad_norm": 0.5630249977111816,
+      "learning_rate": 2.9978047566632887e-05,
+      "loss": 1.0928,
+      "step": 200
+    },
+    {
+      "epoch": 0.34223706176961605,
+      "grad_norm": 0.5413353443145752,
+      "learning_rate": 2.9973340297497384e-05,
+      "loss": 1.0944,
+      "step": 205
+    },
+    {
+      "epoch": 0.35058430717863104,
+      "grad_norm": 0.6307699680328369,
+      "learning_rate": 2.9968176578789874e-05,
+      "loss": 1.1564,
+      "step": 210
+    },
+    {
+      "epoch": 0.3589315525876461,
+      "grad_norm": 0.7229161858558655,
+      "learning_rate": 2.9962556567921933e-05,
+      "loss": 1.0633,
+      "step": 215
+    },
+    {
+      "epoch": 0.3672787979966611,
+      "grad_norm": 0.6549956798553467,
+      "learning_rate": 2.9956480436214828e-05,
+      "loss": 1.0555,
+      "step": 220
+    },
+    {
+      "epoch": 0.3756260434056761,
+      "grad_norm": 0.5410817265510559,
+      "learning_rate": 2.994994836889427e-05,
+      "loss": 1.0631,
+      "step": 225
+    },
+    {
+      "epoch": 0.38397328881469117,
+      "grad_norm": 0.6632435917854309,
+      "learning_rate": 2.9942960565084802e-05,
+      "loss": 1.0573,
+      "step": 230
+    },
+    {
+      "epoch": 0.39232053422370616,
+      "grad_norm": 0.6554168462753296,
+      "learning_rate": 2.9935517237803675e-05,
+      "loss": 1.0133,
+      "step": 235
+    },
+    {
+      "epoch": 0.4006677796327212,
+      "grad_norm": 0.6186609864234924,
+      "learning_rate": 2.9927618613954417e-05,
+      "loss": 1.0114,
+      "step": 240
+    },
+    {
+      "epoch": 0.4090150250417362,
+      "grad_norm": 0.6478000283241272,
+      "learning_rate": 2.9919264934319874e-05,
+      "loss": 1.0423,
+      "step": 245
+    },
+    {
+      "epoch": 0.41736227045075125,
+      "grad_norm": 0.6168914437294006,
+      "learning_rate": 2.9910456453554868e-05,
+      "loss": 1.0181,
+      "step": 250
+    },
+    {
+      "epoch": 0.4257095158597663,
+      "grad_norm": 0.8668388724327087,
+      "learning_rate": 2.990119344017847e-05,
+      "loss": 1.0286,
+      "step": 255
+    },
+    {
+      "epoch": 0.4340567612687813,
+      "grad_norm": 0.6979084610939026,
+      "learning_rate": 2.9891476176565784e-05,
+      "loss": 1.0031,
+      "step": 260
+    },
+    {
+      "epoch": 0.44240400667779634,
+      "grad_norm": 0.7331295609474182,
+      "learning_rate": 2.9881304958939334e-05,
+      "loss": 1.0001,
+      "step": 265
+    },
+    {
+      "epoch": 0.4507512520868113,
+      "grad_norm": 0.8051875829696655,
+      "learning_rate": 2.987068009736007e-05,
+      "loss": 0.9713,
+      "step": 270
+    },
+    {
+      "epoch": 0.4590984974958264,
+      "grad_norm": 0.7588741183280945,
+      "learning_rate": 2.985960191571787e-05,
+      "loss": 0.9831,
+      "step": 275
+    },
+    {
+      "epoch": 0.4674457429048414,
+      "grad_norm": 0.7282049059867859,
+      "learning_rate": 2.9848070751721704e-05,
+      "loss": 0.9665,
+      "step": 280
+    },
+    {
+      "epoch": 0.4757929883138564,
+      "grad_norm": 0.7469468116760254,
+      "learning_rate": 2.9836086956889328e-05,
+      "loss": 1.004,
+      "step": 285
+    },
+    {
+      "epoch": 0.48414023372287146,
+      "grad_norm": 0.7200749516487122,
+      "learning_rate": 2.9823650896536545e-05,
+      "loss": 0.9635,
+      "step": 290
+    },
+    {
+      "epoch": 0.49248747913188645,
+      "grad_norm": 0.6923590898513794,
+      "learning_rate": 2.9810762949766113e-05,
+      "loss": 0.9508,
+      "step": 295
+    },
+    {
+      "epoch": 0.5008347245409015,
+      "grad_norm": 0.738401472568512,
+      "learning_rate": 2.979742350945614e-05,
+      "loss": 0.9086,
+      "step": 300
+    },
+    {
+      "epoch": 0.5091819699499165,
+      "grad_norm": 0.7354897856712341,
+      "learning_rate": 2.9783632982248152e-05,
+      "loss": 0.9975,
+      "step": 305
+    },
+    {
+      "epoch": 0.5175292153589316,
+      "grad_norm": 0.8435232639312744,
+      "learning_rate": 2.9769391788534662e-05,
+      "loss": 0.9406,
+      "step": 310
+    },
+    {
+      "epoch": 0.5258764607679466,
+      "grad_norm": 1.0065906047821045,
+      "learning_rate": 2.9754700362446382e-05,
+      "loss": 0.922,
+      "step": 315
+    },
+    {
+      "epoch": 0.5342237061769616,
+      "grad_norm": 0.7278281450271606,
+      "learning_rate": 2.9739559151838964e-05,
+      "loss": 0.9792,
+      "step": 320
+    },
+    {
+      "epoch": 0.5425709515859767,
+      "grad_norm": 0.8112268447875977,
+      "learning_rate": 2.972396861827936e-05,
+      "loss": 0.9774,
+      "step": 325
+    },
+    {
+      "epoch": 0.5509181969949917,
+      "grad_norm": 0.7046104073524475,
+      "learning_rate": 2.9707929237031765e-05,
+      "loss": 0.9139,
+      "step": 330
+    },
+    {
+      "epoch": 0.5592654424040067,
+      "grad_norm": 0.7925729751586914,
+      "learning_rate": 2.9691441497043098e-05,
+      "loss": 0.8911,
+      "step": 335
+    },
+    {
+      "epoch": 0.5676126878130217,
+      "grad_norm": 0.7873656749725342,
+      "learning_rate": 2.9674505900928118e-05,
+      "loss": 0.9207,
+      "step": 340
+    },
+    {
+      "epoch": 0.5759599332220368,
+      "grad_norm": 0.8177038431167603,
+      "learning_rate": 2.9657122964954107e-05,
+      "loss": 0.9357,
+      "step": 345
+    },
+    {
+      "epoch": 0.5843071786310517,
+      "grad_norm": 0.7847334742546082,
+      "learning_rate": 2.9639293219025116e-05,
+      "loss": 0.8949,
+      "step": 350
+    },
+    {
+      "epoch": 0.5926544240400667,
+      "grad_norm": 0.8220208287239075,
+      "learning_rate": 2.9621017206665807e-05,
+      "loss": 0.9603,
+      "step": 355
+    },
+    {
+      "epoch": 0.6010016694490818,
+      "grad_norm": 0.8926270008087158,
+      "learning_rate": 2.9602295485004922e-05,
+      "loss": 0.9297,
+      "step": 360
+    },
+    {
+      "epoch": 0.6093489148580968,
+      "grad_norm": 0.834158718585968,
+      "learning_rate": 2.9583128624758252e-05,
+      "loss": 0.9139,
+      "step": 365
+    },
+    {
+      "epoch": 0.6176961602671118,
+      "grad_norm": 0.8102491497993469,
+      "learning_rate": 2.9563517210211266e-05,
+      "loss": 0.8262,
+      "step": 370
+    },
+    {
+      "epoch": 0.6260434056761269,
+      "grad_norm": 0.7698463797569275,
+      "learning_rate": 2.954346183920129e-05,
+      "loss": 0.8861,
+      "step": 375
+    },
+    {
+      "epoch": 0.6343906510851419,
+      "grad_norm": 0.7722349166870117,
+      "learning_rate": 2.9522963123099292e-05,
+      "loss": 0.8338,
+      "step": 380
+    },
+    {
+      "epoch": 0.6427378964941569,
+      "grad_norm": 0.9294840097427368,
+      "learning_rate": 2.9502021686791238e-05,
+      "loss": 0.8988,
+      "step": 385
+    },
+    {
+      "epoch": 0.6510851419031719,
+      "grad_norm": 0.8922429084777832,
+      "learning_rate": 2.9480638168659043e-05,
+      "loss": 0.8197,
+      "step": 390
+    },
+    {
+      "epoch": 0.659432387312187,
+      "grad_norm": 0.8714888691902161,
+      "learning_rate": 2.9458813220561107e-05,
+      "loss": 0.9041,
+      "step": 395
+    },
+    {
+      "epoch": 0.667779632721202,
+      "grad_norm": 0.8319898843765259,
+      "learning_rate": 2.9436547507812454e-05,
+      "loss": 0.856,
+      "step": 400
+    },
+    {
+      "epoch": 0.676126878130217,
+      "grad_norm": 0.818681538105011,
+      "learning_rate": 2.941384170916444e-05,
+      "loss": 0.8121,
+      "step": 405
+    },
+    {
+      "epoch": 0.6844741235392321,
+      "grad_norm": 0.8972247838973999,
+      "learning_rate": 2.9390696516784063e-05,
+      "loss": 0.8501,
+      "step": 410
+    },
+    {
+      "epoch": 0.6928213689482471,
+      "grad_norm": 0.9798915982246399,
+      "learning_rate": 2.9367112636232882e-05,
+      "loss": 0.8597,
+      "step": 415
+    },
+    {
+      "epoch": 0.7011686143572621,
+      "grad_norm": 0.8564924001693726,
+      "learning_rate": 2.9343090786445473e-05,
+      "loss": 0.8355,
+      "step": 420
+    },
+    {
+      "epoch": 0.7095158597662772,
+      "grad_norm": 0.895405650138855,
+      "learning_rate": 2.9318631699707543e-05,
+      "loss": 0.8235,
+      "step": 425
+    },
+    {
+      "epoch": 0.7178631051752922,
+      "grad_norm": 0.9582717418670654,
+      "learning_rate": 2.9293736121633595e-05,
+      "loss": 0.8321,
+      "step": 430
+    },
+    {
+      "epoch": 0.7262103505843072,
+      "grad_norm": 0.8186922073364258,
+      "learning_rate": 2.92684048111442e-05,
+      "loss": 0.8084,
+      "step": 435
+    },
+    {
+      "epoch": 0.7345575959933222,
+      "grad_norm": 0.8169116973876953,
+      "learning_rate": 2.924263854044285e-05,
+      "loss": 0.8503,
+      "step": 440
+    },
+    {
+      "epoch": 0.7429048414023373,
+      "grad_norm": 0.9199667572975159,
+      "learning_rate": 2.9216438094992462e-05,
+      "loss": 0.7875,
+      "step": 445
+    },
+    {
+      "epoch": 0.7512520868113522,
+      "grad_norm": 0.8499664068222046,
+      "learning_rate": 2.9189804273491367e-05,
+      "loss": 0.8191,
+      "step": 450
+    },
+    {
+      "epoch": 0.7595993322203672,
+      "grad_norm": 0.9171932935714722,
+      "learning_rate": 2.916273788784901e-05,
+      "loss": 0.8218,
+      "step": 455
+    },
+    {
+      "epoch": 0.7679465776293823,
+      "grad_norm": 0.9836664795875549,
+      "learning_rate": 2.9135239763161193e-05,
+      "loss": 0.8104,
+      "step": 460
+    },
+    {
+      "epoch": 0.7762938230383973,
+      "grad_norm": 0.8993955850601196,
+      "learning_rate": 2.9107310737684916e-05,
+      "loss": 0.8565,
+      "step": 465
+    },
+    {
+      "epoch": 0.7846410684474123,
+      "grad_norm": 0.9021497368812561,
+      "learning_rate": 2.9078951662812802e-05,
+      "loss": 0.8161,
+      "step": 470
+    },
+    {
+      "epoch": 0.7929883138564274,
+      "grad_norm": 1.0624117851257324,
+      "learning_rate": 2.9050163403047198e-05,
+      "loss": 0.7775,
+      "step": 475
+    },
+    {
+      "epoch": 0.8013355592654424,
+      "grad_norm": 0.9302368760108948,
+      "learning_rate": 2.9020946835973768e-05,
+      "loss": 0.741,
+      "step": 480
+    },
+    {
+      "epoch": 0.8096828046744574,
+      "grad_norm": 0.888584554195404,
+      "learning_rate": 2.8991302852234763e-05,
+      "loss": 0.7862,
+      "step": 485
+    },
+    {
+      "epoch": 0.8180300500834724,
+      "grad_norm": 0.9118584394454956,
+      "learning_rate": 2.8961232355501867e-05,
+      "loss": 0.7813,
+      "step": 490
+    },
+    {
+      "epoch": 0.8263772954924875,
+      "grad_norm": 0.9881577491760254,
+      "learning_rate": 2.893073626244866e-05,
+      "loss": 0.7605,
+      "step": 495
+    },
+    {
+      "epoch": 0.8347245409015025,
+      "grad_norm": 0.9539454579353333,
+      "learning_rate": 2.889981550272265e-05,
+      "loss": 0.8336,
+      "step": 500
+    },
+    {
+      "epoch": 0.8430717863105175,
+      "grad_norm": 0.8545875549316406,
+      "learning_rate": 2.886847101891697e-05,
+      "loss": 0.7985,
+      "step": 505
+    },
+    {
+      "epoch": 0.8514190317195326,
+      "grad_norm": 0.9182264804840088,
+      "learning_rate": 2.8836703766541598e-05,
+      "loss": 0.7538,
+      "step": 510
+    },
+    {
+      "epoch": 0.8597662771285476,
+      "grad_norm": 0.8976413607597351,
+      "learning_rate": 2.8804514713994273e-05,
+      "loss": 0.7996,
+      "step": 515
+    },
+    {
+      "epoch": 0.8681135225375626,
+      "grad_norm": 1.0158463716506958,
+      "learning_rate": 2.8771904842530932e-05,
+      "loss": 0.7427,
+      "step": 520
+    },
+    {
+      "epoch": 0.8764607679465777,
+      "grad_norm": 0.9002468585968018,
+      "learning_rate": 2.8738875146235842e-05,
+      "loss": 0.7298,
+      "step": 525
+    },
+    {
+      "epoch": 0.8848080133555927,
+      "grad_norm": 1.0560840368270874,
+      "learning_rate": 2.8705426631991264e-05,
+      "loss": 0.7674,
+      "step": 530
+    },
+    {
+      "epoch": 0.8931552587646077,
+      "grad_norm": 0.9862357974052429,
+      "learning_rate": 2.8671560319446766e-05,
+      "loss": 0.7424,
+      "step": 535
+    },
+    {
+      "epoch": 0.9015025041736227,
+      "grad_norm": 0.9585193395614624,
+      "learning_rate": 2.8637277240988146e-05,
+      "loss": 0.7878,
+      "step": 540
+    },
+    {
+      "epoch": 0.9098497495826378,
+      "grad_norm": 0.9761247038841248,
+      "learning_rate": 2.8602578441705957e-05,
+      "loss": 0.7392,
+      "step": 545
+    },
+    {
+      "epoch": 0.9181969949916527,
+      "grad_norm": 0.8948637843132019,
+      "learning_rate": 2.856746497936365e-05,
+      "loss": 0.742,
+      "step": 550
+    },
+    {
+      "epoch": 0.9265442404006677,
+      "grad_norm": 0.9446806907653809,
+      "learning_rate": 2.8531937924365322e-05,
+      "loss": 0.6997,
+      "step": 555
+    },
+    {
+      "epoch": 0.9348914858096828,
+      "grad_norm": 0.8577098250389099,
+      "learning_rate": 2.8495998359723103e-05,
+      "loss": 0.7768,
+      "step": 560
+    },
+    {
+      "epoch": 0.9432387312186978,
+      "grad_norm": 0.9033234715461731,
+      "learning_rate": 2.8459647381024113e-05,
+      "loss": 0.7318,
+      "step": 565
+    },
+    {
+      "epoch": 0.9515859766277128,
+      "grad_norm": 1.1232320070266724,
+      "learning_rate": 2.8422886096397093e-05,
+      "loss": 0.7179,
+      "step": 570
+    },
+    {
+      "epoch": 0.9599332220367279,
+      "grad_norm": 1.0359822511672974,
+      "learning_rate": 2.8385715626478605e-05,
+      "loss": 0.7476,
+      "step": 575
+    },
+    {
+      "epoch": 0.9682804674457429,
+      "grad_norm": 1.1556510925292969,
+      "learning_rate": 2.8348137104378883e-05,
+      "loss": 0.7362,
+      "step": 580
+    },
+    {
+      "epoch": 0.9766277128547579,
+      "grad_norm": 0.9714240431785583,
+      "learning_rate": 2.8310151675647284e-05,
+      "loss": 0.7242,
+      "step": 585
+    },
+    {
+      "epoch": 0.9849749582637729,
+      "grad_norm": 0.9543178677558899,
+      "learning_rate": 2.8271760498237363e-05,
+      "loss": 0.7009,
+      "step": 590
+    },
+    {
+      "epoch": 0.993322203672788,
+      "grad_norm": 0.8914697766304016,
+      "learning_rate": 2.823296474247158e-05,
+      "loss": 0.7086,
+      "step": 595
+    },
+    {
+      "epoch": 1.001669449081803,
+      "grad_norm": 0.9592460989952087,
+      "learning_rate": 2.8193765591005633e-05,
+      "loss": 0.6801,
+      "step": 600
+    },
+    {
+      "epoch": 1.010016694490818,
+      "grad_norm": 1.1697297096252441,
+      "learning_rate": 2.815416423879239e-05,
+      "loss": 0.6585,
+      "step": 605
+    },
+    {
+      "epoch": 1.018363939899833,
+      "grad_norm": 0.9771791696548462,
+      "learning_rate": 2.8114161893045457e-05,
+      "loss": 0.6201,
+      "step": 610
+    },
+    {
+      "epoch": 1.026711185308848,
+      "grad_norm": 0.9642391800880432,
+      "learning_rate": 2.8073759773202405e-05,
+      "loss": 0.678,
+      "step": 615
+    },
+    {
+      "epoch": 1.0350584307178632,
+      "grad_norm": 0.9048047065734863,
+      "learning_rate": 2.803295911088757e-05,
+      "loss": 0.5851,
+      "step": 620
+    },
+    {
+      "epoch": 1.0434056761268782,
+      "grad_norm": 1.0365580320358276,
+      "learning_rate": 2.799176114987451e-05,
+      "loss": 0.6526,
+      "step": 625
+    },
+    {
+      "epoch": 1.0517529215358932,
+      "grad_norm": 1.0347423553466797,
+      "learning_rate": 2.795016714604811e-05,
+      "loss": 0.6402,
+      "step": 630
+    },
+    {
+      "epoch": 1.0601001669449082,
+      "grad_norm": 1.0601266622543335,
+      "learning_rate": 2.7908178367366275e-05,
+      "loss": 0.6087,
+      "step": 635
+    },
+    {
+      "epoch": 1.0684474123539232,
+      "grad_norm": 1.1169068813323975,
+      "learning_rate": 2.78657960938213e-05,
+      "loss": 0.5888,
+      "step": 640
+    },
+    {
+      "epoch": 1.0767946577629381,
+      "grad_norm": 1.0929909944534302,
+      "learning_rate": 2.782302161740081e-05,
+      "loss": 0.6069,
+      "step": 645
+    },
+    {
+      "epoch": 1.0851419031719534,
+      "grad_norm": 0.9820810556411743,
+      "learning_rate": 2.7779856242048434e-05,
+      "loss": 0.6454,
+      "step": 650
+    },
+    {
+      "epoch": 1.0934891485809684,
+      "grad_norm": 0.9788629412651062,
+      "learning_rate": 2.7736301283624e-05,
+      "loss": 0.581,
+      "step": 655
+    },
+    {
+      "epoch": 1.1018363939899833,
+      "grad_norm": 1.0459963083267212,
+      "learning_rate": 2.7692358069863455e-05,
+      "loss": 0.626,
+      "step": 660
+    },
+    {
+      "epoch": 1.1101836393989983,
+      "grad_norm": 1.0294015407562256,
+      "learning_rate": 2.7648027940338377e-05,
+      "loss": 0.561,
+      "step": 665
+    },
+    {
+      "epoch": 1.1185308848080133,
+      "grad_norm": 1.0476348400115967,
+      "learning_rate": 2.7603312246415142e-05,
+      "loss": 0.5558,
+      "step": 670
+    },
+    {
+      "epoch": 1.1268781302170283,
+      "grad_norm": 1.2524807453155518,
+      "learning_rate": 2.7558212351213744e-05,
+      "loss": 0.5823,
+      "step": 675
+    },
+    {
+      "epoch": 1.1352253756260433,
+      "grad_norm": 0.9370312094688416,
+      "learning_rate": 2.75127296295662e-05,
+      "loss": 0.606,
+      "step": 680
+    },
+    {
+      "epoch": 1.1435726210350585,
+      "grad_norm": 0.9592323899269104,
+      "learning_rate": 2.7466865467974686e-05,
+      "loss": 0.5904,
+      "step": 685
+    },
+    {
+      "epoch": 1.1519198664440735,
+      "grad_norm": 1.0453234910964966,
+      "learning_rate": 2.7420621264569247e-05,
+      "loss": 0.5774,
+      "step": 690
+    },
+    {
+      "epoch": 1.1602671118530885,
+      "grad_norm": 1.0608439445495605,
+      "learning_rate": 2.737399842906518e-05,
+      "loss": 0.5962,
+      "step": 695
+    },
+    {
+      "epoch": 1.1686143572621035,
+      "grad_norm": 1.0199201107025146,
+      "learning_rate": 2.7326998382720062e-05,
