Upload folder using huggingface_hub

32_128_e5_3e-5/checkpoint-1158/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

32_128_e5_3e-5/checkpoint-1158/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": [
+    "up_proj",
+    "gate_proj",
+    "down_proj",
+    "q_proj",
+    "v_proj",
+    "k_proj",
+    "o_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

32_128_e5_3e-5/checkpoint-1158/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e40789539e733b3659b130bf42020911faaf851dd5bf3d3bfc9f8b451edaf3ca
+size 791751704

32_128_e5_3e-5/checkpoint-1158/latest

@@ -0,0 +1 @@
+global_step1158

32_128_e5_3e-5/checkpoint-1158/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8716877232ac411cd8724c54746c501db65188a382b2b5ba05b27e02afa38ad9
+size 15920

32_128_e5_3e-5/checkpoint-1158/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4d29c1342aed5dd046383c525e35387a62c41b85d7dd2bc5db70befefe09b301
+size 15920

32_128_e5_3e-5/checkpoint-1158/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:189d311fde6ff48b0a74c2b7fa7660bd6fbe8100a6ced8cc4e00e3bb7b5b8a2e
+size 15920

32_128_e5_3e-5/checkpoint-1158/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3a6a5c62eeb8fd064b9b75cd7149ae9c3f88004109a90d85a52eb8b7c65f03f1
+size 15920

32_128_e5_3e-5/checkpoint-1158/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a2ba034d44cc1687c80b13a20abdbee4cb43e40c9e4a265948c581554eead869
+size 15920

32_128_e5_3e-5/checkpoint-1158/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3bfd5c5ed3cdf42ce0e5a27f466624385a2a27e46a5e9b0436a643bc049afa33
+size 15920

32_128_e5_3e-5/checkpoint-1158/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2f005332f6182648b308bff9542f57b7ef0d1c36815741554d43ee1c29af3863
+size 15920

32_128_e5_3e-5/checkpoint-1158/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:51bc17cbb2e984639710e723dfd4873502a5f427a997781f2135919bfdf1bfc6
+size 15920

32_128_e5_3e-5/checkpoint-1158/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:133287543726f0752b413ff5a9801311009c764cc344be8d4d94febef2d78f59
+size 1064

32_128_e5_3e-5/checkpoint-1158/special_tokens_map.json

@@ -0,0 +1,51 @@
+{
+  "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": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

32_128_e5_3e-5/checkpoint-1158/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": "<|endoftext|>",
+  "padding_side": "left",
+  "tokenizer_class": "GPT2Tokenizer",
+  "unk_token": "<|endoftext|>",
+  "vocab_size": 49152
+}