+      "loss": 0.5724,
+      "step": 700
+    },
+    {
+      "epoch": 1.1769616026711185,
+      "grad_norm": 1.007466435432434,
+      "learning_rate": 2.7279622558290412e-05,
+      "loss": 0.6032,
+      "step": 705
+    },
+    {
+      "epoch": 1.1853088480801335,
+      "grad_norm": 1.1127582788467407,
+      "learning_rate": 2.7231872399988042e-05,
+      "loss": 0.5602,
+      "step": 710
+    },
+    {
+      "epoch": 1.1936560934891487,
+      "grad_norm": 1.0738898515701294,
+      "learning_rate": 2.7183749363436e-05,
+      "loss": 0.5827,
+      "step": 715
+    },
+    {
+      "epoch": 1.2020033388981637,
+      "grad_norm": 1.0549622774124146,
+      "learning_rate": 2.7135254915624213e-05,
+      "loss": 0.5254,
+      "step": 720
+    },
+    {
+      "epoch": 1.2103505843071787,
+      "grad_norm": 1.1299083232879639,
+      "learning_rate": 2.7086390534864773e-05,
+      "loss": 0.5608,
+      "step": 725
+    },
+    {
+      "epoch": 1.2186978297161937,
+      "grad_norm": 0.998676598072052,
+      "learning_rate": 2.7037157710746855e-05,
+      "loss": 0.5304,
+      "step": 730
+    },
+    {
+      "epoch": 1.2270450751252087,
+      "grad_norm": 1.1849541664123535,
+      "learning_rate": 2.6987557944091327e-05,
+      "loss": 0.56,
+      "step": 735
+    },
+    {
+      "epoch": 1.2353923205342237,
+      "grad_norm": 1.0428630113601685,
+      "learning_rate": 2.6937592746904985e-05,
+      "loss": 0.5368,
+      "step": 740
+    },
+    {
+      "epoch": 1.2437395659432386,
+      "grad_norm": 1.0633243322372437,
+      "learning_rate": 2.6887263642334454e-05,
+      "loss": 0.5711,
+      "step": 745
+    },
+    {
+      "epoch": 1.2520868113522536,
+      "grad_norm": 1.0899866819381714,
+      "learning_rate": 2.683657216461978e-05,
+      "loss": 0.5749,
+      "step": 750
+    },
+    {
+      "epoch": 1.2604340567612689,
+      "grad_norm": 1.0628269910812378,
+      "learning_rate": 2.6785519859047644e-05,
+      "loss": 0.5492,
+      "step": 755
+    },
+    {
+      "epoch": 1.2687813021702838,
+      "grad_norm": 0.9858053922653198,
+      "learning_rate": 2.673410828190425e-05,
+      "loss": 0.6019,
+      "step": 760
+    },
+    {
+      "epoch": 1.2771285475792988,
+      "grad_norm": 1.1756415367126465,
+      "learning_rate": 2.6682339000427908e-05,
+      "loss": 0.5808,
+      "step": 765
+    },
+    {
+      "epoch": 1.2854757929883138,
+      "grad_norm": 0.9808685183525085,
+      "learning_rate": 2.6630213592761222e-05,
+      "loss": 0.5265,
+      "step": 770
+    },
+    {
+      "epoch": 1.2938230383973288,
+      "grad_norm": 1.0026417970657349,
+      "learning_rate": 2.6577733647903014e-05,
+      "loss": 0.5774,
+      "step": 775
+    },
+    {
+      "epoch": 1.302170283806344,
+      "grad_norm": 0.9454689621925354,
+      "learning_rate": 2.6524900765659858e-05,
+      "loss": 0.5496,
+      "step": 780
+    },
+    {
+      "epoch": 1.310517529215359,
+      "grad_norm": 1.0980932712554932,
+      "learning_rate": 2.6471716556597343e-05,
+      "loss": 0.5462,
+      "step": 785
+    },
+    {
+      "epoch": 1.318864774624374,
+      "grad_norm": 1.016065001487732,
+      "learning_rate": 2.6418182641990948e-05,
+      "loss": 0.5434,
+      "step": 790
+    },
+    {
+      "epoch": 1.327212020033389,
+      "grad_norm": 1.130270004272461,
+      "learning_rate": 2.6364300653776618e-05,
+      "loss": 0.5138,
+      "step": 795
+    },
+    {
+      "epoch": 1.335559265442404,
+      "grad_norm": 1.1002886295318604,
+      "learning_rate": 2.631007223450105e-05,
+      "loss": 0.5347,
+      "step": 800
+    },
+    {
+      "epoch": 1.343906510851419,
+      "grad_norm": 1.1232490539550781,
+      "learning_rate": 2.625549903727158e-05,
+      "loss": 0.541,
+      "step": 805
+    },
+    {
+      "epoch": 1.352253756260434,
+      "grad_norm": 0.9805702567100525,
+      "learning_rate": 2.620058272570581e-05,
+      "loss": 0.5234,
+      "step": 810
+    },
+    {
+      "epoch": 1.360601001669449,
+      "grad_norm": 1.0924123525619507,
+      "learning_rate": 2.61453249738809e-05,
+      "loss": 0.5339,
+      "step": 815
+    },
+    {
+      "epoch": 1.3689482470784642,
+      "grad_norm": 0.9931308627128601,
+      "learning_rate": 2.608972746628251e-05,
+      "loss": 0.5596,
+      "step": 820
+    },
+    {
+      "epoch": 1.3772954924874792,
+      "grad_norm": 1.1093590259552002,
+      "learning_rate": 2.6033791897753487e-05,
+      "loss": 0.5169,
+      "step": 825
+    },
+    {
+      "epoch": 1.3856427378964942,
+      "grad_norm": 1.104363203048706,
+      "learning_rate": 2.5977519973442164e-05,
+      "loss": 0.4898,
+      "step": 830
+    },
+    {
+      "epoch": 1.3939899833055092,
+      "grad_norm": 0.9823944568634033,
+      "learning_rate": 2.5920913408750402e-05,
+      "loss": 0.5376,
+      "step": 835
+    },
+    {
+      "epoch": 1.4023372287145242,
+      "grad_norm": 1.0676780939102173,
+      "learning_rate": 2.5863973929281288e-05,
+      "loss": 0.5472,
+      "step": 840
+    },
+    {
+      "epoch": 1.4106844741235394,
+      "grad_norm": 1.1943162679672241,
+      "learning_rate": 2.5806703270786526e-05,
+      "loss": 0.4788,
+      "step": 845
+    },
+    {
+      "epoch": 1.4190317195325544,
+      "grad_norm": 1.1296433210372925,
+      "learning_rate": 2.574910317911354e-05,
+      "loss": 0.4961,
+      "step": 850
+    },
+    {
+      "epoch": 1.4273789649415694,
+      "grad_norm": 1.1892937421798706,
+      "learning_rate": 2.5691175410152253e-05,
+      "loss": 0.4995,
+      "step": 855
+    },
+    {
+      "epoch": 1.4357262103505843,
+      "grad_norm": 1.0098474025726318,
+      "learning_rate": 2.5632921729781528e-05,
+      "loss": 0.4576,
+      "step": 860
+    },
+    {
+      "epoch": 1.4440734557595993,
+      "grad_norm": 0.9864776730537415,
+      "learning_rate": 2.5574343913815387e-05,
+      "loss": 0.5543,
+      "step": 865
+    },
+    {
+      "epoch": 1.4524207011686143,
+      "grad_norm": 1.1109187602996826,
+      "learning_rate": 2.551544374794883e-05,
+      "loss": 0.4911,
+      "step": 870
+    },
+    {
+      "epoch": 1.4607679465776293,
+      "grad_norm": 1.2205853462219238,
+      "learning_rate": 2.5456223027703427e-05,
+      "loss": 0.5013,
+      "step": 875
+    },
+    {
+      "epoch": 1.4691151919866443,
+      "grad_norm": 1.2574927806854248,
+      "learning_rate": 2.539668355837258e-05,
+      "loss": 0.4745,
+      "step": 880
+    },
+    {
+      "epoch": 1.4774624373956593,
+      "grad_norm": 1.2027695178985596,
+      "learning_rate": 2.533682715496649e-05,
+      "loss": 0.485,
+      "step": 885
+    },
+    {
+      "epoch": 1.4858096828046745,
+      "grad_norm": 1.032432198524475,
+      "learning_rate": 2.5276655642156814e-05,
+      "loss": 0.4859,
+      "step": 890
+    },
+    {
+      "epoch": 1.4941569282136895,
+      "grad_norm": 1.0579237937927246,
+      "learning_rate": 2.521617085422105e-05,
+      "loss": 0.489,
+      "step": 895
+    },
+    {
+      "epoch": 1.5025041736227045,
+      "grad_norm": 1.2205932140350342,
+      "learning_rate": 2.515537463498664e-05,
+      "loss": 0.4788,
+      "step": 900
+    },
+    {
+      "epoch": 1.5108514190317195,
+      "grad_norm": 0.9560028314590454,
+      "learning_rate": 2.509426883777473e-05,
+      "loss": 0.4796,
+      "step": 905
+    },
+    {
+      "epoch": 1.5191986644407347,
+      "grad_norm": 1.158149242401123,
+      "learning_rate": 2.5032855325343687e-05,
+      "loss": 0.4771,
+      "step": 910
+    },
+    {
+      "epoch": 1.5275459098497497,
+      "grad_norm": 1.2381256818771362,
+      "learning_rate": 2.497113596983232e-05,
+      "loss": 0.505,
+      "step": 915
+    },
+    {
+      "epoch": 1.5358931552587647,
+      "grad_norm": 1.0872339010238647,
+      "learning_rate": 2.4909112652702804e-05,
+      "loss": 0.4443,
+      "step": 920
+    },
+    {
+      "epoch": 1.5442404006677797,
+      "grad_norm": 1.1831974983215332,
+      "learning_rate": 2.4846787264683334e-05,
+      "loss": 0.4865,
+      "step": 925
+    },
+    {
+      "epoch": 1.5525876460767947,
+      "grad_norm": 1.3985413312911987,
+      "learning_rate": 2.478416170571046e-05,
+      "loss": 0.4794,
+      "step": 930
+    },
+    {
+      "epoch": 1.5609348914858097,
+      "grad_norm": 1.174689531326294,
+      "learning_rate": 2.472123788487121e-05,
+      "loss": 0.4824,
+      "step": 935
+    },
+    {
+      "epoch": 1.5692821368948247,
+      "grad_norm": 1.071053147315979,
+      "learning_rate": 2.4658017720344862e-05,
+      "loss": 0.4365,
+      "step": 940
+    },
+    {
+      "epoch": 1.5776293823038396,
+      "grad_norm": 1.0008587837219238,
+      "learning_rate": 2.459450313934447e-05,
+      "loss": 0.4488,
+      "step": 945
+    },
+    {
+      "epoch": 1.5859766277128546,
+      "grad_norm": 1.2077606916427612,
+      "learning_rate": 2.453069607805815e-05,
+      "loss": 0.4196,
+      "step": 950
+    },
+    {
+      "epoch": 1.5943238731218696,
+      "grad_norm": 1.0523104667663574,
+      "learning_rate": 2.446659848159001e-05,
+      "loss": 0.4244,
+      "step": 955
+    },
+    {
+      "epoch": 1.6026711185308848,
+      "grad_norm": 1.2364376783370972,
+      "learning_rate": 2.4402212303900877e-05,
+      "loss": 0.4857,
+      "step": 960
+    },
+    {
+      "epoch": 1.6110183639398998,
+      "grad_norm": 0.9511405229568481,
+      "learning_rate": 2.433753950774875e-05,
+      "loss": 0.4673,
+      "step": 965
+    },
+    {
+      "epoch": 1.6193656093489148,
+      "grad_norm": 1.0465832948684692,
+      "learning_rate": 2.427258206462892e-05,
+      "loss": 0.4204,
+      "step": 970
+    },
+    {
+      "epoch": 1.62771285475793,
+      "grad_norm": 1.0256762504577637,
+      "learning_rate": 2.4207341954713925e-05,
+      "loss": 0.4369,
+      "step": 975
+    },
+    {
+      "epoch": 1.636060100166945,
+      "grad_norm": 1.137253761291504,
+      "learning_rate": 2.414182116679315e-05,
+      "loss": 0.4575,
+      "step": 980
+    },
+    {
+      "epoch": 1.64440734557596,
+      "grad_norm": 1.1028913259506226,
+      "learning_rate": 2.4076021698212208e-05,
+      "loss": 0.5011,
+      "step": 985
+    },
+    {
+      "epoch": 1.652754590984975,
+      "grad_norm": 1.0733743906021118,
+      "learning_rate": 2.4009945554812064e-05,
+      "loss": 0.4618,
+      "step": 990
+    },
+    {
+      "epoch": 1.66110183639399,
+      "grad_norm": 1.1975722312927246,
+      "learning_rate": 2.394359475086787e-05,
+      "loss": 0.4332,
+      "step": 995
+    },
+    {
+      "epoch": 1.669449081803005,
+      "grad_norm": 1.1416840553283691,
+      "learning_rate": 2.387697130902757e-05,
+      "loss": 0.4704,
+      "step": 1000
+    },
+    {
+      "epoch": 1.67779632721202,
+      "grad_norm": 1.1874338388442993,
+      "learning_rate": 2.3810077260250268e-05,
+      "loss": 0.442,
+      "step": 1005
+    },
+    {
+      "epoch": 1.686143572621035,
+      "grad_norm": 0.9999452829360962,
+      "learning_rate": 2.3742914643744255e-05,
+      "loss": 0.4364,
+      "step": 1010
+    },
+    {
+      "epoch": 1.69449081803005,
+      "grad_norm": 1.1475085020065308,
+      "learning_rate": 2.367548550690491e-05,
+      "loss": 0.428,
+      "step": 1015
+    },
+    {
+      "epoch": 1.702838063439065,
+      "grad_norm": 1.049939513206482,
+      "learning_rate": 2.3607791905252232e-05,
+      "loss": 0.4365,
+      "step": 1020
+    },
+    {
+      "epoch": 1.7111853088480802,
+      "grad_norm": 1.1441938877105713,
+      "learning_rate": 2.3539835902368238e-05,
+      "loss": 0.4367,
+      "step": 1025
+    },
+    {
+      "epoch": 1.7195325542570952,
+      "grad_norm": 1.03773033618927,
+      "learning_rate": 2.347161956983399e-05,
+      "loss": 0.4067,
+      "step": 1030
+    },
+    {
+      "epoch": 1.7278797996661102,
+      "grad_norm": 1.2723736763000488,
+      "learning_rate": 2.3403144987166505e-05,
+      "loss": 0.4251,
+      "step": 1035
+    },
+    {
+      "epoch": 1.7362270450751254,
+      "grad_norm": 1.0961979627609253,
+      "learning_rate": 2.333441424175532e-05,
+      "loss": 0.3885,
+      "step": 1040
+    },
+    {
+      "epoch": 1.7445742904841404,
+      "grad_norm": 1.002756953239441,
+      "learning_rate": 2.3265429428798878e-05,
+      "loss": 0.4228,
+      "step": 1045
+    },
+    {
+      "epoch": 1.7529215358931554,
+      "grad_norm": 1.1268832683563232,
+      "learning_rate": 2.319619265124067e-05,
+      "loss": 0.4268,
+      "step": 1050
+    },
+    {
+      "epoch": 1.7612687813021703,
+      "grad_norm": 1.1599644422531128,
+      "learning_rate": 2.3126706019705093e-05,
+      "loss": 0.425,
+      "step": 1055
+    },
+    {
+      "epoch": 1.7696160267111853,
+      "grad_norm": 1.0869255065917969,
+      "learning_rate": 2.3056971652433145e-05,
+      "loss": 0.4035,
+      "step": 1060
+    },
+    {
+      "epoch": 1.7779632721202003,
+      "grad_norm": 1.1428903341293335,
+      "learning_rate": 2.298699167521783e-05,
+      "loss": 0.4355,
+      "step": 1065
+    },
+    {
+      "epoch": 1.7863105175292153,
+      "grad_norm": 1.090363621711731,
+      "learning_rate": 2.2916768221339372e-05,
+      "loss": 0.3947,
+      "step": 1070
+    },
+    {
+      "epoch": 1.7946577629382303,
+      "grad_norm": 1.0996553897857666,
+      "learning_rate": 2.284630343150016e-05,
+      "loss": 0.3775,
+      "step": 1075
+    },
+    {
+      "epoch": 1.8030050083472453,
+      "grad_norm": 1.1129885911941528,
+      "learning_rate": 2.2775599453759514e-05,
+      "loss": 0.4478,
+      "step": 1080
+    },
+    {
+      "epoch": 1.8113522537562603,
+      "grad_norm": 1.147131085395813,
+      "learning_rate": 2.27046584434682e-05,
+      "loss": 0.3752,
+      "step": 1085
+    },
+    {
+      "epoch": 1.8196994991652755,
+      "grad_norm": 1.1744565963745117,
+      "learning_rate": 2.263348256320271e-05,
+      "loss": 0.4113,
+      "step": 1090
+    },
+    {
+      "epoch": 1.8280467445742905,
+      "grad_norm": 1.2666568756103516,
+      "learning_rate": 2.2562073982699346e-05,
+      "loss": 0.3871,
+      "step": 1095
+    },
+    {
+      "epoch": 1.8363939899833055,
+      "grad_norm": 1.2098952531814575,
+      "learning_rate": 2.2490434878788094e-05,
+      "loss": 0.4084,
+      "step": 1100
+    },
+    {
+      "epoch": 1.8447412353923205,
+      "grad_norm": 1.218492865562439,
+      "learning_rate": 2.2418567435326233e-05,
+      "loss": 0.4108,
+      "step": 1105
+    },
+    {
+      "epoch": 1.8530884808013357,
+      "grad_norm": 1.3567547798156738,
+      "learning_rate": 2.234647384313179e-05,
+      "loss": 0.3972,
+      "step": 1110
+    },
+    {
+      "epoch": 1.8614357262103507,
+      "grad_norm": 1.090145230293274,
+      "learning_rate": 2.2274156299916746e-05,
+      "loss": 0.3838,
+      "step": 1115
+    },
+    {
+      "epoch": 1.8697829716193657,
+      "grad_norm": 1.1858583688735962,
+      "learning_rate": 2.2201617010220023e-05,
+      "loss": 0.3887,
+      "step": 1120
+    },
+    {
+      "epoch": 1.8781302170283807,
+      "grad_norm": 1.036993384361267,
+      "learning_rate": 2.2128858185340322e-05,
+      "loss": 0.4284,
+      "step": 1125
+    },
+    {
+      "epoch": 1.8864774624373957,
+      "grad_norm": 1.1040432453155518,
+      "learning_rate": 2.2055882043268663e-05,
+      "loss": 0.3715,
+      "step": 1130
+    },
+    {
+      "epoch": 1.8948247078464107,
+      "grad_norm": 0.9753995537757874,
+      "learning_rate": 2.198269080862081e-05,
+      "loss": 0.3748,
+      "step": 1135
+    },
+    {
+      "epoch": 1.9031719532554257,
+      "grad_norm": 1.0646343231201172,
+      "learning_rate": 2.1909286712569444e-05,
+      "loss": 0.4245,
+      "step": 1140
+    },
+    {
+      "epoch": 1.9115191986644406,
+      "grad_norm": 1.4246667623519897,
+      "learning_rate": 2.1835671992776132e-05,
+      "loss": 0.3573,
+      "step": 1145
+    },
+    {
+      "epoch": 1.9198664440734556,
+      "grad_norm": 1.0584313869476318,
+      "learning_rate": 2.176184889332314e-05,
+      "loss": 0.3806,
+      "step": 1150
+    },
+    {
+      "epoch": 1.9282136894824706,
+      "grad_norm": 1.0937577486038208,
+      "learning_rate": 2.1687819664644993e-05,
+      "loss": 0.35,
+      "step": 1155
+    },
+    {
+      "epoch": 1.9365609348914858,
+      "grad_norm": 1.183713436126709,
+      "learning_rate": 2.161358656345992e-05,
+      "loss": 0.3668,
+      "step": 1160
+    },
+    {
+      "epoch": 1.9449081803005008,
+      "grad_norm": 1.119342565536499,
+      "learning_rate": 2.153915185270101e-05,
+      "loss": 0.388,
+      "step": 1165
+    },
+    {
+      "epoch": 1.9532554257095158,
+      "grad_norm": 1.2510145902633667,
+      "learning_rate": 2.1464517801447244e-05,
+      "loss": 0.3594,
+      "step": 1170
+    },
+    {
+      "epoch": 1.961602671118531,
+      "grad_norm": 1.1926363706588745,
+      "learning_rate": 2.138968668485434e-05,
+      "loss": 0.4232,
+      "step": 1175
+    },
+    {
+      "epoch": 1.969949916527546,
+      "grad_norm": 1.211121916770935,
+      "learning_rate": 2.1314660784085387e-05,
+      "loss": 0.3796,
+      "step": 1180
+    },
+    {
+      "epoch": 1.978297161936561,
+      "grad_norm": 1.1046868562698364,
+      "learning_rate": 2.123944238624129e-05,
+      "loss": 0.348,
+      "step": 1185
+    },
+    {
+      "epoch": 1.986644407345576,
+      "grad_norm": 1.2281684875488281,
+      "learning_rate": 2.1164033784291092e-05,
+      "loss": 0.4116,
+      "step": 1190
+    },
+    {
+      "epoch": 1.994991652754591,
+      "grad_norm": 1.1758081912994385,
+      "learning_rate": 2.1088437277002014e-05,
+      "loss": 0.3496,
+      "step": 1195
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 2995,
+  "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": 1.738952727704109e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

23_128_e5_3e-5/checkpoint-1198/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:462836112e88c18d92160ccca50fdb00995a79d39c0918ae35a73e6be56e20a2
+size 7736

23_128_e5_3e-5/checkpoint-1198/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)

23_128_e5_3e-5/checkpoint-1797/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

23_128_e5_3e-5/checkpoint-1797/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",
+    "gate_proj",
+    "v_proj",
+    "k_proj",
+    "up_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

23_128_e5_3e-5/checkpoint-1797/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c56d1000f343cb02729317b65dc6380caa49f2bf9e96beb0b1507046f4e0dd29
+size 791751704

23_128_e5_3e-5/checkpoint-1797/latest

@@ -0,0 +1 @@
+global_step1797

23_128_e5_3e-5/checkpoint-1797/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:57df9dd45d86acc9a9cc65db5ddef2aba99875dd0f0d48dd2f79b30e47fe3441
+size 15920

23_128_e5_3e-5/checkpoint-1797/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0193f7b2d69ad94310a3caa0ace8ef98b1b0c7105fff1384785b0c7ccd856445
+size 15920

23_128_e5_3e-5/checkpoint-1797/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b312e01eaa724380879531a9fbf048e390b2412aaee3465a322f2d0b9236ce4d
+size 15920

23_128_e5_3e-5/checkpoint-1797/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7e4ae0bbb3ec3eb12dfdd1ece6001a07b040f057d0f3dd9cff9d49634106efcc
+size 15920

23_128_e5_3e-5/checkpoint-1797/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fd7c709f30b8197ea49b92c942c11683e0f41bca8be48ab5716ee633060e6f99
+size 15920

23_128_e5_3e-5/checkpoint-1797/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5229d83e6b3fa9d1773734e21f5e2198c3ad3439547dcddbd7783e3fbac775f2
+size 15920

23_128_e5_3e-5/checkpoint-1797/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f8385695387396e0a367340909264dd6d21991d842a69dd75c5c9f989d33258
+size 15920

23_128_e5_3e-5/checkpoint-1797/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3d103a6950f24f0031ff5b7a745fbbafe8b28158372aae652e7586b339375f3e
+size 15920

23_128_e5_3e-5/checkpoint-1797/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2dec8e2e9d9ea4c1c4de14d11d24907397459ce7b16fc03e810cb4d21958d14d
+size 1064

23_128_e5_3e-5/checkpoint-1797/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
+  }
+}

23_128_e5_3e-5/checkpoint-1797/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
+}