32_128_e5_3e-5/checkpoint-1158/trainer_state.json

@@ -0,0 +1,1651 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 1158,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01297016861219196,
+      "grad_norm": 1.0690768957138062,
+      "learning_rate": 1.2371134020618557e-06,
+      "loss": 1.3595,
+      "step": 5
+    },
+    {
+      "epoch": 0.02594033722438392,
+      "grad_norm": 0.8587875962257385,
+      "learning_rate": 2.7835051546391753e-06,
+      "loss": 1.3355,
+      "step": 10
+    },
+    {
+      "epoch": 0.038910505836575876,
+      "grad_norm": 0.7017920017242432,
+      "learning_rate": 4.3298969072164946e-06,
+      "loss": 1.3063,
+      "step": 15
+    },
+    {
+      "epoch": 0.05188067444876784,
+      "grad_norm": 0.48842307925224304,
+      "learning_rate": 5.876288659793814e-06,
+      "loss": 1.3547,
+      "step": 20
+    },
+    {
+      "epoch": 0.0648508430609598,
+      "grad_norm": 0.47942179441452026,
+      "learning_rate": 7.422680412371135e-06,
+      "loss": 1.2945,
+      "step": 25
+    },
+    {
+      "epoch": 0.07782101167315175,
+      "grad_norm": 0.5098851323127747,
+      "learning_rate": 8.969072164948454e-06,
+      "loss": 1.3135,
+      "step": 30
+    },
+    {
+      "epoch": 0.0907911802853437,
+      "grad_norm": 0.5397660136222839,
+      "learning_rate": 1.0515463917525773e-05,
+      "loss": 1.2404,
+      "step": 35
+    },
+    {
+      "epoch": 0.10376134889753567,
+      "grad_norm": 0.5113343596458435,
+      "learning_rate": 1.2061855670103093e-05,
+      "loss": 1.2192,
+      "step": 40
+    },
+    {
+      "epoch": 0.11673151750972763,
+      "grad_norm": 0.44067639112472534,
+      "learning_rate": 1.3608247422680413e-05,
+      "loss": 1.1849,
+      "step": 45
+    },
+    {
+      "epoch": 0.1297016861219196,
+      "grad_norm": 0.5046361088752747,
+      "learning_rate": 1.5154639175257731e-05,
+      "loss": 1.2783,
+      "step": 50
+    },
+    {
+      "epoch": 0.14267185473411154,
+      "grad_norm": 0.4127322733402252,
+      "learning_rate": 1.6701030927835052e-05,
+      "loss": 1.1978,
+      "step": 55
+    },
+    {
+      "epoch": 0.1556420233463035,
+      "grad_norm": 0.47095608711242676,
+      "learning_rate": 1.824742268041237e-05,
+      "loss": 1.2134,
+      "step": 60
+    },
+    {
+      "epoch": 0.16861219195849547,
+      "grad_norm": 0.4175727963447571,
+      "learning_rate": 1.9793814432989692e-05,
+      "loss": 1.2058,
+      "step": 65
+    },
+    {
+      "epoch": 0.1815823605706874,
+      "grad_norm": 0.452852725982666,
+      "learning_rate": 2.134020618556701e-05,
+      "loss": 1.1875,
+      "step": 70
+    },
+    {
+      "epoch": 0.19455252918287938,
+      "grad_norm": 0.46958330273628235,
+      "learning_rate": 2.2886597938144328e-05,
+      "loss": 1.168,
+      "step": 75
+    },
+    {
+      "epoch": 0.20752269779507135,
+      "grad_norm": 0.522190272808075,
+      "learning_rate": 2.443298969072165e-05,
+      "loss": 1.1747,
+      "step": 80
+    },
+    {
+      "epoch": 0.2204928664072633,
+      "grad_norm": 0.44514960050582886,
+      "learning_rate": 2.597938144329897e-05,
+      "loss": 1.1504,
+      "step": 85
+    },
+    {
+      "epoch": 0.23346303501945526,
+      "grad_norm": 0.45393964648246765,
+      "learning_rate": 2.7525773195876287e-05,
+      "loss": 1.1557,
+      "step": 90
+    },
+    {
+      "epoch": 0.24643320363164722,
+      "grad_norm": 0.4870307147502899,
+      "learning_rate": 2.907216494845361e-05,
+      "loss": 1.1292,
+      "step": 95
+    },
+    {
+      "epoch": 0.2594033722438392,
+      "grad_norm": 0.5074522495269775,
+      "learning_rate": 2.999991187570717e-05,
+      "loss": 1.1929,
+      "step": 100
+    },
+    {
+      "epoch": 0.2723735408560311,
+      "grad_norm": 0.48062318563461304,
+      "learning_rate": 2.9998920489304306e-05,
+      "loss": 1.1339,
+      "step": 105
+    },
+    {
+      "epoch": 0.2853437094682231,
+      "grad_norm": 0.452762633562088,
+      "learning_rate": 2.99968276341792e-05,
+      "loss": 1.0783,
+      "step": 110
+    },
+    {
+      "epoch": 0.29831387808041504,
+      "grad_norm": 0.46613550186157227,
+      "learning_rate": 2.9993633464023883e-05,
+      "loss": 1.1303,
+      "step": 115
+    },
+    {
+      "epoch": 0.311284046692607,
+      "grad_norm": 0.5356303453445435,
+      "learning_rate": 2.9989338213407134e-05,
+      "loss": 1.0882,
+      "step": 120
+    },
+    {
+      "epoch": 0.324254215304799,
+      "grad_norm": 0.5889877676963806,
+      "learning_rate": 2.9983942197757268e-05,
+      "loss": 1.0996,
+      "step": 125
+    },
+    {
+      "epoch": 0.33722438391699094,
+      "grad_norm": 0.6177087426185608,
+      "learning_rate": 2.9977445813338964e-05,
+      "loss": 1.076,
+      "step": 130
+    },
+    {
+      "epoch": 0.35019455252918286,
+      "grad_norm": 0.6538938879966736,
+      "learning_rate": 2.996984953722418e-05,
+      "loss": 1.0926,
+      "step": 135
+    },
+    {
+      "epoch": 0.3631647211413748,
+      "grad_norm": 0.5603901147842407,
+      "learning_rate": 2.9961153927257093e-05,
+      "loss": 1.0801,
+      "step": 140
+    },
+    {
+      "epoch": 0.3761348897535668,
+      "grad_norm": 0.579160213470459,
+      "learning_rate": 2.995135962201315e-05,
+      "loss": 1.021,
+      "step": 145
+    },
+    {
+      "epoch": 0.38910505836575876,
+      "grad_norm": 0.6291273236274719,
+      "learning_rate": 2.9940467340752182e-05,
+      "loss": 1.029,
+      "step": 150
+    },
+    {
+      "epoch": 0.40207522697795073,
+      "grad_norm": 0.5813262462615967,
+      "learning_rate": 2.9928477883365555e-05,
+      "loss": 1.0386,
+      "step": 155
+    },
+    {
+      "epoch": 0.4150453955901427,
+      "grad_norm": 0.6637552380561829,
+      "learning_rate": 2.9915392130317462e-05,
+      "loss": 1.0496,
+      "step": 160
+    },
+    {
+      "epoch": 0.4280155642023346,
+      "grad_norm": 0.8415825963020325,
+      "learning_rate": 2.990121104258023e-05,
+      "loss": 1.0025,
+      "step": 165
+    },
+    {
+      "epoch": 0.4409857328145266,
+      "grad_norm": 0.6291825771331787,
+      "learning_rate": 2.988593566156379e-05,
+      "loss": 0.9619,
+      "step": 170
+    },
+    {
+      "epoch": 0.45395590142671854,
+      "grad_norm": 0.6472850441932678,
+      "learning_rate": 2.9869567109039164e-05,
+      "loss": 0.9765,
+      "step": 175
+    },
+    {
+      "epoch": 0.4669260700389105,
+      "grad_norm": 0.7866678237915039,
+      "learning_rate": 2.9852106587056098e-05,
+      "loss": 0.9929,
+      "step": 180
+    },
+    {
+      "epoch": 0.4798962386511025,
+      "grad_norm": 0.6654261350631714,
+      "learning_rate": 2.98335553778548e-05,
+      "loss": 0.9523,
+      "step": 185
+    },
+    {
+      "epoch": 0.49286640726329445,
+      "grad_norm": 0.718325674533844,
+      "learning_rate": 2.981391484377176e-05,
+      "loss": 0.9353,
+      "step": 190
+    },
+    {
+      "epoch": 0.5058365758754864,
+      "grad_norm": 0.7905012369155884,
+      "learning_rate": 2.9793186427139704e-05,
+      "loss": 0.9493,
+      "step": 195
+    },
+    {
+      "epoch": 0.5188067444876784,
+      "grad_norm": 0.7448559403419495,
+      "learning_rate": 2.977137165018169e-05,
+      "loss": 0.9394,
+      "step": 200
+    },
+    {
+      "epoch": 0.5317769130998703,
+      "grad_norm": 0.7643033266067505,
+      "learning_rate": 2.9748472114899316e-05,
+      "loss": 0.9119,
+      "step": 205
+    },
+    {
+      "epoch": 0.5447470817120622,
+      "grad_norm": 0.7649075984954834,
+      "learning_rate": 2.9724489502955062e-05,
+      "loss": 0.9382,
+      "step": 210
+    },
+    {
+      "epoch": 0.5577172503242542,
+      "grad_norm": 0.7593182325363159,
+      "learning_rate": 2.9699425575548802e-05,
+      "loss": 0.9519,
+      "step": 215
+    },
+    {
+      "epoch": 0.5706874189364461,
+      "grad_norm": 0.7075984477996826,
+      "learning_rate": 2.9673282173288477e-05,
+      "loss": 0.8989,
+      "step": 220
+    },
+    {
+      "epoch": 0.5836575875486382,
+      "grad_norm": 0.7443664073944092,
+      "learning_rate": 2.9646061216054914e-05,
+      "loss": 0.8752,
+      "step": 225
+    },
+    {
+      "epoch": 0.5966277561608301,
+      "grad_norm": 0.8497354984283447,
+      "learning_rate": 2.961776470286084e-05,
+      "loss": 0.8977,
+      "step": 230
+    },
+    {
+      "epoch": 0.6095979247730221,
+      "grad_norm": 0.8709704875946045,
+      "learning_rate": 2.958839471170409e-05,
+      "loss": 0.8699,
+      "step": 235
+    },
+    {
+      "epoch": 0.622568093385214,
+      "grad_norm": 0.79726642370224,
+      "learning_rate": 2.9557953399414987e-05,
+      "loss": 0.8906,
+      "step": 240
+    },
+    {
+      "epoch": 0.6355382619974059,
+      "grad_norm": 0.8190026879310608,
+      "learning_rate": 2.9526443001497974e-05,
+      "loss": 0.8847,
+      "step": 245
+    },
+    {
+      "epoch": 0.648508430609598,
+      "grad_norm": 0.9261983036994934,
+      "learning_rate": 2.9493865831967436e-05,
+      "loss": 0.9022,
+      "step": 250
+    },
+    {
+      "epoch": 0.6614785992217899,
+      "grad_norm": 0.9094107747077942,
+      "learning_rate": 2.9460224283177763e-05,
+      "loss": 0.8016,
+      "step": 255
+    },
+    {
+      "epoch": 0.6744487678339819,
+      "grad_norm": 0.854141891002655,
+      "learning_rate": 2.942552082564767e-05,
+      "loss": 0.8197,
+      "step": 260
+    },
+    {
+      "epoch": 0.6874189364461738,
+      "grad_norm": 0.8813712000846863,
+      "learning_rate": 2.938975800787876e-05,
+      "loss": 0.8538,
+      "step": 265
+    },
+    {
+      "epoch": 0.7003891050583657,
+      "grad_norm": 0.8427942395210266,
+      "learning_rate": 2.9352938456168393e-05,
+      "loss": 0.7911,
+      "step": 270
+    },
+    {
+      "epoch": 0.7133592736705577,
+      "grad_norm": 0.9274711012840271,
+      "learning_rate": 2.931506487441681e-05,
+      "loss": 0.8352,
+      "step": 275
+    },
+    {
+      "epoch": 0.7263294422827496,
+      "grad_norm": 0.8024734258651733,
+      "learning_rate": 2.9276140043928533e-05,
+      "loss": 0.8198,
+      "step": 280
+    },
+    {
+      "epoch": 0.7392996108949417,
+      "grad_norm": 0.8227505087852478,
+      "learning_rate": 2.9236166823208184e-05,
+      "loss": 0.8043,
+      "step": 285
+    },
+    {
+      "epoch": 0.7522697795071336,
+      "grad_norm": 0.8620085716247559,
+      "learning_rate": 2.9195148147750504e-05,
+      "loss": 0.8011,
+      "step": 290
+    },
+    {
+      "epoch": 0.7652399481193255,
+      "grad_norm": 0.8576732873916626,
+      "learning_rate": 2.915308702982482e-05,
+      "loss": 0.7978,
+      "step": 295
+    },
+    {
+      "epoch": 0.7782101167315175,
+      "grad_norm": 1.038345217704773,
+      "learning_rate": 2.9109986558253814e-05,
+      "loss": 0.7844,
+      "step": 300
+    },
+    {
+      "epoch": 0.7911802853437094,
+      "grad_norm": 0.8221902251243591,
+      "learning_rate": 2.9065849898186693e-05,
+      "loss": 0.8405,
+      "step": 305
+    },
+    {
+      "epoch": 0.8041504539559015,
+      "grad_norm": 0.9288234114646912,
+      "learning_rate": 2.902068029086676e-05,
+      "loss": 0.7909,
+      "step": 310
+    },
+    {
+      "epoch": 0.8171206225680934,
+      "grad_norm": 0.9620817303657532,
+      "learning_rate": 2.8974481053393386e-05,
+      "loss": 0.7999,
+      "step": 315
+    },
+    {
+      "epoch": 0.8300907911802854,
+      "grad_norm": 1.018082857131958,
+      "learning_rate": 2.89272555784784e-05,
+      "loss": 0.7731,
+      "step": 320
+    },
+    {
+      "epoch": 0.8430609597924773,
+      "grad_norm": 0.9036654233932495,
+      "learning_rate": 2.8879007334196963e-05,
+      "loss": 0.775,
+      "step": 325
+    },
+    {
+      "epoch": 0.8560311284046692,
+      "grad_norm": 0.826240062713623,
+      "learning_rate": 2.882973986373287e-05,
+      "loss": 0.7607,
+      "step": 330
+    },
+    {
+      "epoch": 0.8690012970168612,
+      "grad_norm": 0.8781949281692505,
+      "learning_rate": 2.877945678511834e-05,
+      "loss": 0.7673,
+      "step": 335
+    },
+    {
+      "epoch": 0.8819714656290532,
+      "grad_norm": 1.0212292671203613,
+      "learning_rate": 2.8728161790968352e-05,
+      "loss": 0.7506,
+      "step": 340
+    },
+    {
+      "epoch": 0.8949416342412452,
+      "grad_norm": 0.9420115351676941,
+      "learning_rate": 2.8675858648209442e-05,
+      "loss": 0.7218,
+      "step": 345
+    },
+    {
+      "epoch": 0.9079118028534371,
+      "grad_norm": 1.0438215732574463,
+      "learning_rate": 2.8622551197803093e-05,
+      "loss": 0.7349,
+      "step": 350
+    },
+    {
+      "epoch": 0.920881971465629,
+      "grad_norm": 1.0321903228759766,
+      "learning_rate": 2.8568243354463657e-05,
+      "loss": 0.7259,
+      "step": 355
+    },
+    {
+      "epoch": 0.933852140077821,
+      "grad_norm": 1.0026477575302124,
+      "learning_rate": 2.8512939106370864e-05,
+      "loss": 0.7264,
+      "step": 360
+    },
+    {
+      "epoch": 0.9468223086900129,
+      "grad_norm": 1.0209153890609741,
+      "learning_rate": 2.8456642514876976e-05,
+      "loss": 0.706,
+      "step": 365
+    },
+    {
+      "epoch": 0.959792477302205,
+      "grad_norm": 0.9709693789482117,
+      "learning_rate": 2.8399357714208498e-05,
+      "loss": 0.735,
+      "step": 370
+    },
+    {
+      "epoch": 0.9727626459143969,
+      "grad_norm": 1.0189666748046875,
+      "learning_rate": 2.8341088911162594e-05,
+      "loss": 0.7149,
+      "step": 375
+    },
+    {
+      "epoch": 0.9857328145265889,
+      "grad_norm": 1.0608593225479126,
+      "learning_rate": 2.8281840384798147e-05,
+      "loss": 0.7506,
+      "step": 380
+    },
+    {
+      "epoch": 0.9987029831387808,
+      "grad_norm": 1.1326305866241455,
+      "learning_rate": 2.822161648612154e-05,
+      "loss": 0.7307,
+      "step": 385
+    },
+    {
+      "epoch": 1.0103761348897535,
+      "grad_norm": 1.1239498853683472,
+      "learning_rate": 2.8160421637767105e-05,
+      "loss": 0.624,
+      "step": 390
+    },
+    {
+      "epoch": 1.0233463035019454,
+      "grad_norm": 0.9470804929733276,
+      "learning_rate": 2.809826033367235e-05,
+      "loss": 0.5788,
+      "step": 395
+    },
+    {
+      "epoch": 1.0363164721141376,
+      "grad_norm": 1.092249870300293,
+      "learning_rate": 2.803513713874794e-05,
+      "loss": 0.6766,
+      "step": 400
+    },
+    {
+      "epoch": 1.0492866407263295,
+      "grad_norm": 1.013177752494812,
+      "learning_rate": 2.797105668854249e-05,
+      "loss": 0.6742,
+      "step": 405
+    },
+    {
+      "epoch": 1.0622568093385214,
+      "grad_norm": 0.9360693097114563,
+      "learning_rate": 2.790602368890209e-05,
+      "loss": 0.6108,
+      "step": 410
+    },
+    {
+      "epoch": 1.0752269779507133,
+      "grad_norm": 0.9409477710723877,
+      "learning_rate": 2.7840042915624794e-05,
+      "loss": 0.6305,
+      "step": 415
+    },
+    {
+      "epoch": 1.0881971465629052,
+      "grad_norm": 1.3407989740371704,
+      "learning_rate": 2.777311921410983e-05,
+      "loss": 0.5923,
+      "step": 420
+    },
+    {
+      "epoch": 1.1011673151750974,
+      "grad_norm": 1.1192437410354614,
+      "learning_rate": 2.7705257499001837e-05,
+      "loss": 0.5941,
+      "step": 425
+    },
+    {
+      "epoch": 1.1141374837872893,
+      "grad_norm": 1.3298574686050415,
+      "learning_rate": 2.763646275382991e-05,
+      "loss": 0.5569,
+      "step": 430
+    },
+    {
+      "epoch": 1.1271076523994812,
+      "grad_norm": 0.9660617113113403,
+      "learning_rate": 2.7566740030641628e-05,
+      "loss": 0.5473,
+      "step": 435
+    },
+    {
+      "epoch": 1.140077821011673,
+      "grad_norm": 1.0812163352966309,
+      "learning_rate": 2.7496094449632064e-05,
+      "loss": 0.5982,
+      "step": 440
+    },
+    {
+      "epoch": 1.153047989623865,
+      "grad_norm": 1.1607142686843872,
+      "learning_rate": 2.7424531198767783e-05,
+      "loss": 0.6103,
+      "step": 445
+    },
+    {
+      "epoch": 1.1660181582360571,
+      "grad_norm": 1.1823639869689941,
+      "learning_rate": 2.7352055533405828e-05,
+      "loss": 0.5997,
+      "step": 450
+    },
+    {
+      "epoch": 1.178988326848249,
+      "grad_norm": 1.0552186965942383,
+      "learning_rate": 2.72786727759078e-05,
+      "loss": 0.5379,
+      "step": 455
+    },
+    {
+      "epoch": 1.191958495460441,
+      "grad_norm": 1.2875730991363525,
+      "learning_rate": 2.7204388315249007e-05,
+      "loss": 0.5581,
+      "step": 460
+    },
+    {
+      "epoch": 1.2049286640726329,