23_128_e5_3e-5/checkpoint-1797/trainer_state.json

@@ -0,0 +1,2547 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 1797,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.008347245409015025,
+      "grad_norm": 0.9250032305717468,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 1.3897,
+      "step": 5
+    },
+    {
+      "epoch": 0.01669449081803005,
+      "grad_norm": 1.0577523708343506,
+      "learning_rate": 1.8e-06,
+      "loss": 1.3743,
+      "step": 10
+    },
+    {
+      "epoch": 0.025041736227045076,
+      "grad_norm": 0.7588217258453369,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 1.3647,
+      "step": 15
+    },
+    {
+      "epoch": 0.0333889816360601,
+      "grad_norm": 0.7027095556259155,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 1.3301,
+      "step": 20
+    },
+    {
+      "epoch": 0.041736227045075125,
+      "grad_norm": 0.47005781531333923,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.2668,
+      "step": 25
+    },
+    {
+      "epoch": 0.05008347245409015,
+      "grad_norm": 0.49903222918510437,
+      "learning_rate": 5.8e-06,
+      "loss": 1.3468,
+      "step": 30
+    },
+    {
+      "epoch": 0.05843071786310518,
+      "grad_norm": 0.541542649269104,
+      "learning_rate": 6.8e-06,
+      "loss": 1.2821,
+      "step": 35
+    },
+    {
+      "epoch": 0.0667779632721202,
+      "grad_norm": 0.5814635157585144,
+      "learning_rate": 7.8e-06,
+      "loss": 1.2576,
+      "step": 40
+    },
+    {
+      "epoch": 0.07512520868113523,
+      "grad_norm": 0.4928892254829407,
+      "learning_rate": 8.8e-06,
+      "loss": 1.2328,
+      "step": 45
+    },
+    {
+      "epoch": 0.08347245409015025,
+      "grad_norm": 0.44546306133270264,
+      "learning_rate": 9.8e-06,
+      "loss": 1.2874,
+      "step": 50
+    },
+    {
+      "epoch": 0.09181969949916527,
+      "grad_norm": 0.41543567180633545,
+      "learning_rate": 1.08e-05,
+      "loss": 1.2451,
+      "step": 55
+    },
+    {
+      "epoch": 0.1001669449081803,
+      "grad_norm": 0.4427853524684906,
+      "learning_rate": 1.1799999999999999e-05,
+      "loss": 1.2629,
+      "step": 60
+    },
+    {
+      "epoch": 0.10851419031719532,
+      "grad_norm": 0.36455604434013367,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 1.2237,
+      "step": 65
+    },
+    {
+      "epoch": 0.11686143572621036,
+      "grad_norm": 0.4628913998603821,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 1.2426,
+      "step": 70
+    },
+    {
+      "epoch": 0.12520868113522537,
+      "grad_norm": 0.4718121588230133,
+      "learning_rate": 1.48e-05,
+      "loss": 1.2996,
+      "step": 75
+    },
+    {
+      "epoch": 0.1335559265442404,
+      "grad_norm": 0.4145875573158264,
+      "learning_rate": 1.5799999999999998e-05,
+      "loss": 1.2439,
+      "step": 80
+    },
+    {
+      "epoch": 0.1419031719532554,
+      "grad_norm": 0.46866002678871155,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 1.2307,
+      "step": 85
+    },
+    {
+      "epoch": 0.15025041736227046,
+      "grad_norm": 0.4653532803058624,
+      "learning_rate": 1.7800000000000002e-05,
+      "loss": 1.2022,
+      "step": 90
+    },
+    {
+      "epoch": 0.15859766277128548,
+      "grad_norm": 0.5028699636459351,
+      "learning_rate": 1.8800000000000003e-05,
+      "loss": 1.2042,
+      "step": 95
+    },
+    {
+      "epoch": 0.1669449081803005,
+      "grad_norm": 0.4724132716655731,
+      "learning_rate": 1.98e-05,
+      "loss": 1.1189,
+      "step": 100
+    },
+    {
+      "epoch": 0.17529215358931552,
+      "grad_norm": 0.49735012650489807,
+      "learning_rate": 2.08e-05,
+      "loss": 1.2435,
+      "step": 105
+    },
+    {
+      "epoch": 0.18363939899833054,
+      "grad_norm": 0.5972133874893188,
+      "learning_rate": 2.18e-05,
+      "loss": 1.1465,
+      "step": 110
+    },
+    {
+      "epoch": 0.19198664440734559,
+      "grad_norm": 0.47499340772628784,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 1.207,
+      "step": 115
+    },
+    {
+      "epoch": 0.2003338898163606,
+      "grad_norm": 0.4662191867828369,
+      "learning_rate": 2.38e-05,
+      "loss": 1.18,
+      "step": 120
+    },
+    {
+      "epoch": 0.20868113522537562,
+      "grad_norm": 0.8468295931816101,
+      "learning_rate": 2.48e-05,
+      "loss": 1.1867,
+      "step": 125
+    },
+    {
+      "epoch": 0.21702838063439064,
+      "grad_norm": 0.528782069683075,
+      "learning_rate": 2.58e-05,
+      "loss": 1.1633,
+      "step": 130
+    },
+    {
+      "epoch": 0.22537562604340566,
+      "grad_norm": 0.5306044220924377,
+      "learning_rate": 2.68e-05,
+      "loss": 1.1775,
+      "step": 135
+    },
+    {
+      "epoch": 0.2337228714524207,
+      "grad_norm": 0.5375636219978333,
+      "learning_rate": 2.78e-05,
+      "loss": 1.188,
+      "step": 140
+    },
+    {
+      "epoch": 0.24207011686143573,
+      "grad_norm": 0.5145397782325745,
+      "learning_rate": 2.88e-05,
+      "loss": 1.1268,
+      "step": 145
+    },
+    {
+      "epoch": 0.25041736227045075,
+      "grad_norm": 0.5362802147865295,
+      "learning_rate": 2.98e-05,
+      "loss": 1.1542,
+      "step": 150
+    },
+    {
+      "epoch": 0.2587646076794658,
+      "grad_norm": 0.5439884662628174,
+      "learning_rate": 2.999985367593965e-05,
+      "loss": 1.13,
+      "step": 155
+    },
+    {
+      "epoch": 0.2671118530884808,
+      "grad_norm": 0.487518310546875,
+      "learning_rate": 2.999925923933716e-05,
+      "loss": 1.109,
+      "step": 160
+    },
+    {
+      "epoch": 0.27545909849749584,
+      "grad_norm": 0.5048381686210632,
+      "learning_rate": 2.999820756304571e-05,
+      "loss": 1.1214,
+      "step": 165
+    },
+    {
+      "epoch": 0.2838063439065108,
+      "grad_norm": 0.5778033137321472,
+      "learning_rate": 2.999669867912476e-05,
+      "loss": 1.1179,
+      "step": 170
+    },
+    {
+      "epoch": 0.2921535893155259,
+      "grad_norm": 0.5533081293106079,
+      "learning_rate": 2.999473263357136e-05,
+      "loss": 1.0759,
+      "step": 175
+    },
+    {
+      "epoch": 0.3005008347245409,
+      "grad_norm": 0.6201685070991516,
+      "learning_rate": 2.999230948631874e-05,
+      "loss": 1.058,
+      "step": 180
+    },
+    {
+      "epoch": 0.3088480801335559,
+      "grad_norm": 0.5643416047096252,
+      "learning_rate": 2.9989429311234485e-05,
+      "loss": 1.0972,
+      "step": 185
+    },
+    {
+      "epoch": 0.31719532554257096,
+      "grad_norm": 0.5617562532424927,
+      "learning_rate": 2.9986092196118284e-05,
+      "loss": 1.0984,
+      "step": 190
+    },
+    {
+      "epoch": 0.32554257095158595,
+      "grad_norm": 0.5762996673583984,
+      "learning_rate": 2.998229824269927e-05,
+      "loss": 1.0559,
+      "step": 195
+    },
+    {
+      "epoch": 0.333889816360601,
+      "grad_norm": 0.5630249977111816,
+      "learning_rate": 2.9978047566632887e-05,
+      "loss": 1.0928,
+      "step": 200
+    },
+    {
+      "epoch": 0.34223706176961605,
+      "grad_norm": 0.5413353443145752,
+      "learning_rate": 2.9973340297497384e-05,
+      "loss": 1.0944,
+      "step": 205
+    },
+    {
+      "epoch": 0.35058430717863104,
+      "grad_norm": 0.6307699680328369,
+      "learning_rate": 2.9968176578789874e-05,
+      "loss": 1.1564,
+      "step": 210
+    },
+    {
+      "epoch": 0.3589315525876461,
+      "grad_norm": 0.7229161858558655,
+      "learning_rate": 2.9962556567921933e-05,
+      "loss": 1.0633,
+      "step": 215
+    },
+    {
+      "epoch": 0.3672787979966611,
+      "grad_norm": 0.6549956798553467,
+      "learning_rate": 2.9956480436214828e-05,
+      "loss": 1.0555,
+      "step": 220
+    },
+    {
+      "epoch": 0.3756260434056761,
+      "grad_norm": 0.5410817265510559,
+      "learning_rate": 2.994994836889427e-05,
+      "loss": 1.0631,
+      "step": 225
+    },
+    {
+      "epoch": 0.38397328881469117,
+      "grad_norm": 0.6632435917854309,
+      "learning_rate": 2.9942960565084802e-05,
+      "loss": 1.0573,
+      "step": 230
+    },
+    {
+      "epoch": 0.39232053422370616,
+      "grad_norm": 0.6554168462753296,
+      "learning_rate": 2.9935517237803675e-05,
+      "loss": 1.0133,
+      "step": 235
+    },
+    {
+      "epoch": 0.4006677796327212,
+      "grad_norm": 0.6186609864234924,
+      "learning_rate": 2.9927618613954417e-05,
+      "loss": 1.0114,
+      "step": 240
+    },
+    {
+      "epoch": 0.4090150250417362,
+      "grad_norm": 0.6478000283241272,
+      "learning_rate": 2.9919264934319874e-05,
+      "loss": 1.0423,
+      "step": 245
+    },
+    {
+      "epoch": 0.41736227045075125,
+      "grad_norm": 0.6168914437294006,
+      "learning_rate": 2.9910456453554868e-05,
+      "loss": 1.0181,
+      "step": 250
+    },
+    {
+      "epoch": 0.4257095158597663,
+      "grad_norm": 0.8668388724327087,
+      "learning_rate": 2.990119344017847e-05,
+      "loss": 1.0286,
+      "step": 255
+    },
+    {
+      "epoch": 0.4340567612687813,
+      "grad_norm": 0.6979084610939026,
+      "learning_rate": 2.9891476176565784e-05,
+      "loss": 1.0031,
+      "step": 260
+    },
+    {
+      "epoch": 0.44240400667779634,
+      "grad_norm": 0.7331295609474182,
+      "learning_rate": 2.9881304958939334e-05,
+      "loss": 1.0001,
+      "step": 265
+    },
+    {
+      "epoch": 0.4507512520868113,
+      "grad_norm": 0.8051875829696655,
+      "learning_rate": 2.987068009736007e-05,
+      "loss": 0.9713,
+      "step": 270
+    },
+    {
+      "epoch": 0.4590984974958264,
+      "grad_norm": 0.7588741183280945,
+      "learning_rate": 2.985960191571787e-05,
+      "loss": 0.9831,
+      "step": 275
+    },
+    {
+      "epoch": 0.4674457429048414,
+      "grad_norm": 0.7282049059867859,
+      "learning_rate": 2.9848070751721704e-05,
+      "loss": 0.9665,
+      "step": 280
+    },
+    {
+      "epoch": 0.4757929883138564,
+      "grad_norm": 0.7469468116760254,
+      "learning_rate": 2.9836086956889328e-05,
+      "loss": 1.004,
+      "step": 285
+    },
+    {
+      "epoch": 0.48414023372287146,
+      "grad_norm": 0.7200749516487122,
+      "learning_rate": 2.9823650896536545e-05,
+      "loss": 0.9635,
+      "step": 290
+    },
+    {
+      "epoch": 0.49248747913188645,
+      "grad_norm": 0.6923590898513794,
+      "learning_rate": 2.9810762949766113e-05,
+      "loss": 0.9508,
+      "step": 295
+    },
+    {
+      "epoch": 0.5008347245409015,
+      "grad_norm": 0.738401472568512,
+      "learning_rate": 2.979742350945614e-05,
+      "loss": 0.9086,
+      "step": 300
+    },
+    {
+      "epoch": 0.5091819699499165,
+      "grad_norm": 0.7354897856712341,
+      "learning_rate": 2.9783632982248152e-05,
+      "loss": 0.9975,
+      "step": 305
+    },
+    {
+      "epoch": 0.5175292153589316,
+      "grad_norm": 0.8435232639312744,
+      "learning_rate": 2.9769391788534662e-05,
+      "loss": 0.9406,
+      "step": 310
+    },
+    {
+      "epoch": 0.5258764607679466,
+      "grad_norm": 1.0065906047821045,
+      "learning_rate": 2.9754700362446382e-05,
+      "loss": 0.922,
+      "step": 315
+    },
+    {
+      "epoch": 0.5342237061769616,
+      "grad_norm": 0.7278281450271606,
+      "learning_rate": 2.9739559151838964e-05,
+      "loss": 0.9792,
+      "step": 320
+    },
+    {
+      "epoch": 0.5425709515859767,
+      "grad_norm": 0.8112268447875977,
+      "learning_rate": 2.972396861827936e-05,
+      "loss": 0.9774,
+      "step": 325
+    },
+    {
+      "epoch": 0.5509181969949917,
+      "grad_norm": 0.7046104073524475,
+      "learning_rate": 2.9707929237031765e-05,
+      "loss": 0.9139,
+      "step": 330
+    },
+    {
+      "epoch": 0.5592654424040067,
+      "grad_norm": 0.7925729751586914,
+      "learning_rate": 2.9691441497043098e-05,
+      "loss": 0.8911,
+      "step": 335
+    },
+    {
+      "epoch": 0.5676126878130217,
+      "grad_norm": 0.7873656749725342,
+      "learning_rate": 2.9674505900928118e-05,
+      "loss": 0.9207,
+      "step": 340
+    },
+    {
+      "epoch": 0.5759599332220368,
+      "grad_norm": 0.8177038431167603,
+      "learning_rate": 2.9657122964954107e-05,
+      "loss": 0.9357,
+      "step": 345
+    },
+    {
+      "epoch": 0.5843071786310517,
+      "grad_norm": 0.7847334742546082,
+      "learning_rate": 2.9639293219025116e-05,
+      "loss": 0.8949,
+      "step": 350
+    },
+    {
+      "epoch": 0.5926544240400667,
+      "grad_norm": 0.8220208287239075,
+      "learning_rate": 2.9621017206665807e-05,
+      "loss": 0.9603,
+      "step": 355
+    },
+    {
+      "epoch": 0.6010016694490818,
+      "grad_norm": 0.8926270008087158,
+      "learning_rate": 2.9602295485004922e-05,
+      "loss": 0.9297,
+      "step": 360
+    },
+    {
+      "epoch": 0.6093489148580968,
+      "grad_norm": 0.834158718585968,
+      "learning_rate": 2.9583128624758252e-05,
+      "loss": 0.9139,
+      "step": 365
+    },
+    {
+      "epoch": 0.6176961602671118,
+      "grad_norm": 0.8102491497993469,
+      "learning_rate": 2.9563517210211266e-05,
+      "loss": 0.8262,
+      "step": 370
+    },
+    {
+      "epoch": 0.6260434056761269,
+      "grad_norm": 0.7698463797569275,
+      "learning_rate": 2.954346183920129e-05,
+      "loss": 0.8861,
+      "step": 375
+    },
+    {
+      "epoch": 0.6343906510851419,
+      "grad_norm": 0.7722349166870117,
+      "learning_rate": 2.9522963123099292e-05,
+      "loss": 0.8338,
+      "step": 380
+    },
+    {
+      "epoch": 0.6427378964941569,
+      "grad_norm": 0.9294840097427368,
+      "learning_rate": 2.9502021686791238e-05,
+      "loss": 0.8988,
+      "step": 385
+    },
+    {
+      "epoch": 0.6510851419031719,
+      "grad_norm": 0.8922429084777832,
+      "learning_rate": 2.9480638168659043e-05,
+      "loss": 0.8197,
+      "step": 390
+    },
+    {
+      "epoch": 0.659432387312187,
+      "grad_norm": 0.8714888691902161,
+      "learning_rate": 2.9458813220561107e-05,
+      "loss": 0.9041,
+      "step": 395
+    },
+    {
+      "epoch": 0.667779632721202,
+      "grad_norm": 0.8319898843765259,
+      "learning_rate": 2.9436547507812454e-05,
+      "loss": 0.856,
+      "step": 400
+    },
+    {
+      "epoch": 0.676126878130217,
+      "grad_norm": 0.818681538105011,
+      "learning_rate": 2.941384170916444e-05,
+      "loss": 0.8121,
+      "step": 405
+    },
+    {
+      "epoch": 0.6844741235392321,
+      "grad_norm": 0.8972247838973999,
+      "learning_rate": 2.9390696516784063e-05,
+      "loss": 0.8501,
+      "step": 410
+    },
+    {
+      "epoch": 0.6928213689482471,
+      "grad_norm": 0.9798915982246399,
+      "learning_rate": 2.9367112636232882e-05,
+      "loss": 0.8597,
+      "step": 415
+    },
+    {
+      "epoch": 0.7011686143572621,
+      "grad_norm": 0.8564924001693726,
+      "learning_rate": 2.9343090786445473e-05,
+      "loss": 0.8355,
+      "step": 420
+    },
+    {
+      "epoch": 0.7095158597662772,
+      "grad_norm": 0.895405650138855,
+      "learning_rate": 2.9318631699707543e-05,
+      "loss": 0.8235,
+      "step": 425
+    },
+    {
+      "epoch": 0.7178631051752922,
+      "grad_norm": 0.9582717418670654,
+      "learning_rate": 2.9293736121633595e-05,
+      "loss": 0.8321,
+      "step": 430
+    },
+    {
+      "epoch": 0.7262103505843072,
+      "grad_norm": 0.8186922073364258,
+      "learning_rate": 2.92684048111442e-05,
+      "loss": 0.8084,
+      "step": 435
+    },
+    {
+      "epoch": 0.7345575959933222,
+      "grad_norm": 0.8169116973876953,
+      "learning_rate": 2.924263854044285e-05,
+      "loss": 0.8503,
+      "step": 440
+    },
+    {
+      "epoch": 0.7429048414023373,
+      "grad_norm": 0.9199667572975159,
+      "learning_rate": 2.9216438094992462e-05,
+      "loss": 0.7875,
+      "step": 445
+    },
+    {
+      "epoch": 0.7512520868113522,
+      "grad_norm": 0.8499664068222046,
+      "learning_rate": 2.9189804273491367e-05,
+      "loss": 0.8191,
+      "step": 450
+    },
+    {
+      "epoch": 0.7595993322203672,
+      "grad_norm": 0.9171932935714722,
+      "learning_rate": 2.916273788784901e-05,