+      "grad_norm": 1.1362204551696777,
+      "learning_rate": 2.712920760662271e-05,
+      "loss": 0.5421,
+      "step": 465
+    },
+    {
+      "epoch": 1.217898832684825,
+      "grad_norm": 0.9040284156799316,
+      "learning_rate": 2.705313617103951e-05,
+      "loss": 0.5178,
+      "step": 470
+    },
+    {
+      "epoch": 1.230869001297017,
+      "grad_norm": 0.9417663216590881,
+      "learning_rate": 2.6976179594921915e-05,
+      "loss": 0.5913,
+      "step": 475
+    },
+    {
+      "epoch": 1.2438391699092088,
+      "grad_norm": 1.0625457763671875,
+      "learning_rate": 2.6898343529694077e-05,
+      "loss": 0.5811,
+      "step": 480
+    },
+    {
+      "epoch": 1.2568093385214008,
+      "grad_norm": 1.1155017614364624,
+      "learning_rate": 2.6819633691366784e-05,
+      "loss": 0.5369,
+      "step": 485
+    },
+    {
+      "epoch": 1.2697795071335927,
+      "grad_norm": 1.141462802886963,
+      "learning_rate": 2.674005586011769e-05,
+      "loss": 0.5635,
+      "step": 490
+    },
+    {
+      "epoch": 1.2827496757457846,
+      "grad_norm": 1.132550597190857,
+      "learning_rate": 2.6659615879866844e-05,
+      "loss": 0.5135,
+      "step": 495
+    },
+    {
+      "epoch": 1.2957198443579767,
+      "grad_norm": 1.1744695901870728,
+      "learning_rate": 2.6578319657847523e-05,
+      "loss": 0.5586,
+      "step": 500
+    },
+    {
+      "epoch": 1.3086900129701686,
+      "grad_norm": 1.0014809370040894,
+      "learning_rate": 2.649617316417243e-05,
+      "loss": 0.5548,
+      "step": 505
+    },
+    {
+      "epoch": 1.3216601815823605,
+      "grad_norm": 1.0711734294891357,
+      "learning_rate": 2.6413182431395273e-05,
+      "loss": 0.4958,
+      "step": 510
+    },
+    {
+      "epoch": 1.3346303501945525,
+      "grad_norm": 1.3235042095184326,
+      "learning_rate": 2.6329353554067758e-05,
+      "loss": 0.4948,
+      "step": 515
+    },
+    {
+      "epoch": 1.3476005188067446,
+      "grad_norm": 1.1252357959747314,
+      "learning_rate": 2.6244692688292015e-05,
+      "loss": 0.4991,
+      "step": 520
+    },
+    {
+      "epoch": 1.3605706874189365,
+      "grad_norm": 1.1571067571640015,
+      "learning_rate": 2.615920605126852e-05,
+      "loss": 0.5401,
+      "step": 525
+    },
+    {
+      "epoch": 1.3735408560311284,
+      "grad_norm": 1.0673496723175049,
+      "learning_rate": 2.607289992083954e-05,
+      "loss": 0.5342,
+      "step": 530
+    },
+    {
+      "epoch": 1.3865110246433203,
+      "grad_norm": 1.0683461427688599,
+      "learning_rate": 2.598578063502807e-05,
+      "loss": 0.4935,
+      "step": 535
+    },
+    {
+      "epoch": 1.3994811932555122,
+      "grad_norm": 1.0260779857635498,
+      "learning_rate": 2.5897854591572448e-05,
+      "loss": 0.4831,
+      "step": 540
+    },
+    {
+      "epoch": 1.4124513618677041,
+      "grad_norm": 0.9995431900024414,
+      "learning_rate": 2.5809128247456478e-05,
+      "loss": 0.5238,
+      "step": 545
+    },
+    {
+      "epoch": 1.4254215304798963,
+      "grad_norm": 1.1451547145843506,
+      "learning_rate": 2.5719608118435266e-05,
+      "loss": 0.5088,
+      "step": 550
+    },
+    {
+      "epoch": 1.4383916990920882,
+      "grad_norm": 1.133154273033142,
+      "learning_rate": 2.5629300778556746e-05,
+      "loss": 0.4793,
+      "step": 555
+    },
+    {
+      "epoch": 1.45136186770428,
+      "grad_norm": 1.1944438219070435,
+      "learning_rate": 2.5538212859678876e-05,
+      "loss": 0.5004,
+      "step": 560
+    },
+    {
+      "epoch": 1.4643320363164722,
+      "grad_norm": 1.150473713874817,
+      "learning_rate": 2.544635105098264e-05,
+      "loss": 0.5104,
+      "step": 565
+    },
+    {
+      "epoch": 1.4773022049286642,
+      "grad_norm": 1.1394140720367432,
+      "learning_rate": 2.5353722098480788e-05,
+      "loss": 0.5064,
+      "step": 570
+    },
+    {
+      "epoch": 1.490272373540856,
+      "grad_norm": 1.1548528671264648,
+      "learning_rate": 2.5260332804522478e-05,
+      "loss": 0.5266,
+      "step": 575
+    },
+    {
+      "epoch": 1.503242542153048,
+      "grad_norm": 1.0762923955917358,
+      "learning_rate": 2.5166190027293684e-05,
+      "loss": 0.446,
+      "step": 580
+    },
+    {
+      "epoch": 1.51621271076524,
+      "grad_norm": 1.1425952911376953,
+      "learning_rate": 2.50713006803136e-05,
+      "loss": 0.4396,
+      "step": 585
+    },
+    {
+      "epoch": 1.5291828793774318,
+      "grad_norm": 1.0265759229660034,
+      "learning_rate": 2.4975671731926892e-05,
+      "loss": 0.4541,
+      "step": 590
+    },
+    {
+      "epoch": 1.5421530479896237,
+      "grad_norm": 1.0789086818695068,
+      "learning_rate": 2.4879310204792012e-05,
+      "loss": 0.4674,
+      "step": 595
+    },
+    {
+      "epoch": 1.5551232166018158,
+      "grad_norm": 1.0244364738464355,
+      "learning_rate": 2.478222317536544e-05,
+      "loss": 0.4654,
+      "step": 600
+    },
+    {
+      "epoch": 1.5680933852140078,
+      "grad_norm": 1.1058456897735596,
+      "learning_rate": 2.4684417773382037e-05,
+      "loss": 0.4599,
+      "step": 605
+    },
+    {
+      "epoch": 1.5810635538262,
+      "grad_norm": 1.031064510345459,
+      "learning_rate": 2.458590118133145e-05,
+      "loss": 0.417,
+      "step": 610
+    },
+    {
+      "epoch": 1.5940337224383918,
+      "grad_norm": 1.0740469694137573,
+      "learning_rate": 2.4486680633930658e-05,
+      "loss": 0.4712,
+      "step": 615
+    },
+    {
+      "epoch": 1.6070038910505837,
+      "grad_norm": 1.1247283220291138,
+      "learning_rate": 2.4386763417592688e-05,
+      "loss": 0.5005,
+      "step": 620
+    },
+    {
+      "epoch": 1.6199740596627756,
+      "grad_norm": 1.0776716470718384,
+      "learning_rate": 2.428615686989151e-05,
+      "loss": 0.4437,
+      "step": 625
+    },
+    {
+      "epoch": 1.6329442282749675,
+      "grad_norm": 1.0497078895568848,
+      "learning_rate": 2.4184868379023204e-05,
+      "loss": 0.436,
+      "step": 630
+    },
+    {
+      "epoch": 1.6459143968871595,
+      "grad_norm": 1.2037216424942017,
+      "learning_rate": 2.4082905383263387e-05,
+      "loss": 0.4378,
+      "step": 635
+    },
+    {
+      "epoch": 1.6588845654993514,
+      "grad_norm": 1.0372422933578491,
+      "learning_rate": 2.3980275370420986e-05,
+      "loss": 0.4469,
+      "step": 640
+    },
+    {
+      "epoch": 1.6718547341115433,
+      "grad_norm": 1.1579840183258057,
+      "learning_rate": 2.387698587728834e-05,
+      "loss": 0.434,
+      "step": 645
+    },
+    {
+      "epoch": 1.6848249027237354,
+      "grad_norm": 1.1483240127563477,
+      "learning_rate": 2.377304448908775e-05,
+      "loss": 0.4355,
+      "step": 650
+    },
+    {
+      "epoch": 1.6977950713359273,
+      "grad_norm": 1.0725358724594116,
+      "learning_rate": 2.3668458838914398e-05,
+      "loss": 0.4579,
+      "step": 655
+    },
+    {
+      "epoch": 1.7107652399481195,
+      "grad_norm": 1.1938557624816895,
+      "learning_rate": 2.3563236607175877e-05,
+      "loss": 0.4397,
+      "step": 660
+    },
+    {
+      "epoch": 1.7237354085603114,
+      "grad_norm": 1.0592831373214722,
+      "learning_rate": 2.3457385521028085e-05,
+      "loss": 0.4342,
+      "step": 665
+    },
+    {
+      "epoch": 1.7367055771725033,
+      "grad_norm": 1.191778302192688,
+      "learning_rate": 2.3350913353807826e-05,
+      "loss": 0.4108,
+      "step": 670
+    },
+    {
+      "epoch": 1.7496757457846952,
+      "grad_norm": 1.1862207651138306,
+      "learning_rate": 2.324382792446194e-05,
+      "loss": 0.4462,
+      "step": 675
+    },
+    {
+      "epoch": 1.7626459143968871,
+      "grad_norm": 1.0156222581863403,
+      "learning_rate": 2.3136137096973124e-05,
+      "loss": 0.4245,
+      "step": 680
+    },
+    {
+      "epoch": 1.775616083009079,
+      "grad_norm": 1.2515448331832886,
+      "learning_rate": 2.3027848779782388e-05,
+      "loss": 0.3718,
+      "step": 685
+    },
+    {
+      "epoch": 1.788586251621271,
+      "grad_norm": 1.0088294744491577,
+      "learning_rate": 2.291897092520834e-05,
+      "loss": 0.41,
+      "step": 690
+    },
+    {
+      "epoch": 1.801556420233463,
+      "grad_norm": 1.4383920431137085,
+      "learning_rate": 2.280951152886316e-05,
+      "loss": 0.4075,
+      "step": 695
+    },
+    {
+      "epoch": 1.814526588845655,
+      "grad_norm": 1.1313135623931885,
+      "learning_rate": 2.269947862906544e-05,
+      "loss": 0.4139,
+      "step": 700
+    },
+    {
+      "epoch": 1.827496757457847,
+      "grad_norm": 1.1307212114334106,
+      "learning_rate": 2.2588880306249873e-05,
+      "loss": 0.3916,
+      "step": 705
+    },
+    {
+      "epoch": 1.840466926070039,
+      "grad_norm": 1.10785710811615,
+      "learning_rate": 2.2477724682373868e-05,
+      "loss": 0.4109,
+      "step": 710
+    },
+    {
+      "epoch": 1.853437094682231,
+      "grad_norm": 1.2819660902023315,
+      "learning_rate": 2.236601992032108e-05,
+      "loss": 0.3913,
+      "step": 715
+    },
+    {
+      "epoch": 1.8664072632944229,
+      "grad_norm": 1.2175217866897583,
+      "learning_rate": 2.2253774223301984e-05,
+      "loss": 0.3831,
+      "step": 720
+    },
+    {
+      "epoch": 1.8793774319066148,
+      "grad_norm": 1.0192928314208984,
+      "learning_rate": 2.2140995834251427e-05,
+      "loss": 0.3897,
+      "step": 725
+    },
+    {
+      "epoch": 1.8923476005188067,
+      "grad_norm": 1.0274885892868042,
+      "learning_rate": 2.2027693035223323e-05,
+      "loss": 0.4076,
+      "step": 730
+    },
+    {
+      "epoch": 1.9053177691309986,
+      "grad_norm": 1.0526772737503052,
+      "learning_rate": 2.1913874146782417e-05,
+      "loss": 0.3463,
+      "step": 735
+    },
+    {
+      "epoch": 1.9182879377431905,
+      "grad_norm": 1.1391632556915283,
+      "learning_rate": 2.1799547527393313e-05,
+      "loss": 0.3822,
+      "step": 740
+    },
+    {
+      "epoch": 1.9312581063553826,
+      "grad_norm": 1.1479544639587402,
+      "learning_rate": 2.1684721572806573e-05,
+      "loss": 0.361,
+      "step": 745
+    },
+    {
+      "epoch": 1.9442282749675746,
+      "grad_norm": 1.2531360387802124,
+      "learning_rate": 2.1569404715442223e-05,
+      "loss": 0.3892,
+      "step": 750
+    },
+    {
+      "epoch": 1.9571984435797667,
+      "grad_norm": 1.17800772190094,
+      "learning_rate": 2.145360542377049e-05,
+      "loss": 0.4036,
+      "step": 755
+    },
+    {
+      "epoch": 1.9701686121919586,
+      "grad_norm": 1.2836965322494507,
+      "learning_rate": 2.1337332201689896e-05,
+      "loss": 0.3661,
+      "step": 760
+    },
+    {
+      "epoch": 1.9831387808041505,
+      "grad_norm": 1.1780281066894531,
+      "learning_rate": 2.1220593587902772e-05,
+      "loss": 0.3602,
+      "step": 765
+    },
+    {
+      "epoch": 1.9961089494163424,
+      "grad_norm": 1.1417920589447021,
+      "learning_rate": 2.1103398155288207e-05,
+      "loss": 0.3628,
+      "step": 770
+    },
+    {
+      "epoch": 2.007782101167315,
+      "grad_norm": 1.0694905519485474,
+      "learning_rate": 2.0985754510272474e-05,
+      "loss": 0.3022,
+      "step": 775
+    },
+    {
+      "epoch": 2.020752269779507,
+      "grad_norm": 1.2128925323486328,
+      "learning_rate": 2.0867671292197013e-05,
+      "loss": 0.3188,
+      "step": 780
+    },
+    {
+      "epoch": 2.033722438391699,
+      "grad_norm": 1.137680172920227,
+      "learning_rate": 2.0749157172683994e-05,
+      "loss": 0.3077,
+      "step": 785
+    },
+    {
+      "epoch": 2.046692607003891,
+      "grad_norm": 1.3190040588378906,
+      "learning_rate": 2.063022085499948e-05,
+      "loss": 0.292,
+      "step": 790
+    },
+    {
+      "epoch": 2.059662775616083,
+      "grad_norm": 1.097852110862732,
+      "learning_rate": 2.0510871073414315e-05,
+      "loss": 0.2982,
+      "step": 795
+    },
+    {
+      "epoch": 2.072632944228275,
+      "grad_norm": 1.1585649251937866,
+      "learning_rate": 2.039111659256269e-05,
+      "loss": 0.2592,
+      "step": 800
+    },
+    {
+      "epoch": 2.085603112840467,
+      "grad_norm": 1.065697193145752,
+      "learning_rate": 2.0270966206798523e-05,
+      "loss": 0.2759,
+      "step": 805
+    },
+    {
+      "epoch": 2.098573281452659,
+      "grad_norm": 1.011765718460083,
+      "learning_rate": 2.0150428739549595e-05,
+      "loss": 0.2843,
+      "step": 810
+    },
+    {
+      "epoch": 2.111543450064851,
+      "grad_norm": 1.0232386589050293,
+      "learning_rate": 2.002951304266962e-05,
+      "loss": 0.3209,
+      "step": 815
+    },
+    {
+      "epoch": 2.124513618677043,
+      "grad_norm": 1.1222460269927979,
+      "learning_rate": 1.990822799578819e-05,
+      "loss": 0.2832,
+      "step": 820
+    },
+    {
+      "epoch": 2.1374837872892347,
+      "grad_norm": 1.1336065530776978,
+      "learning_rate": 1.9786582505658668e-05,
+      "loss": 0.2981,
+      "step": 825
+    },
+    {
+      "epoch": 2.1504539559014266,
+      "grad_norm": 1.2448867559432983,
+      "learning_rate": 1.9664585505504134e-05,
+      "loss": 0.2898,
+      "step": 830
+    },
+    {
+      "epoch": 2.1634241245136185,
+      "grad_norm": 1.1029194593429565,
+      "learning_rate": 1.9542245954361333e-05,
+      "loss": 0.2748,
+      "step": 835
+    },
+    {
+      "epoch": 2.1763942931258105,
+      "grad_norm": 1.2409709692001343,
+      "learning_rate": 1.941957283642277e-05,
+      "loss": 0.2791,
+      "step": 840
+    },
+    {
+      "epoch": 2.189364461738003,
+      "grad_norm": 1.0632833242416382,
+      "learning_rate": 1.9296575160376953e-05,
+      "loss": 0.2774,
+      "step": 845
+    },
+    {
+      "epoch": 2.2023346303501947,
+      "grad_norm": 1.166345238685608,
+      "learning_rate": 1.9173261958746793e-05,
+      "loss": 0.2695,
+      "step": 850
+    },
+    {
+      "epoch": 2.2153047989623866,
+      "grad_norm": 1.0866575241088867,
+      "learning_rate": 1.9049642287226307e-05,
+      "loss": 0.2739,
+      "step": 855
+    },
+    {
+      "epoch": 2.2282749675745785,
+      "grad_norm": 1.0431710481643677,
+      "learning_rate": 1.8925725224015616e-05,
+      "loss": 0.256,
+      "step": 860
+    },
+    {
+      "epoch": 2.2412451361867705,
+      "grad_norm": 1.2127219438552856,
+      "learning_rate": 1.8801519869154237e-05,
+      "loss": 0.2638,
+      "step": 865
+    },
+    {
+      "epoch": 2.2542153047989624,
+      "grad_norm": 1.1814619302749634,
+      "learning_rate": 1.867703534385284e-05,
+      "loss": 0.2781,
+      "step": 870
+    },
+    {
+      "epoch": 2.2671854734111543,
+      "grad_norm": 1.1074808835983276,
+      "learning_rate": 1.8552280789823394e-05,
+      "loss": 0.2788,
+      "step": 875
+    },
+    {
+      "epoch": 2.280155642023346,
+      "grad_norm": 1.226494312286377,
+      "learning_rate": 1.8427265368607844e-05,
+      "loss": 0.2707,
+      "step": 880
+    },
+    {
+      "epoch": 2.293125810635538,
+      "grad_norm": 1.1561022996902466,
+      "learning_rate": 1.830199826090532e-05,
+      "loss": 0.2879,
+      "step": 885
+    },
+    {
+      "epoch": 2.30609597924773,
+      "grad_norm": 1.1817694902420044,
+      "learning_rate": 1.8176488665897945e-05,
+      "loss": 0.277,
+      "step": 890
+    },
+    {
+      "epoch": 2.319066147859922,
+      "grad_norm": 1.3267537355422974,
+      "learning_rate": 1.805074580057524e-05,
+      "loss": 0.2832,
+      "step": 895
+    },
+    {
+      "epoch": 2.3320363164721143,
+      "grad_norm": 1.1436495780944824,
+      "learning_rate": 1.7924778899057316e-05,
+      "loss": 0.2822,
+      "step": 900
+    },
+    {
+      "epoch": 2.345006485084306,
+      "grad_norm": 1.1241085529327393,
+      "learning_rate": 1.779859721191671e-05,
+      "loss": 0.29,
+      "step": 905
+    },
+    {
+      "epoch": 2.357976653696498,
+      "grad_norm": 1.5335941314697266,
+      "learning_rate": 1.7672210005499088e-05,
+      "loss": 0.2511,
+      "step": 910
+    },
+    {
+      "epoch": 2.37094682230869,
+      "grad_norm": 1.233126163482666,
+      "learning_rate": 1.754562656124272e-05,
+      "loss": 0.2595,
+      "step": 915
+    },
+    {
+      "epoch": 2.383916990920882,
+      "grad_norm": 1.1877148151397705,
+      "learning_rate": 1.7418856174996915e-05,
+      "loss": 0.2536,
+      "step": 920
+    },
+    {
+      "epoch": 2.396887159533074,
+      "grad_norm": 1.3099548816680908,
+      "learning_rate": 1.729190815633935e-05,
+      "loss": 0.2719,
+      "step": 925
+    },
+    {
+      "epoch": 2.4098573281452658,
+      "grad_norm": 1.1045476198196411,