+      "loss": 0.8218,
+      "step": 455
+    },
+    {
+      "epoch": 0.7679465776293823,
+      "grad_norm": 0.9836664795875549,
+      "learning_rate": 2.9135239763161193e-05,
+      "loss": 0.8104,
+      "step": 460
+    },
+    {
+      "epoch": 0.7762938230383973,
+      "grad_norm": 0.8993955850601196,
+      "learning_rate": 2.9107310737684916e-05,
+      "loss": 0.8565,
+      "step": 465
+    },
+    {
+      "epoch": 0.7846410684474123,
+      "grad_norm": 0.9021497368812561,
+      "learning_rate": 2.9078951662812802e-05,
+      "loss": 0.8161,
+      "step": 470
+    },
+    {
+      "epoch": 0.7929883138564274,
+      "grad_norm": 1.0624117851257324,
+      "learning_rate": 2.9050163403047198e-05,
+      "loss": 0.7775,
+      "step": 475
+    },
+    {
+      "epoch": 0.8013355592654424,
+      "grad_norm": 0.9302368760108948,
+      "learning_rate": 2.9020946835973768e-05,
+      "loss": 0.741,
+      "step": 480
+    },
+    {
+      "epoch": 0.8096828046744574,
+      "grad_norm": 0.888584554195404,
+      "learning_rate": 2.8991302852234763e-05,
+      "loss": 0.7862,
+      "step": 485
+    },
+    {
+      "epoch": 0.8180300500834724,
+      "grad_norm": 0.9118584394454956,
+      "learning_rate": 2.8961232355501867e-05,
+      "loss": 0.7813,
+      "step": 490
+    },
+    {
+      "epoch": 0.8263772954924875,
+      "grad_norm": 0.9881577491760254,
+      "learning_rate": 2.893073626244866e-05,
+      "loss": 0.7605,
+      "step": 495
+    },
+    {
+      "epoch": 0.8347245409015025,
+      "grad_norm": 0.9539454579353333,
+      "learning_rate": 2.889981550272265e-05,
+      "loss": 0.8336,
+      "step": 500
+    },
+    {
+      "epoch": 0.8430717863105175,
+      "grad_norm": 0.8545875549316406,
+      "learning_rate": 2.886847101891697e-05,
+      "loss": 0.7985,
+      "step": 505
+    },
+    {
+      "epoch": 0.8514190317195326,
+      "grad_norm": 0.9182264804840088,
+      "learning_rate": 2.8836703766541598e-05,
+      "loss": 0.7538,
+      "step": 510
+    },
+    {
+      "epoch": 0.8597662771285476,
+      "grad_norm": 0.8976413607597351,
+      "learning_rate": 2.8804514713994273e-05,
+      "loss": 0.7996,
+      "step": 515
+    },
+    {
+      "epoch": 0.8681135225375626,
+      "grad_norm": 1.0158463716506958,
+      "learning_rate": 2.8771904842530932e-05,
+      "loss": 0.7427,
+      "step": 520
+    },
+    {
+      "epoch": 0.8764607679465777,
+      "grad_norm": 0.9002468585968018,
+      "learning_rate": 2.8738875146235842e-05,
+      "loss": 0.7298,
+      "step": 525
+    },
+    {
+      "epoch": 0.8848080133555927,
+      "grad_norm": 1.0560840368270874,
+      "learning_rate": 2.8705426631991264e-05,
+      "loss": 0.7674,
+      "step": 530
+    },
+    {
+      "epoch": 0.8931552587646077,
+      "grad_norm": 0.9862357974052429,
+      "learning_rate": 2.8671560319446766e-05,
+      "loss": 0.7424,
+      "step": 535
+    },
+    {
+      "epoch": 0.9015025041736227,
+      "grad_norm": 0.9585193395614624,
+      "learning_rate": 2.8637277240988146e-05,
+      "loss": 0.7878,
+      "step": 540
+    },
+    {
+      "epoch": 0.9098497495826378,
+      "grad_norm": 0.9761247038841248,
+      "learning_rate": 2.8602578441705957e-05,
+      "loss": 0.7392,
+      "step": 545
+    },
+    {
+      "epoch": 0.9181969949916527,
+      "grad_norm": 0.8948637843132019,
+      "learning_rate": 2.856746497936365e-05,
+      "loss": 0.742,
+      "step": 550
+    },
+    {
+      "epoch": 0.9265442404006677,
+      "grad_norm": 0.9446806907653809,
+      "learning_rate": 2.8531937924365322e-05,
+      "loss": 0.6997,
+      "step": 555
+    },
+    {
+      "epoch": 0.9348914858096828,
+      "grad_norm": 0.8577098250389099,
+      "learning_rate": 2.8495998359723103e-05,
+      "loss": 0.7768,
+      "step": 560
+    },
+    {
+      "epoch": 0.9432387312186978,
+      "grad_norm": 0.9033234715461731,
+      "learning_rate": 2.8459647381024113e-05,
+      "loss": 0.7318,
+      "step": 565
+    },
+    {
+      "epoch": 0.9515859766277128,
+      "grad_norm": 1.1232320070266724,
+      "learning_rate": 2.8422886096397093e-05,
+      "loss": 0.7179,
+      "step": 570
+    },
+    {
+      "epoch": 0.9599332220367279,
+      "grad_norm": 1.0359822511672974,
+      "learning_rate": 2.8385715626478605e-05,
+      "loss": 0.7476,
+      "step": 575
+    },
+    {
+      "epoch": 0.9682804674457429,
+      "grad_norm": 1.1556510925292969,
+      "learning_rate": 2.8348137104378883e-05,
+      "loss": 0.7362,
+      "step": 580
+    },
+    {
+      "epoch": 0.9766277128547579,
+      "grad_norm": 0.9714240431785583,
+      "learning_rate": 2.8310151675647284e-05,
+      "loss": 0.7242,
+      "step": 585
+    },
+    {
+      "epoch": 0.9849749582637729,
+      "grad_norm": 0.9543178677558899,
+      "learning_rate": 2.8271760498237363e-05,
+      "loss": 0.7009,
+      "step": 590
+    },
+    {
+      "epoch": 0.993322203672788,
+      "grad_norm": 0.8914697766304016,
+      "learning_rate": 2.823296474247158e-05,
+      "loss": 0.7086,
+      "step": 595
+    },
+    {
+      "epoch": 1.001669449081803,
+      "grad_norm": 0.9592460989952087,
+      "learning_rate": 2.8193765591005633e-05,
+      "loss": 0.6801,
+      "step": 600
+    },
+    {
+      "epoch": 1.010016694490818,
+      "grad_norm": 1.1697297096252441,
+      "learning_rate": 2.815416423879239e-05,
+      "loss": 0.6585,
+      "step": 605
+    },
+    {
+      "epoch": 1.018363939899833,
+      "grad_norm": 0.9771791696548462,
+      "learning_rate": 2.8114161893045457e-05,
+      "loss": 0.6201,
+      "step": 610
+    },
+    {
+      "epoch": 1.026711185308848,
+      "grad_norm": 0.9642391800880432,
+      "learning_rate": 2.8073759773202405e-05,
+      "loss": 0.678,
+      "step": 615
+    },
+    {
+      "epoch": 1.0350584307178632,
+      "grad_norm": 0.9048047065734863,
+      "learning_rate": 2.803295911088757e-05,
+      "loss": 0.5851,
+      "step": 620
+    },
+    {
+      "epoch": 1.0434056761268782,
+      "grad_norm": 1.0365580320358276,
+      "learning_rate": 2.799176114987451e-05,
+      "loss": 0.6526,
+      "step": 625
+    },
+    {
+      "epoch": 1.0517529215358932,
+      "grad_norm": 1.0347423553466797,
+      "learning_rate": 2.795016714604811e-05,
+      "loss": 0.6402,
+      "step": 630
+    },
+    {
+      "epoch": 1.0601001669449082,
+      "grad_norm": 1.0601266622543335,
+      "learning_rate": 2.7908178367366275e-05,
+      "loss": 0.6087,
+      "step": 635
+    },
+    {
+      "epoch": 1.0684474123539232,
+      "grad_norm": 1.1169068813323975,
+      "learning_rate": 2.78657960938213e-05,
+      "loss": 0.5888,
+      "step": 640
+    },
+    {
+      "epoch": 1.0767946577629381,
+      "grad_norm": 1.0929909944534302,
+      "learning_rate": 2.782302161740081e-05,
+      "loss": 0.6069,
+      "step": 645
+    },
+    {
+      "epoch": 1.0851419031719534,
+      "grad_norm": 0.9820810556411743,
+      "learning_rate": 2.7779856242048434e-05,
+      "loss": 0.6454,
+      "step": 650
+    },
+    {
+      "epoch": 1.0934891485809684,
+      "grad_norm": 0.9788629412651062,
+      "learning_rate": 2.7736301283624e-05,
+      "loss": 0.581,
+      "step": 655
+    },
+    {
+      "epoch": 1.1018363939899833,
+      "grad_norm": 1.0459963083267212,
+      "learning_rate": 2.7692358069863455e-05,
+      "loss": 0.626,
+      "step": 660
+    },
+    {
+      "epoch": 1.1101836393989983,
+      "grad_norm": 1.0294015407562256,
+      "learning_rate": 2.7648027940338377e-05,
+      "loss": 0.561,
+      "step": 665
+    },
+    {
+      "epoch": 1.1185308848080133,
+      "grad_norm": 1.0476348400115967,
+      "learning_rate": 2.7603312246415142e-05,
+      "loss": 0.5558,
+      "step": 670
+    },
+    {
+      "epoch": 1.1268781302170283,
+      "grad_norm": 1.2524807453155518,
+      "learning_rate": 2.7558212351213744e-05,
+      "loss": 0.5823,
+      "step": 675
+    },
+    {
+      "epoch": 1.1352253756260433,
+      "grad_norm": 0.9370312094688416,
+      "learning_rate": 2.75127296295662e-05,
+      "loss": 0.606,
+      "step": 680
+    },
+    {
+      "epoch": 1.1435726210350585,
+      "grad_norm": 0.9592323899269104,
+      "learning_rate": 2.7466865467974686e-05,
+      "loss": 0.5904,
+      "step": 685
+    },
+    {
+      "epoch": 1.1519198664440735,
+      "grad_norm": 1.0453234910964966,
+      "learning_rate": 2.7420621264569247e-05,
+      "loss": 0.5774,
+      "step": 690
+    },
+    {
+      "epoch": 1.1602671118530885,
+      "grad_norm": 1.0608439445495605,
+      "learning_rate": 2.737399842906518e-05,
+      "loss": 0.5962,
+      "step": 695
+    },
+    {
+      "epoch": 1.1686143572621035,
+      "grad_norm": 1.0199201107025146,
+      "learning_rate": 2.7326998382720062e-05,
+      "loss": 0.5724,
+      "step": 700
+    },
+    {
+      "epoch": 1.1769616026711185,
+      "grad_norm": 1.007466435432434,
+      "learning_rate": 2.7279622558290412e-05,
+      "loss": 0.6032,
+      "step": 705
+    },
+    {
+      "epoch": 1.1853088480801335,
+      "grad_norm": 1.1127582788467407,
+      "learning_rate": 2.7231872399988042e-05,
+      "loss": 0.5602,
+      "step": 710
+    },
+    {
+      "epoch": 1.1936560934891487,
+      "grad_norm": 1.0738898515701294,
+      "learning_rate": 2.7183749363436e-05,
+      "loss": 0.5827,
+      "step": 715
+    },
+    {
+      "epoch": 1.2020033388981637,
+      "grad_norm": 1.0549622774124146,
+      "learning_rate": 2.7135254915624213e-05,
+      "loss": 0.5254,
+      "step": 720
+    },
+    {
+      "epoch": 1.2103505843071787,
+      "grad_norm": 1.1299083232879639,
+      "learning_rate": 2.7086390534864773e-05,
+      "loss": 0.5608,
+      "step": 725
+    },
+    {
+      "epoch": 1.2186978297161937,
+      "grad_norm": 0.998676598072052,
+      "learning_rate": 2.7037157710746855e-05,
+      "loss": 0.5304,
+      "step": 730
+    },
+    {
+      "epoch": 1.2270450751252087,
+      "grad_norm": 1.1849541664123535,
+      "learning_rate": 2.6987557944091327e-05,
+      "loss": 0.56,
+      "step": 735
+    },
+    {
+      "epoch": 1.2353923205342237,
+      "grad_norm": 1.0428630113601685,
+      "learning_rate": 2.6937592746904985e-05,
+      "loss": 0.5368,
+      "step": 740
+    },
+    {
+      "epoch": 1.2437395659432386,
+      "grad_norm": 1.0633243322372437,
+      "learning_rate": 2.6887263642334454e-05,
+      "loss": 0.5711,
+      "step": 745
+    },
+    {
+      "epoch": 1.2520868113522536,
+      "grad_norm": 1.0899866819381714,
+      "learning_rate": 2.683657216461978e-05,
+      "loss": 0.5749,
+      "step": 750
+    },
+    {
+      "epoch": 1.2604340567612689,
+      "grad_norm": 1.0628269910812378,
+      "learning_rate": 2.6785519859047644e-05,
+      "loss": 0.5492,
+      "step": 755
+    },
+    {
+      "epoch": 1.2687813021702838,
+      "grad_norm": 0.9858053922653198,
+      "learning_rate": 2.673410828190425e-05,
+      "loss": 0.6019,
+      "step": 760
+    },
+    {
+      "epoch": 1.2771285475792988,
+      "grad_norm": 1.1756415367126465,
+      "learning_rate": 2.6682339000427908e-05,
+      "loss": 0.5808,
+      "step": 765
+    },
+    {
+      "epoch": 1.2854757929883138,
+      "grad_norm": 0.9808685183525085,
+      "learning_rate": 2.6630213592761222e-05,
+      "loss": 0.5265,
+      "step": 770
+    },
+    {
+      "epoch": 1.2938230383973288,
+      "grad_norm": 1.0026417970657349,
+      "learning_rate": 2.6577733647903014e-05,
+      "loss": 0.5774,
+      "step": 775
+    },
+    {
+      "epoch": 1.302170283806344,
+      "grad_norm": 0.9454689621925354,
+      "learning_rate": 2.6524900765659858e-05,
+      "loss": 0.5496,
+      "step": 780
+    },
+    {
+      "epoch": 1.310517529215359,
+      "grad_norm": 1.0980932712554932,
+      "learning_rate": 2.6471716556597343e-05,
+      "loss": 0.5462,
+      "step": 785
+    },
+    {
+      "epoch": 1.318864774624374,
+      "grad_norm": 1.016065001487732,
+      "learning_rate": 2.6418182641990948e-05,
+      "loss": 0.5434,
+      "step": 790
+    },
+    {
+      "epoch": 1.327212020033389,
+      "grad_norm": 1.130270004272461,
+      "learning_rate": 2.6364300653776618e-05,
+      "loss": 0.5138,
+      "step": 795
+    },
+    {
+      "epoch": 1.335559265442404,
+      "grad_norm": 1.1002886295318604,
+      "learning_rate": 2.631007223450105e-05,
+      "loss": 0.5347,
+      "step": 800
+    },
+    {
+      "epoch": 1.343906510851419,
+      "grad_norm": 1.1232490539550781,
+      "learning_rate": 2.625549903727158e-05,
+      "loss": 0.541,
+      "step": 805
+    },
+    {
+      "epoch": 1.352253756260434,
+      "grad_norm": 0.9805702567100525,
+      "learning_rate": 2.620058272570581e-05,
+      "loss": 0.5234,
+      "step": 810
+    },
+    {
+      "epoch": 1.360601001669449,
+      "grad_norm": 1.0924123525619507,
+      "learning_rate": 2.61453249738809e-05,
+      "loss": 0.5339,
+      "step": 815
+    },
+    {
+      "epoch": 1.3689482470784642,
+      "grad_norm": 0.9931308627128601,
+      "learning_rate": 2.608972746628251e-05,
+      "loss": 0.5596,
+      "step": 820
+    },
+    {
+      "epoch": 1.3772954924874792,
+      "grad_norm": 1.1093590259552002,
+      "learning_rate": 2.6033791897753487e-05,
+      "loss": 0.5169,
+      "step": 825
+    },
+    {
+      "epoch": 1.3856427378964942,
+      "grad_norm": 1.104363203048706,
+      "learning_rate": 2.5977519973442164e-05,
+      "loss": 0.4898,
+      "step": 830
+    },
+    {
+      "epoch": 1.3939899833055092,
+      "grad_norm": 0.9823944568634033,
+      "learning_rate": 2.5920913408750402e-05,
+      "loss": 0.5376,
+      "step": 835
+    },
+    {
+      "epoch": 1.4023372287145242,
+      "grad_norm": 1.0676780939102173,
+      "learning_rate": 2.5863973929281288e-05,
+      "loss": 0.5472,
+      "step": 840
+    },
+    {
+      "epoch": 1.4106844741235394,
+      "grad_norm": 1.1943162679672241,
+      "learning_rate": 2.5806703270786526e-05,
+      "loss": 0.4788,
+      "step": 845
+    },
+    {
+      "epoch": 1.4190317195325544,
+      "grad_norm": 1.1296433210372925,
+      "learning_rate": 2.574910317911354e-05,
+      "loss": 0.4961,
+      "step": 850
+    },
+    {
+      "epoch": 1.4273789649415694,
+      "grad_norm": 1.1892937421798706,
+      "learning_rate": 2.5691175410152253e-05,
+      "loss": 0.4995,
+      "step": 855
+    },
+    {
+      "epoch": 1.4357262103505843,
+      "grad_norm": 1.0098474025726318,
+      "learning_rate": 2.5632921729781528e-05,
+      "loss": 0.4576,
+      "step": 860
+    },
+    {
+      "epoch": 1.4440734557595993,
+      "grad_norm": 0.9864776730537415,
+      "learning_rate": 2.5574343913815387e-05,
+      "loss": 0.5543,
+      "step": 865
+    },
+    {
+      "epoch": 1.4524207011686143,
+      "grad_norm": 1.1109187602996826,
+      "learning_rate": 2.551544374794883e-05,
+      "loss": 0.4911,
+      "step": 870
+    },
+    {
+      "epoch": 1.4607679465776293,
+      "grad_norm": 1.2205853462219238,
+      "learning_rate": 2.5456223027703427e-05,
+      "loss": 0.5013,
+      "step": 875
+    },
+    {
+      "epoch": 1.4691151919866443,
+      "grad_norm": 1.2574927806854248,
+      "learning_rate": 2.539668355837258e-05,
+      "loss": 0.4745,
+      "step": 880
+    },
+    {
+      "epoch": 1.4774624373956593,
+      "grad_norm": 1.2027695178985596,
+      "learning_rate": 2.533682715496649e-05,
+      "loss": 0.485,
+      "step": 885
+    },
+    {
+      "epoch": 1.4858096828046745,
+      "grad_norm": 1.032432198524475,
+      "learning_rate": 2.5276655642156814e-05,
+      "loss": 0.4859,
+      "step": 890
+    },
+    {
+      "epoch": 1.4941569282136895,
+      "grad_norm": 1.0579237937927246,
+      "learning_rate": 2.521617085422105e-05,
+      "loss": 0.489,
+      "step": 895
+    },
+    {
+      "epoch": 1.5025041736227045,
+      "grad_norm": 1.2205932140350342,
+      "learning_rate": 2.515537463498664e-05,
+      "loss": 0.4788,
+      "step": 900
+    },
+    {
+      "epoch": 1.5108514190317195,
+      "grad_norm": 0.9560028314590454,
+      "learning_rate": 2.509426883777473e-05,
+      "loss": 0.4796,