+      "learning_rate": 1.71647918278924e-05,
+      "loss": 0.2438,
+      "step": 930
+    },
+    {
+      "epoch": 2.4228274967574577,
+      "grad_norm": 1.1141847372055054,
+      "learning_rate": 1.7037516524638548e-05,
+      "loss": 0.2439,
+      "step": 935
+    },
+    {
+      "epoch": 2.43579766536965,
+      "grad_norm": 1.0792944431304932,
+      "learning_rate": 1.6910091593234823e-05,
+      "loss": 0.2265,
+      "step": 940
+    },
+    {
+      "epoch": 2.448767833981842,
+      "grad_norm": 1.1306582689285278,
+      "learning_rate": 1.6782526391326425e-05,
+      "loss": 0.2525,
+      "step": 945
+    },
+    {
+      "epoch": 2.461738002594034,
+      "grad_norm": 1.1637719869613647,
+      "learning_rate": 1.6654830286859533e-05,
+      "loss": 0.2327,
+      "step": 950
+    },
+    {
+      "epoch": 2.4747081712062258,
+      "grad_norm": 1.0156742334365845,
+      "learning_rate": 1.6527012657393372e-05,
+      "loss": 0.2229,
+      "step": 955
+    },
+    {
+      "epoch": 2.4876783398184177,
+      "grad_norm": 1.0547713041305542,
+      "learning_rate": 1.6399082889411525e-05,
+      "loss": 0.2477,
+      "step": 960
+    },
+    {
+      "epoch": 2.5006485084306096,
+      "grad_norm": 1.1955467462539673,
+      "learning_rate": 1.627105037763264e-05,
+      "loss": 0.2333,
+      "step": 965
+    },
+    {
+      "epoch": 2.5136186770428015,
+      "grad_norm": 1.1450884342193604,
+      "learning_rate": 1.614292452432052e-05,
+      "loss": 0.2504,
+      "step": 970
+    },
+    {
+      "epoch": 2.5265888456549934,
+      "grad_norm": 1.1756523847579956,
+      "learning_rate": 1.6014714738593637e-05,
+      "loss": 0.2415,
+      "step": 975
+    },
+    {
+      "epoch": 2.5395590142671853,
+      "grad_norm": 1.1718183755874634,
+      "learning_rate": 1.588643043573419e-05,
+      "loss": 0.2289,
+      "step": 980
+    },
+    {
+      "epoch": 2.5525291828793772,
+      "grad_norm": 1.236280918121338,
+      "learning_rate": 1.575808103649663e-05,
+      "loss": 0.2355,
+      "step": 985
+    },
+    {
+      "epoch": 2.565499351491569,
+      "grad_norm": 1.082622766494751,
+      "learning_rate": 1.5629675966415877e-05,
+      "loss": 0.2434,
+      "step": 990
+    },
+    {
+      "epoch": 2.5784695201037615,
+      "grad_norm": 1.3068712949752808,
+      "learning_rate": 1.5501224655115118e-05,
+      "loss": 0.2186,
+      "step": 995
+    },
+    {
+      "epoch": 2.5914396887159534,
+      "grad_norm": 1.0971678495407104,
+      "learning_rate": 1.5372736535613347e-05,
+      "loss": 0.2442,
+      "step": 1000
+    },
+    {
+      "epoch": 2.6044098573281453,
+      "grad_norm": 1.1903417110443115,
+      "learning_rate": 1.5244221043632605e-05,
+      "loss": 0.2455,
+      "step": 1005
+    },
+    {
+      "epoch": 2.6173800259403373,
+      "grad_norm": 1.1289457082748413,
+      "learning_rate": 1.511568761690509e-05,
+      "loss": 0.2531,
+      "step": 1010
+    },
+    {
+      "epoch": 2.630350194552529,
+      "grad_norm": 1.1403872966766357,
+      "learning_rate": 1.4987145694480054e-05,
+      "loss": 0.2355,
+      "step": 1015
+    },
+    {
+      "epoch": 2.643320363164721,
+      "grad_norm": 1.3029553890228271,
+      "learning_rate": 1.4858604716030655e-05,
+      "loss": 0.2496,
+      "step": 1020
+    },
+    {
+      "epoch": 2.656290531776913,
+      "grad_norm": 0.9976205825805664,
+      "learning_rate": 1.4730074121160718e-05,
+      "loss": 0.2097,
+      "step": 1025
+    },
+    {
+      "epoch": 2.669260700389105,
+      "grad_norm": 1.191940426826477,
+      "learning_rate": 1.4601563348711543e-05,
+      "loss": 0.2129,
+      "step": 1030
+    },
+    {
+      "epoch": 2.6822308690012973,
+      "grad_norm": 1.2161098718643188,
+      "learning_rate": 1.4473081836068728e-05,
+      "loss": 0.2274,
+      "step": 1035
+    },
+    {
+      "epoch": 2.695201037613489,
+      "grad_norm": 1.191472053527832,
+      "learning_rate": 1.4344639018469145e-05,
+      "loss": 0.2332,
+      "step": 1040
+    },
+    {
+      "epoch": 2.708171206225681,
+      "grad_norm": 1.1076041460037231,
+      "learning_rate": 1.4216244328308029e-05,
+      "loss": 0.2337,
+      "step": 1045
+    },
+    {
+      "epoch": 2.721141374837873,
+      "grad_norm": 1.1517136096954346,
+      "learning_rate": 1.4087907194446307e-05,
+      "loss": 0.227,
+      "step": 1050
+    },
+    {
+      "epoch": 2.734111543450065,
+      "grad_norm": 1.2073043584823608,
+      "learning_rate": 1.395963704151817e-05,
+      "loss": 0.2006,
+      "step": 1055
+    },
+    {
+      "epoch": 2.747081712062257,
+      "grad_norm": 1.2052357196807861,
+      "learning_rate": 1.3831443289238951e-05,
+      "loss": 0.2119,
+      "step": 1060
+    },
+    {
+      "epoch": 2.7600518806744487,
+      "grad_norm": 1.129930853843689,
+      "learning_rate": 1.3703335351713403e-05,
+      "loss": 0.1809,
+      "step": 1065
+    },
+    {
+      "epoch": 2.7730220492866406,
+      "grad_norm": 1.2224830389022827,
+      "learning_rate": 1.3575322636744325e-05,
+      "loss": 0.2393,
+      "step": 1070
+    },
+    {
+      "epoch": 2.7859922178988326,
+      "grad_norm": 1.2128647565841675,
+      "learning_rate": 1.3447414545141712e-05,
+      "loss": 0.2124,
+      "step": 1075
+    },
+    {
+      "epoch": 2.7989623865110245,
+      "grad_norm": 1.084707260131836,
+      "learning_rate": 1.3319620470032379e-05,
+      "loss": 0.2222,
+      "step": 1080
+    },
+    {
+      "epoch": 2.8119325551232164,
+      "grad_norm": 1.155920147895813,
+      "learning_rate": 1.3191949796170156e-05,
+      "loss": 0.2154,
+      "step": 1085
+    },
+    {
+      "epoch": 2.8249027237354083,
+      "grad_norm": 1.206000566482544,
+      "learning_rate": 1.3064411899246728e-05,
+      "loss": 0.1929,
+      "step": 1090
+    },
+    {
+      "epoch": 2.8378728923476007,
+      "grad_norm": 1.216287612915039,
+      "learning_rate": 1.293701614520309e-05,
+      "loss": 0.2301,
+      "step": 1095
+    },
+    {
+      "epoch": 2.8508430609597926,
+      "grad_norm": 1.1954846382141113,
+      "learning_rate": 1.280977188954176e-05,
+      "loss": 0.2108,
+      "step": 1100
+    },
+    {
+      "epoch": 2.8638132295719845,
+      "grad_norm": 1.132566213607788,
+      "learning_rate": 1.2682688476639753e-05,
+      "loss": 0.2138,
+      "step": 1105
+    },
+    {
+      "epoch": 2.8767833981841764,
+      "grad_norm": 1.0993660688400269,
+      "learning_rate": 1.2555775239062323e-05,
+      "loss": 0.2069,
+      "step": 1110
+    },
+    {
+      "epoch": 2.8897535667963683,
+      "grad_norm": 1.035963535308838,
+      "learning_rate": 1.242904149687766e-05,
+      "loss": 0.2059,
+      "step": 1115
+    },
+    {
+      "epoch": 2.90272373540856,
+      "grad_norm": 1.0626498460769653,
+      "learning_rate": 1.2302496556972424e-05,
+      "loss": 0.2147,
+      "step": 1120
+    },
+    {
+      "epoch": 2.915693904020752,
+      "grad_norm": 1.0666407346725464,
+      "learning_rate": 1.2176149712368298e-05,
+      "loss": 0.1955,
+      "step": 1125
+    },
+    {
+      "epoch": 2.9286640726329445,
+      "grad_norm": 1.1655992269515991,
+      "learning_rate": 1.2050010241539524e-05,
+      "loss": 0.2158,
+      "step": 1130
+    },
+    {
+      "epoch": 2.9416342412451364,
+      "grad_norm": 1.0219794511795044,
+      "learning_rate": 1.1924087407731552e-05,
+      "loss": 0.2001,
+      "step": 1135
+    },
+    {
+      "epoch": 2.9546044098573283,
+      "grad_norm": 1.1955561637878418,
+      "learning_rate": 1.1798390458280721e-05,
+      "loss": 0.2017,
+      "step": 1140
+    },
+    {
+      "epoch": 2.96757457846952,
+      "grad_norm": 1.2346644401550293,
+      "learning_rate": 1.1672928623935246e-05,
+      "loss": 0.1987,
+      "step": 1145
+    },
+    {
+      "epoch": 2.980544747081712,
+      "grad_norm": 1.2184339761734009,
+      "learning_rate": 1.1547711118177288e-05,
+      "loss": 0.185,
+      "step": 1150
+    },
+    {
+      "epoch": 2.993514915693904,
+      "grad_norm": 1.226433277130127,
+      "learning_rate": 1.1422747136546358e-05,
+      "loss": 0.1663,
+      "step": 1155
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1930,
+  "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": 1.7583085006031421e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