+      "step": 905
+    },
+    {
+      "epoch": 1.5191986644407347,
+      "grad_norm": 1.158149242401123,
+      "learning_rate": 2.5032855325343687e-05,
+      "loss": 0.4771,
+      "step": 910
+    },
+    {
+      "epoch": 1.5275459098497497,
+      "grad_norm": 1.2381256818771362,
+      "learning_rate": 2.497113596983232e-05,
+      "loss": 0.505,
+      "step": 915
+    },
+    {
+      "epoch": 1.5358931552587647,
+      "grad_norm": 1.0872339010238647,
+      "learning_rate": 2.4909112652702804e-05,
+      "loss": 0.4443,
+      "step": 920
+    },
+    {
+      "epoch": 1.5442404006677797,
+      "grad_norm": 1.1831974983215332,
+      "learning_rate": 2.4846787264683334e-05,
+      "loss": 0.4865,
+      "step": 925
+    },
+    {
+      "epoch": 1.5525876460767947,
+      "grad_norm": 1.3985413312911987,
+      "learning_rate": 2.478416170571046e-05,
+      "loss": 0.4794,
+      "step": 930
+    },
+    {
+      "epoch": 1.5609348914858097,
+      "grad_norm": 1.174689531326294,
+      "learning_rate": 2.472123788487121e-05,
+      "loss": 0.4824,
+      "step": 935
+    },
+    {
+      "epoch": 1.5692821368948247,
+      "grad_norm": 1.071053147315979,
+      "learning_rate": 2.4658017720344862e-05,
+      "loss": 0.4365,
+      "step": 940
+    },
+    {
+      "epoch": 1.5776293823038396,
+      "grad_norm": 1.0008587837219238,
+      "learning_rate": 2.459450313934447e-05,
+      "loss": 0.4488,
+      "step": 945
+    },
+    {
+      "epoch": 1.5859766277128546,
+      "grad_norm": 1.2077606916427612,
+      "learning_rate": 2.453069607805815e-05,
+      "loss": 0.4196,
+      "step": 950
+    },
+    {
+      "epoch": 1.5943238731218696,
+      "grad_norm": 1.0523104667663574,
+      "learning_rate": 2.446659848159001e-05,
+      "loss": 0.4244,
+      "step": 955
+    },
+    {
+      "epoch": 1.6026711185308848,
+      "grad_norm": 1.2364376783370972,
+      "learning_rate": 2.4402212303900877e-05,
+      "loss": 0.4857,
+      "step": 960
+    },
+    {
+      "epoch": 1.6110183639398998,
+      "grad_norm": 0.9511405229568481,
+      "learning_rate": 2.433753950774875e-05,
+      "loss": 0.4673,
+      "step": 965
+    },
+    {
+      "epoch": 1.6193656093489148,
+      "grad_norm": 1.0465832948684692,
+      "learning_rate": 2.427258206462892e-05,
+      "loss": 0.4204,
+      "step": 970
+    },
+    {
+      "epoch": 1.62771285475793,
+      "grad_norm": 1.0256762504577637,
+      "learning_rate": 2.4207341954713925e-05,
+      "loss": 0.4369,
+      "step": 975
+    },
+    {
+      "epoch": 1.636060100166945,
+      "grad_norm": 1.137253761291504,
+      "learning_rate": 2.414182116679315e-05,
+      "loss": 0.4575,
+      "step": 980
+    },
+    {
+      "epoch": 1.64440734557596,
+      "grad_norm": 1.1028913259506226,
+      "learning_rate": 2.4076021698212208e-05,
+      "loss": 0.5011,
+      "step": 985
+    },
+    {
+      "epoch": 1.652754590984975,
+      "grad_norm": 1.0733743906021118,
+      "learning_rate": 2.4009945554812064e-05,
+      "loss": 0.4618,
+      "step": 990
+    },
+    {
+      "epoch": 1.66110183639399,
+      "grad_norm": 1.1975722312927246,
+      "learning_rate": 2.394359475086787e-05,
+      "loss": 0.4332,
+      "step": 995
+    },
+    {
+      "epoch": 1.669449081803005,
+      "grad_norm": 1.1416840553283691,
+      "learning_rate": 2.387697130902757e-05,
+      "loss": 0.4704,
+      "step": 1000
+    },
+    {
+      "epoch": 1.67779632721202,
+      "grad_norm": 1.1874338388442993,
+      "learning_rate": 2.3810077260250268e-05,
+      "loss": 0.442,
+      "step": 1005
+    },
+    {
+      "epoch": 1.686143572621035,
+      "grad_norm": 0.9999452829360962,
+      "learning_rate": 2.3742914643744255e-05,
+      "loss": 0.4364,
+      "step": 1010
+    },
+    {
+      "epoch": 1.69449081803005,
+      "grad_norm": 1.1475085020065308,
+      "learning_rate": 2.367548550690491e-05,
+      "loss": 0.428,
+      "step": 1015
+    },
+    {
+      "epoch": 1.702838063439065,
+      "grad_norm": 1.049939513206482,
+      "learning_rate": 2.3607791905252232e-05,
+      "loss": 0.4365,
+      "step": 1020
+    },
+    {
+      "epoch": 1.7111853088480802,
+      "grad_norm": 1.1441938877105713,
+      "learning_rate": 2.3539835902368238e-05,
+      "loss": 0.4367,
+      "step": 1025
+    },
+    {
+      "epoch": 1.7195325542570952,
+      "grad_norm": 1.03773033618927,
+      "learning_rate": 2.347161956983399e-05,
+      "loss": 0.4067,
+      "step": 1030
+    },
+    {
+      "epoch": 1.7278797996661102,
+      "grad_norm": 1.2723736763000488,
+      "learning_rate": 2.3403144987166505e-05,
+      "loss": 0.4251,
+      "step": 1035
+    },
+    {
+      "epoch": 1.7362270450751254,
+      "grad_norm": 1.0961979627609253,
+      "learning_rate": 2.333441424175532e-05,
+      "loss": 0.3885,
+      "step": 1040
+    },
+    {
+      "epoch": 1.7445742904841404,
+      "grad_norm": 1.002756953239441,
+      "learning_rate": 2.3265429428798878e-05,
+      "loss": 0.4228,
+      "step": 1045
+    },
+    {
+      "epoch": 1.7529215358931554,
+      "grad_norm": 1.1268832683563232,
+      "learning_rate": 2.319619265124067e-05,
+      "loss": 0.4268,
+      "step": 1050
+    },
+    {
+      "epoch": 1.7612687813021703,
+      "grad_norm": 1.1599644422531128,
+      "learning_rate": 2.3126706019705093e-05,
+      "loss": 0.425,
+      "step": 1055
+    },
+    {
+      "epoch": 1.7696160267111853,
+      "grad_norm": 1.0869255065917969,
+      "learning_rate": 2.3056971652433145e-05,
+      "loss": 0.4035,
+      "step": 1060
+    },
+    {
+      "epoch": 1.7779632721202003,
+      "grad_norm": 1.1428903341293335,
+      "learning_rate": 2.298699167521783e-05,
+      "loss": 0.4355,
+      "step": 1065
+    },
+    {
+      "epoch": 1.7863105175292153,
+      "grad_norm": 1.090363621711731,
+      "learning_rate": 2.2916768221339372e-05,
+      "loss": 0.3947,
+      "step": 1070
+    },
+    {
+      "epoch": 1.7946577629382303,
+      "grad_norm": 1.0996553897857666,
+      "learning_rate": 2.284630343150016e-05,
+      "loss": 0.3775,
+      "step": 1075
+    },
+    {
+      "epoch": 1.8030050083472453,
+      "grad_norm": 1.1129885911941528,
+      "learning_rate": 2.2775599453759514e-05,
+      "loss": 0.4478,
+      "step": 1080
+    },
+    {
+      "epoch": 1.8113522537562603,
+      "grad_norm": 1.147131085395813,
+      "learning_rate": 2.27046584434682e-05,
+      "loss": 0.3752,
+      "step": 1085
+    },
+    {
+      "epoch": 1.8196994991652755,
+      "grad_norm": 1.1744565963745117,
+      "learning_rate": 2.263348256320271e-05,
+      "loss": 0.4113,
+      "step": 1090
+    },
+    {
+      "epoch": 1.8280467445742905,
+      "grad_norm": 1.2666568756103516,
+      "learning_rate": 2.2562073982699346e-05,
+      "loss": 0.3871,
+      "step": 1095
+    },
+    {
+      "epoch": 1.8363939899833055,
+      "grad_norm": 1.2098952531814575,
+      "learning_rate": 2.2490434878788094e-05,
+      "loss": 0.4084,
+      "step": 1100
+    },
+    {
+      "epoch": 1.8447412353923205,
+      "grad_norm": 1.218492865562439,
+      "learning_rate": 2.2418567435326233e-05,
+      "loss": 0.4108,
+      "step": 1105
+    },
+    {
+      "epoch": 1.8530884808013357,
+      "grad_norm": 1.3567547798156738,
+      "learning_rate": 2.234647384313179e-05,
+      "loss": 0.3972,
+      "step": 1110
+    },
+    {
+      "epoch": 1.8614357262103507,
+      "grad_norm": 1.090145230293274,
+      "learning_rate": 2.2274156299916746e-05,
+      "loss": 0.3838,
+      "step": 1115
+    },
+    {
+      "epoch": 1.8697829716193657,
+      "grad_norm": 1.1858583688735962,
+      "learning_rate": 2.2201617010220023e-05,
+      "loss": 0.3887,
+      "step": 1120
+    },
+    {
+      "epoch": 1.8781302170283807,
+      "grad_norm": 1.036993384361267,
+      "learning_rate": 2.2128858185340322e-05,
+      "loss": 0.4284,
+      "step": 1125
+    },
+    {
+      "epoch": 1.8864774624373957,
+      "grad_norm": 1.1040432453155518,
+      "learning_rate": 2.2055882043268663e-05,
+      "loss": 0.3715,
+      "step": 1130
+    },
+    {
+      "epoch": 1.8948247078464107,
+      "grad_norm": 0.9753995537757874,
+      "learning_rate": 2.198269080862081e-05,
+      "loss": 0.3748,
+      "step": 1135
+    },
+    {
+      "epoch": 1.9031719532554257,
+      "grad_norm": 1.0646343231201172,
+      "learning_rate": 2.1909286712569444e-05,
+      "loss": 0.4245,
+      "step": 1140
+    },
+    {
+      "epoch": 1.9115191986644406,
+      "grad_norm": 1.4246667623519897,
+      "learning_rate": 2.1835671992776132e-05,
+      "loss": 0.3573,
+      "step": 1145
+    },
+    {
+      "epoch": 1.9198664440734556,
+      "grad_norm": 1.0584313869476318,
+      "learning_rate": 2.176184889332314e-05,
+      "loss": 0.3806,
+      "step": 1150
+    },
+    {
+      "epoch": 1.9282136894824706,
+      "grad_norm": 1.0937577486038208,
+      "learning_rate": 2.1687819664644993e-05,
+      "loss": 0.35,
+      "step": 1155
+    },
+    {
+      "epoch": 1.9365609348914858,
+      "grad_norm": 1.183713436126709,
+      "learning_rate": 2.161358656345992e-05,
+      "loss": 0.3668,
+      "step": 1160
+    },
+    {
+      "epoch": 1.9449081803005008,
+      "grad_norm": 1.119342565536499,
+      "learning_rate": 2.153915185270101e-05,
+      "loss": 0.388,
+      "step": 1165
+    },
+    {
+      "epoch": 1.9532554257095158,
+      "grad_norm": 1.2510145902633667,
+      "learning_rate": 2.1464517801447244e-05,
+      "loss": 0.3594,
+      "step": 1170
+    },
+    {
+      "epoch": 1.961602671118531,
+      "grad_norm": 1.1926363706588745,
+      "learning_rate": 2.138968668485434e-05,
+      "loss": 0.4232,
+      "step": 1175
+    },
+    {
+      "epoch": 1.969949916527546,
+      "grad_norm": 1.211121916770935,
+      "learning_rate": 2.1314660784085387e-05,
+      "loss": 0.3796,
+      "step": 1180
+    },
+    {
+      "epoch": 1.978297161936561,
+      "grad_norm": 1.1046868562698364,
+      "learning_rate": 2.123944238624129e-05,
+      "loss": 0.348,
+      "step": 1185
+    },
+    {
+      "epoch": 1.986644407345576,
+      "grad_norm": 1.2281684875488281,
+      "learning_rate": 2.1164033784291092e-05,
+      "loss": 0.4116,
+      "step": 1190
+    },
+    {
+      "epoch": 1.994991652754591,
+      "grad_norm": 1.1758081912994385,
+      "learning_rate": 2.1088437277002014e-05,
+      "loss": 0.3496,
+      "step": 1195
+    },
+    {
+      "epoch": 2.003338898163606,
+      "grad_norm": 1.0964444875717163,
+      "learning_rate": 2.101265516886943e-05,
+      "loss": 0.337,
+      "step": 1200
+    },
+    {
+      "epoch": 2.011686143572621,
+      "grad_norm": 1.2007455825805664,
+      "learning_rate": 2.09366897700466e-05,
+      "loss": 0.3296,
+      "step": 1205
+    },
+    {
+      "epoch": 2.020033388981636,
+      "grad_norm": 1.2061660289764404,
+      "learning_rate": 2.0860543396274233e-05,
+      "loss": 0.2743,
+      "step": 1210
+    },
+    {
+      "epoch": 2.028380634390651,
+      "grad_norm": 1.1202778816223145,
+      "learning_rate": 2.0784218368809918e-05,
+      "loss": 0.3055,
+      "step": 1215
+    },
+    {
+      "epoch": 2.036727879799666,
+      "grad_norm": 1.1539779901504517,
+      "learning_rate": 2.0707717014357335e-05,
+      "loss": 0.2914,
+      "step": 1220
+    },
+    {
+      "epoch": 2.045075125208681,
+      "grad_norm": 1.307861089706421,
+      "learning_rate": 2.0631041664995367e-05,
+      "loss": 0.3264,
+      "step": 1225
+    },
+    {
+      "epoch": 2.053422370617696,
+      "grad_norm": 1.0708540678024292,
+      "learning_rate": 2.055419465810696e-05,
+      "loss": 0.2901,
+      "step": 1230
+    },
+    {
+      "epoch": 2.0617696160267114,
+      "grad_norm": 1.087788701057434,
+      "learning_rate": 2.0477178336307904e-05,
+      "loss": 0.3046,
+      "step": 1235
+    },
+    {
+      "epoch": 2.0701168614357264,
+      "grad_norm": 1.05051851272583,
+      "learning_rate": 2.039999504737541e-05,
+      "loss": 0.2583,
+      "step": 1240
+    },
+    {
+      "epoch": 2.0784641068447414,
+      "grad_norm": 1.3561017513275146,
+      "learning_rate": 2.0322647144176547e-05,
+      "loss": 0.2748,
+      "step": 1245
+    },
+    {
+      "epoch": 2.0868113522537564,
+      "grad_norm": 1.234081506729126,
+      "learning_rate": 2.0245136984596494e-05,
+      "loss": 0.297,
+      "step": 1250
+    },
+    {
+      "epoch": 2.0951585976627713,
+      "grad_norm": 1.157702922821045,
+      "learning_rate": 2.01674669314667e-05,
+      "loss": 0.3029,
+      "step": 1255
+    },
+    {
+      "epoch": 2.1035058430717863,
+      "grad_norm": 1.1084997653961182,
+      "learning_rate": 2.0089639352492807e-05,
+      "loss": 0.3053,
+      "step": 1260
+    },
+    {
+      "epoch": 2.1118530884808013,
+      "grad_norm": 1.0160917043685913,
+      "learning_rate": 2.001165662018252e-05,
+      "loss": 0.2435,
+      "step": 1265
+    },
+    {
+      "epoch": 2.1202003338898163,
+      "grad_norm": 1.20888352394104,
+      "learning_rate": 1.9933521111773257e-05,
+      "loss": 0.2823,
+      "step": 1270
+    },
+    {
+      "epoch": 2.1285475792988313,
+      "grad_norm": 1.155145287513733,
+      "learning_rate": 1.9855235209159674e-05,
+      "loss": 0.2955,
+      "step": 1275
+    },
+    {
+      "epoch": 2.1368948247078463,
+      "grad_norm": 1.166378140449524,
+      "learning_rate": 1.9776801298821087e-05,
+      "loss": 0.292,
+      "step": 1280
+    },
+    {
+      "epoch": 2.1452420701168613,
+      "grad_norm": 1.082502841949463,
+      "learning_rate": 1.9698221771748677e-05,
+      "loss": 0.2896,
+      "step": 1285
+    },
+    {
+      "epoch": 2.1535893155258763,
+      "grad_norm": 1.045149326324463,
+      "learning_rate": 1.9619499023372656e-05,
+      "loss": 0.2915,
+      "step": 1290
+    },
+    {
+      "epoch": 2.1619365609348913,
+      "grad_norm": 1.1643922328948975,
+      "learning_rate": 1.9540635453489188e-05,
+      "loss": 0.2715,
+      "step": 1295
+    },
+    {
+      "epoch": 2.1702838063439067,
+      "grad_norm": 1.145524024963379,
+      "learning_rate": 1.9461633466187283e-05,
+      "loss": 0.2686,
+      "step": 1300
+    },
+    {
+      "epoch": 2.1786310517529217,
+      "grad_norm": 1.1801557540893555,
+      "learning_rate": 1.9382495469775476e-05,
+      "loss": 0.2985,
+      "step": 1305
+    },
+    {
+      "epoch": 2.1869782971619367,
+      "grad_norm": 1.1388278007507324,
+      "learning_rate": 1.930322387670843e-05,
+      "loss": 0.2872,
+      "step": 1310
+    },
+    {
+      "epoch": 2.1953255425709517,
+      "grad_norm": 1.1367162466049194,
+      "learning_rate": 1.922382110351338e-05,
+      "loss": 0.2686,
+      "step": 1315
+    },
+    {
+      "epoch": 2.2036727879799667,
+      "grad_norm": 1.258169174194336,
+      "learning_rate": 1.9144289570716494e-05,
+      "loss": 0.2435,
+      "step": 1320
+    },
+    {
+      "epoch": 2.2120200333889817,
+      "grad_norm": 1.1344308853149414,
+      "learning_rate": 1.906463170276905e-05,
+      "loss": 0.2975,
+      "step": 1325
+    },
+    {
+      "epoch": 2.2203672787979967,
+      "grad_norm": 1.1693185567855835,
+      "learning_rate": 1.8984849927973544e-05,
+      "loss": 0.2712,
+      "step": 1330
+    },
+    {
+      "epoch": 2.2287145242070117,
+      "grad_norm": 1.0323889255523682,
+      "learning_rate": 1.890494667840969e-05,
+      "loss": 0.2941,
+      "step": 1335
+    },
+    {
+      "epoch": 2.2370617696160267,
+      "grad_norm": 1.0409553050994873,
+      "learning_rate": 1.882492438986023e-05,
+      "loss": 0.2622,
+      "step": 1340
+    },
+    {
+      "epoch": 2.2454090150250416,
+      "grad_norm": 1.1228262186050415,
+      "learning_rate": 1.874478550173673e-05,
+      "loss": 0.284,
+      "step": 1345
+    },
+    {
+      "epoch": 2.2537562604340566,
+      "grad_norm": 0.9944053292274475,
+      "learning_rate": 1.8664532457005173e-05,
+      "loss": 0.2577,
+      "step": 1350
+    },
+    {
+      "epoch": 2.2621035058430716,
+      "grad_norm": 1.2135941982269287,
+      "learning_rate": 1.8584167702111533e-05,