32_128_e5_3e-5/checkpoint-1158/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b54d15b6a19c1b0185a13af26c266453a69d60551948a83f1872822676a12e8
+size 7736

32_128_e5_3e-5/checkpoint-1158/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)

32_128_e5_3e-5/checkpoint-1544/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

32_128_e5_3e-5/checkpoint-1544/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": [
+    "up_proj",
+    "gate_proj",
+    "down_proj",
+    "q_proj",
+    "v_proj",
+    "k_proj",
+    "o_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

32_128_e5_3e-5/checkpoint-1544/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6c789df42918ede936b3d04025b970fbd3012fd460d5d73457b3a914b958dbc9
+size 791751704

32_128_e5_3e-5/checkpoint-1544/latest

@@ -0,0 +1 @@
+global_step1544

32_128_e5_3e-5/checkpoint-1544/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7fc1b25c845e502895c278c74ae9b3eee40cfe2f29e06de7c3cff80c1f98538d
+size 15920

32_128_e5_3e-5/checkpoint-1544/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f0e9acbf41997fb87b1d49e43830aac75fc2e9ad3cd987f7f6a78e0023fa533f
+size 15920

32_128_e5_3e-5/checkpoint-1544/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:82090b3c4488805c64c72360d0cba6c877cf972c6be63fb85740df0ca3c7f0ca
+size 15920

32_128_e5_3e-5/checkpoint-1544/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e9c77f4e07933c9b66b12e7fb21332b2a9e73336420ed1ee5b2d31fd3d79de6
+size 15920

32_128_e5_3e-5/checkpoint-1544/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a6d8b7889be86d3ec1e4f8153e132ce42fdd454386ed365e41598fd54f013f28
+size 15920