+      "loss": 0.2709,
+      "step": 1355
+    },
+    {
+      "epoch": 2.2704507512520866,
+      "grad_norm": 1.1230733394622803,
+      "learning_rate": 1.8503693686907157e-05,
+      "loss": 0.2639,
+      "step": 1360
+    },
+    {
+      "epoch": 2.278797996661102,
+      "grad_norm": 1.2838828563690186,
+      "learning_rate": 1.842311286457411e-05,
+      "loss": 0.2661,
+      "step": 1365
+    },
+    {
+      "epoch": 2.287145242070117,
+      "grad_norm": 1.2004976272583008,
+      "learning_rate": 1.8342427691550363e-05,
+      "loss": 0.2421,
+      "step": 1370
+    },
+    {
+      "epoch": 2.295492487479132,
+      "grad_norm": 1.2496306896209717,
+      "learning_rate": 1.8261640627454957e-05,
+      "loss": 0.26,
+      "step": 1375
+    },
+    {
+      "epoch": 2.303839732888147,
+      "grad_norm": 1.0600461959838867,
+      "learning_rate": 1.8180754135012984e-05,
+      "loss": 0.2702,
+      "step": 1380
+    },
+    {
+      "epoch": 2.312186978297162,
+      "grad_norm": 1.3516604900360107,
+      "learning_rate": 1.8099770679980513e-05,
+      "loss": 0.2805,
+      "step": 1385
+    },
+    {
+      "epoch": 2.320534223706177,
+      "grad_norm": 1.1592249870300293,
+      "learning_rate": 1.801869273106945e-05,
+      "loss": 0.2577,
+      "step": 1390
+    },
+    {
+      "epoch": 2.328881469115192,
+      "grad_norm": 1.0440796613693237,
+      "learning_rate": 1.7937522759872253e-05,
+      "loss": 0.2616,
+      "step": 1395
+    },
+    {
+      "epoch": 2.337228714524207,
+      "grad_norm": 1.150465488433838,
+      "learning_rate": 1.7856263240786624e-05,
+      "loss": 0.3081,
+      "step": 1400
+    },
+    {
+      "epoch": 2.345575959933222,
+      "grad_norm": 1.055363416671753,
+      "learning_rate": 1.7774916650940038e-05,
+      "loss": 0.2664,
+      "step": 1405
+    },
+    {
+      "epoch": 2.353923205342237,
+      "grad_norm": 1.064753532409668,
+      "learning_rate": 1.7693485470114245e-05,
+      "loss": 0.258,
+      "step": 1410
+    },
+    {
+      "epoch": 2.362270450751252,
+      "grad_norm": 1.169392466545105,
+      "learning_rate": 1.7611972180669692e-05,
+      "loss": 0.2544,
+      "step": 1415
+    },
+    {
+      "epoch": 2.370617696160267,
+      "grad_norm": 1.180547833442688,
+      "learning_rate": 1.7530379267469833e-05,
+      "loss": 0.2581,
+      "step": 1420
+    },
+    {
+      "epoch": 2.378964941569282,
+      "grad_norm": 1.127563238143921,
+      "learning_rate": 1.7448709217805387e-05,
+      "loss": 0.2474,
+      "step": 1425
+    },
+    {
+      "epoch": 2.3873121869782974,
+      "grad_norm": 1.012845516204834,
+      "learning_rate": 1.7366964521318502e-05,
+      "loss": 0.2861,
+      "step": 1430
+    },
+    {
+      "epoch": 2.395659432387312,
+      "grad_norm": 1.2285345792770386,
+      "learning_rate": 1.7285147669926883e-05,
+      "loss": 0.283,
+      "step": 1435
+    },
+    {
+      "epoch": 2.4040066777963274,
+      "grad_norm": 1.0634362697601318,
+      "learning_rate": 1.720326115774781e-05,
+      "loss": 0.2574,
+      "step": 1440
+    },
+    {
+      "epoch": 2.4123539232053424,
+      "grad_norm": 1.1493803262710571,
+      "learning_rate": 1.7121307481022117e-05,
+      "loss": 0.2742,
+      "step": 1445
+    },
+    {
+      "epoch": 2.4207011686143574,
+      "grad_norm": 1.0936006307601929,
+      "learning_rate": 1.703928913803809e-05,
+      "loss": 0.2665,
+      "step": 1450
+    },
+    {
+      "epoch": 2.4290484140233723,
+      "grad_norm": 1.2558296918869019,
+      "learning_rate": 1.695720862905531e-05,
+      "loss": 0.2586,
+      "step": 1455
+    },
+    {
+      "epoch": 2.4373956594323873,
+      "grad_norm": 1.0479841232299805,
+      "learning_rate": 1.6875068456228435e-05,
+      "loss": 0.2355,
+      "step": 1460
+    },
+    {
+      "epoch": 2.4457429048414023,
+      "grad_norm": 1.185439109802246,
+      "learning_rate": 1.679287112353093e-05,
+      "loss": 0.2604,
+      "step": 1465
+    },
+    {
+      "epoch": 2.4540901502504173,
+      "grad_norm": 1.0795245170593262,
+      "learning_rate": 1.6710619136678725e-05,
+      "loss": 0.2397,
+      "step": 1470
+    },
+    {
+      "epoch": 2.4624373956594323,
+      "grad_norm": 1.1054483652114868,
+      "learning_rate": 1.6628315003053834e-05,
+      "loss": 0.2342,
+      "step": 1475
+    },
+    {
+      "epoch": 2.4707846410684473,
+      "grad_norm": 1.1193106174468994,
+      "learning_rate": 1.6545961231627923e-05,
+      "loss": 0.2405,
+      "step": 1480
+    },
+    {
+      "epoch": 2.4791318864774623,
+      "grad_norm": 1.1593073606491089,
+      "learning_rate": 1.6463560332885832e-05,
+      "loss": 0.2557,
+      "step": 1485
+    },
+    {
+      "epoch": 2.4874791318864773,
+      "grad_norm": 0.9820372462272644,
+      "learning_rate": 1.638111481874903e-05,
+      "loss": 0.2248,
+      "step": 1490
+    },
+    {
+      "epoch": 2.4958263772954927,
+      "grad_norm": 1.1056580543518066,
+      "learning_rate": 1.6298627202499042e-05,
+      "loss": 0.2298,
+      "step": 1495
+    },
+    {
+      "epoch": 2.5041736227045073,
+      "grad_norm": 1.265946388244629,
+      "learning_rate": 1.6216099998700854e-05,
+      "loss": 0.2252,
+      "step": 1500
+    },
+    {
+      "epoch": 2.5125208681135227,
+      "grad_norm": 1.22787344455719,
+      "learning_rate": 1.613353572312623e-05,
+      "loss": 0.2285,
+      "step": 1505
+    },
+    {
+      "epoch": 2.5208681135225377,
+      "grad_norm": 1.1988939046859741,
+      "learning_rate": 1.6050936892677053e-05,
+      "loss": 0.2588,
+      "step": 1510
+    },
+    {
+      "epoch": 2.5292153589315527,
+      "grad_norm": 1.1179131269454956,
+      "learning_rate": 1.5968306025308573e-05,
+      "loss": 0.2396,
+      "step": 1515
+    },
+    {
+      "epoch": 2.5375626043405677,
+      "grad_norm": 1.1556321382522583,
+      "learning_rate": 1.588564563995265e-05,
+      "loss": 0.2257,
+      "step": 1520
+    },
+    {
+      "epoch": 2.5459098497495827,
+      "grad_norm": 1.1104613542556763,
+      "learning_rate": 1.5802958256440983e-05,
+      "loss": 0.2525,
+      "step": 1525
+    },
+    {
+      "epoch": 2.5542570951585977,
+      "grad_norm": 1.3291963338851929,
+      "learning_rate": 1.572024639542829e-05,
+      "loss": 0.2477,
+      "step": 1530
+    },
+    {
+      "epoch": 2.5626043405676127,
+      "grad_norm": 1.1854147911071777,
+      "learning_rate": 1.5637512578315465e-05,
+      "loss": 0.222,
+      "step": 1535
+    },
+    {
+      "epoch": 2.5709515859766277,
+      "grad_norm": 1.1772565841674805,
+      "learning_rate": 1.555475932717271e-05,
+      "loss": 0.2467,
+      "step": 1540
+    },
+    {
+      "epoch": 2.5792988313856426,
+      "grad_norm": 1.2394180297851562,
+      "learning_rate": 1.5471989164662653e-05,
+      "loss": 0.2339,
+      "step": 1545
+    },
+    {
+      "epoch": 2.5876460767946576,
+      "grad_norm": 1.1910455226898193,
+      "learning_rate": 1.5389204613963456e-05,
+      "loss": 0.2369,
+      "step": 1550
+    },
+    {
+      "epoch": 2.5959933222036726,
+      "grad_norm": 1.0397979021072388,
+      "learning_rate": 1.53064081986919e-05,
+      "loss": 0.2484,
+      "step": 1555
+    },
+    {
+      "epoch": 2.604340567612688,
+      "grad_norm": 1.131709337234497,
+      "learning_rate": 1.5223602442826438e-05,
+      "loss": 0.2163,
+      "step": 1560
+    },
+    {
+      "epoch": 2.6126878130217026,
+      "grad_norm": 1.3339247703552246,
+      "learning_rate": 1.5140789870630264e-05,
+      "loss": 0.2534,
+      "step": 1565
+    },
+    {
+      "epoch": 2.621035058430718,
+      "grad_norm": 1.4098116159439087,
+      "learning_rate": 1.505797300657436e-05,
+      "loss": 0.2434,
+      "step": 1570
+    },
+    {
+      "epoch": 2.629382303839733,
+      "grad_norm": 1.2102770805358887,
+      "learning_rate": 1.4975154375260554e-05,
+      "loss": 0.2169,
+      "step": 1575
+    },
+    {
+      "epoch": 2.637729549248748,
+      "grad_norm": 1.1393612623214722,
+      "learning_rate": 1.489233650134453e-05,
+      "loss": 0.2204,
+      "step": 1580
+    },
+    {
+      "epoch": 2.646076794657763,
+      "grad_norm": 1.4402921199798584,
+      "learning_rate": 1.4809521909458896e-05,
+      "loss": 0.2354,
+      "step": 1585
+    },
+    {
+      "epoch": 2.654424040066778,
+      "grad_norm": 1.074196696281433,
+      "learning_rate": 1.4726713124136207e-05,
+      "loss": 0.2147,
+      "step": 1590
+    },
+    {
+      "epoch": 2.662771285475793,
+      "grad_norm": 1.1807732582092285,
+      "learning_rate": 1.4643912669732006e-05,
+      "loss": 0.2403,
+      "step": 1595
+    },
+    {
+      "epoch": 2.671118530884808,
+      "grad_norm": 1.1444352865219116,
+      "learning_rate": 1.4561123070347872e-05,
+      "loss": 0.2298,
+      "step": 1600
+    },
+    {
+      "epoch": 2.679465776293823,
+      "grad_norm": 1.2918682098388672,
+      "learning_rate": 1.4478346849754495e-05,
+      "loss": 0.2269,
+      "step": 1605
+    },
+    {
+      "epoch": 2.687813021702838,
+      "grad_norm": 1.2147592306137085,
+      "learning_rate": 1.4395586531314711e-05,
+      "loss": 0.2525,
+      "step": 1610
+    },
+    {
+      "epoch": 2.696160267111853,
+      "grad_norm": 1.0803178548812866,
+      "learning_rate": 1.431284463790659e-05,
+      "loss": 0.2604,
+      "step": 1615
+    },
+    {
+      "epoch": 2.704507512520868,
+      "grad_norm": 1.1420807838439941,
+      "learning_rate": 1.4230123691846536e-05,
+      "loss": 0.229,
+      "step": 1620
+    },
+    {
+      "epoch": 2.7128547579298834,
+      "grad_norm": 1.0387948751449585,
+      "learning_rate": 1.4147426214812386e-05,
+      "loss": 0.2177,
+      "step": 1625
+    },
+    {
+      "epoch": 2.721202003338898,
+      "grad_norm": 1.08343505859375,
+      "learning_rate": 1.4064754727766553e-05,
+      "loss": 0.2216,
+      "step": 1630
+    },
+    {
+      "epoch": 2.7295492487479134,
+      "grad_norm": 1.3237481117248535,
+      "learning_rate": 1.3982111750879158e-05,
+      "loss": 0.2122,
+      "step": 1635
+    },
+    {
+      "epoch": 2.7378964941569284,
+      "grad_norm": 1.0259819030761719,
+      "learning_rate": 1.3899499803451219e-05,
+      "loss": 0.1989,
+      "step": 1640
+    },
+    {
+      "epoch": 2.7462437395659434,
+      "grad_norm": 1.110953688621521,
+      "learning_rate": 1.3816921403837837e-05,
+      "loss": 0.2167,
+      "step": 1645
+    },
+    {
+      "epoch": 2.7545909849749584,
+      "grad_norm": 1.1509097814559937,
+      "learning_rate": 1.373437906937146e-05,
+      "loss": 0.2043,
+      "step": 1650
+    },
+    {
+      "epoch": 2.7629382303839733,
+      "grad_norm": 1.3061741590499878,
+      "learning_rate": 1.3651875316285098e-05,
+      "loss": 0.2166,
+      "step": 1655
+    },
+    {
+      "epoch": 2.7712854757929883,
+      "grad_norm": 1.1735879182815552,
+      "learning_rate": 1.3569412659635648e-05,
+      "loss": 0.2195,
+      "step": 1660
+    },
+    {
+      "epoch": 2.7796327212020033,
+      "grad_norm": 1.2539076805114746,
+      "learning_rate": 1.3486993613227216e-05,
+      "loss": 0.2376,
+      "step": 1665
+    },
+    {
+      "epoch": 2.7879799666110183,
+      "grad_norm": 1.204086184501648,
+      "learning_rate": 1.3404620689534491e-05,
+      "loss": 0.2096,
+      "step": 1670
+    },
+    {
+      "epoch": 2.7963272120200333,
+      "grad_norm": 1.2498582601547241,
+      "learning_rate": 1.3322296399626143e-05,
+      "loss": 0.2396,
+      "step": 1675
+    },
+    {
+      "epoch": 2.8046744574290483,
+      "grad_norm": 1.021296501159668,
+      "learning_rate": 1.3240023253088286e-05,
+      "loss": 0.1833,
+      "step": 1680
+    },
+    {
+      "epoch": 2.8130217028380633,
+      "grad_norm": 1.1932337284088135,
+      "learning_rate": 1.3157803757947974e-05,
+      "loss": 0.229,
+      "step": 1685
+    },
+    {
+      "epoch": 2.8213689482470787,
+      "grad_norm": 1.1836450099945068,
+      "learning_rate": 1.3075640420596743e-05,
+      "loss": 0.1927,
+      "step": 1690
+    },
+    {
+      "epoch": 2.8297161936560933,
+      "grad_norm": 1.1409358978271484,
+      "learning_rate": 1.2993535745714203e-05,
+      "loss": 0.2148,
+      "step": 1695
+    },
+    {
+      "epoch": 2.8380634390651087,
+      "grad_norm": 1.243720293045044,
+      "learning_rate": 1.2911492236191691e-05,
+      "loss": 0.2128,
+      "step": 1700
+    },
+    {
+      "epoch": 2.8464106844741233,
+      "grad_norm": 1.1428468227386475,
+      "learning_rate": 1.2829512393055972e-05,
+      "loss": 0.2175,
+      "step": 1705
+    },
+    {
+      "epoch": 2.8547579298831387,
+      "grad_norm": 1.0778887271881104,
+      "learning_rate": 1.2747598715392991e-05,
+      "loss": 0.2189,
+      "step": 1710
+    },
+    {
+      "epoch": 2.8631051752921537,
+      "grad_norm": 1.095584511756897,
+      "learning_rate": 1.2665753700271697e-05,
+      "loss": 0.2003,
+      "step": 1715
+    },
+    {
+      "epoch": 2.8714524207011687,
+      "grad_norm": 1.1165109872817993,
+      "learning_rate": 1.2583979842667916e-05,
+      "loss": 0.1878,
+      "step": 1720
+    },
+    {
+      "epoch": 2.8797996661101837,
+      "grad_norm": 1.0806440114974976,
+      "learning_rate": 1.2502279635388304e-05,
+      "loss": 0.1922,
+      "step": 1725
+    },
+    {
+      "epoch": 2.8881469115191987,
+      "grad_norm": 1.2143396139144897,
+      "learning_rate": 1.2420655568994345e-05,
+      "loss": 0.1986,
+      "step": 1730
+    },
+    {
+      "epoch": 2.8964941569282137,
+      "grad_norm": 1.1222453117370605,
+      "learning_rate": 1.2339110131726426e-05,
+      "loss": 0.2256,
+      "step": 1735
+    },
+    {
+      "epoch": 2.9048414023372287,
+      "grad_norm": 1.2317746877670288,
+      "learning_rate": 1.2257645809428014e-05,
+      "loss": 0.1881,
+      "step": 1740
+    },
+    {
+      "epoch": 2.9131886477462436,
+      "grad_norm": 1.1264094114303589,
+      "learning_rate": 1.2176265085469834e-05,
+      "loss": 0.1926,
+      "step": 1745
+    },
+    {
+      "epoch": 2.9215358931552586,
+      "grad_norm": 1.2146402597427368,
+      "learning_rate": 1.209497044067419e-05,
+      "loss": 0.2102,
+      "step": 1750
+    },
+    {
+      "epoch": 2.9298831385642736,
+      "grad_norm": 1.0171878337860107,
+      "learning_rate": 1.2013764353239337e-05,
+      "loss": 0.2061,
+      "step": 1755
+    },
+    {
+      "epoch": 2.9382303839732886,
+      "grad_norm": 1.2336195707321167,
+      "learning_rate": 1.1932649298663947e-05,
+      "loss": 0.1985,
+      "step": 1760
+    },
+    {
+      "epoch": 2.946577629382304,
+      "grad_norm": 1.179914951324463,
+      "learning_rate": 1.1851627749671613e-05,
+      "loss": 0.1978,
+      "step": 1765
+    },
+    {
+      "epoch": 2.9549248747913186,
+      "grad_norm": 1.1733237504959106,
+      "learning_rate": 1.17707021761355e-05,
+      "loss": 0.1945,
+      "step": 1770
+    },
+    {
+      "epoch": 2.963272120200334,
+      "grad_norm": 1.1671475172042847,
+      "learning_rate": 1.1689875045003033e-05,
+      "loss": 0.1801,
+      "step": 1775
+    },
+    {
+      "epoch": 2.971619365609349,
+      "grad_norm": 1.1967194080352783,
+      "learning_rate": 1.1609148820220724e-05,
+      "loss": 0.2052,
+      "step": 1780
+    },
+    {
+      "epoch": 2.979966611018364,
+      "grad_norm": 1.0906327962875366,
+      "learning_rate": 1.1528525962659023e-05,
+      "loss": 0.18,
+      "step": 1785
+    },
+    {
+      "epoch": 2.988313856427379,
+      "grad_norm": 1.1686508655548096,
+      "learning_rate": 1.144800893003732e-05,
+      "loss": 0.203,
+      "step": 1790
+    },
+    {
+      "epoch": 2.996661101836394,
+      "grad_norm": 1.2239857912063599,
+      "learning_rate": 1.1367600176849032e-05,
+      "loss": 0.1971,
+      "step": 1795
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 2995,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 2000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.607849924074668e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

23_128_e5_3e-5/checkpoint-1797/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:462836112e88c18d92160ccca50fdb00995a79d39c0918ae35a73e6be56e20a2
+size 7736

23_128_e5_3e-5/checkpoint-1797/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)

23_128_e5_3e-5/checkpoint-2396/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

23_128_e5_3e-5/checkpoint-2396/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",
+    "gate_proj",
+    "v_proj",
+    "k_proj",
+    "up_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

23_128_e5_3e-5/checkpoint-2396/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc323d75c89ac24a0a850cf68327cd775f5ee10d0f214dce3a38179e07f7fb81
+size 791751704

23_128_e5_3e-5/checkpoint-2396/latest

@@ -0,0 +1 @@
+global_step2396

23_128_e5_3e-5/checkpoint-2396/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8d92bc2000eab468b997be392890e84070ac2b026b0ca7be4bf444f6da09f658
+size 15920

23_128_e5_3e-5/checkpoint-2396/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:24065c6e5e61208d9fe3df7dd22e57c6ffd96d983a72954dee5ac5bdca536cd1
+size 15920

23_128_e5_3e-5/checkpoint-2396/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e8e9dbe54257ddbaaa0ec94d1b8b38265f9da06c36b3f2e7f60c20cc626ddf0e
+size 15920