32_128_e5_3e-5/checkpoint-1544/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d37db9962b61eb3c82832720a2a6a8c3b18c42b863b18bd236271738a7261fff
+size 15920

32_128_e5_3e-5/checkpoint-1544/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:286839b7015a7e8bed527c91ea068b24db85507eed1b8209253b7c37325a90ce
+size 15920

32_128_e5_3e-5/checkpoint-1544/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1a3a77017f483a934ce25cd2a24fd493386d14a4c2cb3edc67538cdff42e54c7
+size 15920

32_128_e5_3e-5/checkpoint-1544/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1b13346042d8c64b847bd983ed0e07884ef613b8eeaab81ef809d74c9dc8b2fc
+size 1064

32_128_e5_3e-5/checkpoint-1544/special_tokens_map.json

@@ -0,0 +1,51 @@
+{
+  "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": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

32_128_e5_3e-5/checkpoint-1544/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": "<|endoftext|>",
+  "padding_side": "left",
+  "tokenizer_class": "GPT2Tokenizer",
+  "unk_token": "<|endoftext|>",
+  "vocab_size": 49152
+}

32_128_e5_3e-5/checkpoint-1544/trainer_state.json

@@ -0,0 +1,2190 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 1544,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01297016861219196,
+      "grad_norm": 1.0690768957138062,
+      "learning_rate": 1.2371134020618557e-06,
+      "loss": 1.3595,
+      "step": 5
+    },
+    {
+      "epoch": 0.02594033722438392,
+      "grad_norm": 0.8587875962257385,
+      "learning_rate": 2.7835051546391753e-06,
+      "loss": 1.3355,
+      "step": 10
+    },
+    {
+      "epoch": 0.038910505836575876,
+      "grad_norm": 0.7017920017242432,
+      "learning_rate": 4.3298969072164946e-06,
+      "loss": 1.3063,
+      "step": 15
+    },
+    {
+      "epoch": 0.05188067444876784,
+      "grad_norm": 0.48842307925224304,
+      "learning_rate": 5.876288659793814e-06,
+      "loss": 1.3547,
+      "step": 20
+    },
+    {
+      "epoch": 0.0648508430609598,
+      "grad_norm": 0.47942179441452026,
+      "learning_rate": 7.422680412371135e-06,
+      "loss": 1.2945,
+      "step": 25
+    },
+    {
+      "epoch": 0.07782101167315175,
+      "grad_norm": 0.5098851323127747,
+      "learning_rate": 8.969072164948454e-06,
+      "loss": 1.3135,
+      "step": 30
+    },
+    {
+      "epoch": 0.0907911802853437,
+      "grad_norm": 0.5397660136222839,
+      "learning_rate": 1.0515463917525773e-05,
+      "loss": 1.2404,
+      "step": 35
+    },
+    {
+      "epoch": 0.10376134889753567,
+      "grad_norm": 0.5113343596458435,
+      "learning_rate": 1.2061855670103093e-05,
+      "loss": 1.2192,
+      "step": 40
+    },
+    {
+      "epoch": 0.11673151750972763,
+      "grad_norm": 0.44067639112472534,
+      "learning_rate": 1.3608247422680413e-05,
+      "loss": 1.1849,
+      "step": 45
+    },
+    {
+      "epoch": 0.1297016861219196,
+      "grad_norm": 0.5046361088752747,
+      "learning_rate": 1.5154639175257731e-05,
+      "loss": 1.2783,
+      "step": 50
+    },
+    {
+      "epoch": 0.14267185473411154,
+      "grad_norm": 0.4127322733402252,
+      "learning_rate": 1.6701030927835052e-05,
+      "loss": 1.1978,
+      "step": 55
+    },
+    {
+      "epoch": 0.1556420233463035,
+      "grad_norm": 0.47095608711242676,
+      "learning_rate": 1.824742268041237e-05,
+      "loss": 1.2134,
+      "step": 60
+    },
+    {
+      "epoch": 0.16861219195849547,
+      "grad_norm": 0.4175727963447571,
+      "learning_rate": 1.9793814432989692e-05,
+      "loss": 1.2058,
+      "step": 65
+    },
+    {
+      "epoch": 0.1815823605706874,
+      "grad_norm": 0.452852725982666,
+      "learning_rate": 2.134020618556701e-05,
+      "loss": 1.1875,
+      "step": 70
+    },
+    {
+      "epoch": 0.19455252918287938,
+      "grad_norm": 0.46958330273628235,
+      "learning_rate": 2.2886597938144328e-05,
+      "loss": 1.168,
+      "step": 75
+    },
+    {
+      "epoch": 0.20752269779507135,
+      "grad_norm": 0.522190272808075,
+      "learning_rate": 2.443298969072165e-05,
+      "loss": 1.1747,
+      "step": 80
+    },
+    {
+      "epoch": 0.2204928664072633,
+      "grad_norm": 0.44514960050582886,
+      "learning_rate": 2.597938144329897e-05,
+      "loss": 1.1504,
+      "step": 85
+    },
+    {
+      "epoch": 0.23346303501945526,
+      "grad_norm": 0.45393964648246765,
+      "learning_rate": 2.7525773195876287e-05,
+      "loss": 1.1557,
+      "step": 90
+    },
+    {
+      "epoch": 0.24643320363164722,
+      "grad_norm": 0.4870307147502899,
+      "learning_rate": 2.907216494845361e-05,
+      "loss": 1.1292,
+      "step": 95
+    },
+    {
+      "epoch": 0.2594033722438392,
+      "grad_norm": 0.5074522495269775,
+      "learning_rate": 2.999991187570717e-05,
+      "loss": 1.1929,
+      "step": 100
+    },
+    {
+      "epoch": 0.2723735408560311,
+      "grad_norm": 0.48062318563461304,
+      "learning_rate": 2.9998920489304306e-05,
+      "loss": 1.1339,
+      "step": 105
+    },
+    {
+      "epoch": 0.2853437094682231,
+      "grad_norm": 0.452762633562088,
+      "learning_rate": 2.99968276341792e-05,
+      "loss": 1.0783,
+      "step": 110
+    },
+    {
+      "epoch": 0.29831387808041504,
+      "grad_norm": 0.46613550186157227,
+      "learning_rate": 2.9993633464023883e-05,
+      "loss": 1.1303,
+      "step": 115
+    },
+    {
+      "epoch": 0.311284046692607,
+      "grad_norm": 0.5356303453445435,
+      "learning_rate": 2.9989338213407134e-05,
+      "loss": 1.0882,
+      "step": 120
+    },
+    {
+      "epoch": 0.324254215304799,
+      "grad_norm": 0.5889877676963806,
+      "learning_rate": 2.9983942197757268e-05,
+      "loss": 1.0996,
+      "step": 125
+    },
+    {
+      "epoch": 0.33722438391699094,
+      "grad_norm": 0.6177087426185608,
+      "learning_rate": 2.9977445813338964e-05,
+      "loss": 1.076,
+      "step": 130
+    },
+    {
+      "epoch": 0.35019455252918286,
+      "grad_norm": 0.6538938879966736,
+      "learning_rate": 2.996984953722418e-05,
+      "loss": 1.0926,
+      "step": 135
+    },
+    {
+      "epoch": 0.3631647211413748,
+      "grad_norm": 0.5603901147842407,
+      "learning_rate": 2.9961153927257093e-05,
+      "loss": 1.0801,
+      "step": 140
+    },
+    {
+      "epoch": 0.3761348897535668,
+      "grad_norm": 0.579160213470459,
+      "learning_rate": 2.995135962201315e-05,
+      "loss": 1.021,
+      "step": 145
+    },
+    {
+      "epoch": 0.38910505836575876,
+      "grad_norm": 0.6291273236274719,
+      "learning_rate": 2.9940467340752182e-05,
+      "loss": 1.029,
+      "step": 150
+    },
+    {
+      "epoch": 0.40207522697795073,
+      "grad_norm": 0.5813262462615967,
+      "learning_rate": 2.9928477883365555e-05,
+      "loss": 1.0386,
+      "step": 155
+    },
+    {
+      "epoch": 0.4150453955901427,
+      "grad_norm": 0.6637552380561829,
+      "learning_rate": 2.9915392130317462e-05,
+      "loss": 1.0496,
+      "step": 160
+    },
+    {
+      "epoch": 0.4280155642023346,
+      "grad_norm": 0.8415825963020325,
+      "learning_rate": 2.990121104258023e-05,
+      "loss": 1.0025,
+      "step": 165
+    },
+    {
+      "epoch": 0.4409857328145266,
+      "grad_norm": 0.6291825771331787,
+      "learning_rate": 2.988593566156379e-05,
+      "loss": 0.9619,
+      "step": 170
+    },
+    {
+      "epoch": 0.45395590142671854,
+      "grad_norm": 0.6472850441932678,
+      "learning_rate": 2.9869567109039164e-05,
+      "loss": 0.9765,
+      "step": 175
+    },
+    {
+      "epoch": 0.4669260700389105,
+      "grad_norm": 0.7866678237915039,
+      "learning_rate": 2.9852106587056098e-05,
+      "loss": 0.9929,
+      "step": 180
+    },
+    {
+      "epoch": 0.4798962386511025,
+      "grad_norm": 0.6654261350631714,
+      "learning_rate": 2.98335553778548e-05,
+      "loss": 0.9523,
+      "step": 185
+    },
+    {
+      "epoch": 0.49286640726329445,
+      "grad_norm": 0.718325674533844,
+      "learning_rate": 2.981391484377176e-05,
+      "loss": 0.9353,
+      "step": 190
+    },
+    {
+      "epoch": 0.5058365758754864,
+      "grad_norm": 0.7905012369155884,
+      "learning_rate": 2.9793186427139704e-05,
+      "loss": 0.9493,
+      "step": 195
+    },
+    {
+      "epoch": 0.5188067444876784,
+      "grad_norm": 0.7448559403419495,
+      "learning_rate": 2.977137165018169e-05,
+      "loss": 0.9394,
+      "step": 200
+    },
+    {
+      "epoch": 0.5317769130998703,
+      "grad_norm": 0.7643033266067505,
+      "learning_rate": 2.9748472114899316e-05,
+      "loss": 0.9119,
+      "step": 205
+    },
+    {
+      "epoch": 0.5447470817120622,
+      "grad_norm": 0.7649075984954834,
+      "learning_rate": 2.9724489502955062e-05,
+      "loss": 0.9382,
+      "step": 210
+    },
+    {
+      "epoch": 0.5577172503242542,
+      "grad_norm": 0.7593182325363159,
+      "learning_rate": 2.9699425575548802e-05,
+      "loss": 0.9519,
+      "step": 215
+    },
+    {
+      "epoch": 0.5706874189364461,
+      "grad_norm": 0.7075984477996826,
+      "learning_rate": 2.9673282173288477e-05,
+      "loss": 0.8989,
+      "step": 220
+    },
+    {
+      "epoch": 0.5836575875486382,
+      "grad_norm": 0.7443664073944092,
+      "learning_rate": 2.9646061216054914e-05,
+      "loss": 0.8752,
+      "step": 225
+    },
+    {
+      "epoch": 0.5966277561608301,
+      "grad_norm": 0.8497354984283447,
+      "learning_rate": 2.961776470286084e-05,
+      "loss": 0.8977,
+      "step": 230
+    },
+    {
+      "epoch": 0.6095979247730221,
+      "grad_norm": 0.8709704875946045,
+      "learning_rate": 2.958839471170409e-05,
+      "loss": 0.8699,
+      "step": 235
+    },
+    {
+      "epoch": 0.622568093385214,
+      "grad_norm": 0.79726642370224,
+      "learning_rate": 2.9557953399414987e-05,
+      "loss": 0.8906,
+      "step": 240
+    },
+    {
+      "epoch": 0.6355382619974059,
+      "grad_norm": 0.8190026879310608,
+      "learning_rate": 2.9526443001497974e-05,
+      "loss": 0.8847,
+      "step": 245
+    },
+    {
+      "epoch": 0.648508430609598,
+      "grad_norm": 0.9261983036994934,
+      "learning_rate": 2.9493865831967436e-05,
+      "loss": 0.9022,
+      "step": 250
+    },
+    {
+      "epoch": 0.6614785992217899,
+      "grad_norm": 0.9094107747077942,
+      "learning_rate": 2.9460224283177763e-05,
+      "loss": 0.8016,
+      "step": 255
+    },
+    {
+      "epoch": 0.6744487678339819,
+      "grad_norm": 0.854141891002655,
+      "learning_rate": 2.942552082564767e-05,
+      "loss": 0.8197,
+      "step": 260
+    },
+    {
+      "epoch": 0.6874189364461738,
+      "grad_norm": 0.8813712000846863,
+      "learning_rate": 2.938975800787876e-05,
+      "loss": 0.8538,
+      "step": 265
+    },
+    {
+      "epoch": 0.7003891050583657,
+      "grad_norm": 0.8427942395210266,
+      "learning_rate": 2.9352938456168393e-05,
+      "loss": 0.7911,
+      "step": 270
+    },
+    {
+      "epoch": 0.7133592736705577,
+      "grad_norm": 0.9274711012840271,
+      "learning_rate": 2.931506487441681e-05,
+      "loss": 0.8352,
+      "step": 275
+    },
+    {
+      "epoch": 0.7263294422827496,
+      "grad_norm": 0.8024734258651733,
+      "learning_rate": 2.9276140043928533e-05,
+      "loss": 0.8198,
+      "step": 280
+    },
+    {
+      "epoch": 0.7392996108949417,
+      "grad_norm": 0.8227505087852478,
+      "learning_rate": 2.9236166823208184e-05,
+      "loss": 0.8043,
+      "step": 285
+    },
+    {
+      "epoch": 0.7522697795071336,
+      "grad_norm": 0.8620085716247559,
+      "learning_rate": 2.9195148147750504e-05,
+      "loss": 0.8011,
+      "step": 290
+    },
+    {
+      "epoch": 0.7652399481193255,
+      "grad_norm": 0.8576732873916626,
+      "learning_rate": 2.915308702982482e-05,
+      "loss": 0.7978,
+      "step": 295
+    },
+    {
+      "epoch": 0.7782101167315175,
+      "grad_norm": 1.038345217704773,
+      "learning_rate": 2.9109986558253814e-05,
+      "loss": 0.7844,
+      "step": 300
+    },
+    {
+      "epoch": 0.7911802853437094,
+      "grad_norm": 0.8221902251243591,
+      "learning_rate": 2.9065849898186693e-05,
+      "loss": 0.8405,
+      "step": 305
+    },
+    {
+      "epoch": 0.8041504539559015,
+      "grad_norm": 0.9288234114646912,
+      "learning_rate": 2.902068029086676e-05,
+      "loss": 0.7909,
+      "step": 310
+    },
+    {
+      "epoch": 0.8171206225680934,
+      "grad_norm": 0.9620817303657532,
+      "learning_rate": 2.8974481053393386e-05,
+      "loss": 0.7999,
+      "step": 315
+    },
+    {
+      "epoch": 0.8300907911802854,
+      "grad_norm": 1.018082857131958,
+      "learning_rate": 2.89272555784784e-05,
+      "loss": 0.7731,
+      "step": 320
+    },
+    {
+      "epoch": 0.8430609597924773,
+      "grad_norm": 0.9036654233932495,
+      "learning_rate": 2.8879007334196963e-05,
+      "loss": 0.775,
+      "step": 325
+    },
+    {
+      "epoch": 0.8560311284046692,
+      "grad_norm": 0.826240062713623,
+      "learning_rate": 2.882973986373287e-05,
+      "loss": 0.7607,
+      "step": 330
+    },
+    {
+      "epoch": 0.8690012970168612,
+      "grad_norm": 0.8781949281692505,
+      "learning_rate": 2.877945678511834e-05,
+      "loss": 0.7673,
+      "step": 335
+    },
+    {
+      "epoch": 0.8819714656290532,
+      "grad_norm": 1.0212292671203613,
+      "learning_rate": 2.8728161790968352e-05,
+      "loss": 0.7506,
+      "step": 340
+    },
+    {
+      "epoch": 0.8949416342412452,
+      "grad_norm": 0.9420115351676941,
+      "learning_rate": 2.8675858648209442e-05,
+      "loss": 0.7218,
+      "step": 345
+    },
+    {
+      "epoch": 0.9079118028534371,
+      "grad_norm": 1.0438215732574463,
+      "learning_rate": 2.8622551197803093e-05,
+      "loss": 0.7349,
+      "step": 350
+    },
+    {
+      "epoch": 0.920881971465629,
+      "grad_norm": 1.0321903228759766,
+      "learning_rate": 2.8568243354463657e-05,
+      "loss": 0.7259,
+      "step": 355
+    },
+    {
+      "epoch": 0.933852140077821,
+      "grad_norm": 1.0026477575302124,
+      "learning_rate": 2.8512939106370864e-05,
+      "loss": 0.7264,
+      "step": 360
+    },
+    {
+      "epoch": 0.9468223086900129,
+      "grad_norm": 1.0209153890609741,
+      "learning_rate": 2.8456642514876976e-05,
+      "loss": 0.706,
+      "step": 365
+    },
+    {
+      "epoch": 0.959792477302205,
+      "grad_norm": 0.9709693789482117,
+      "learning_rate": 2.8399357714208498e-05,
+      "loss": 0.735,
+      "step": 370
+    },
+    {
+      "epoch": 0.9727626459143969,
+      "grad_norm": 1.0189666748046875,
+      "learning_rate": 2.8341088911162594e-05,
+      "loss": 0.7149,
+      "step": 375
+    },
+    {
+      "epoch": 0.9857328145265889,
+      "grad_norm": 1.0608593225479126,
+      "learning_rate": 2.8281840384798147e-05,
+      "loss": 0.7506,
+      "step": 380
+    },
+    {
+      "epoch": 0.9987029831387808,
+      "grad_norm": 1.1326305866241455,
+      "learning_rate": 2.822161648612154e-05,
+      "loss": 0.7307,
+      "step": 385
+    },
+    {
+      "epoch": 1.0103761348897535,
+      "grad_norm": 1.1239498853683472,
+      "learning_rate": 2.8160421637767105e-05,
+      "loss": 0.624,
+      "step": 390
+    },
+    {
+      "epoch": 1.0233463035019454,
+      "grad_norm": 0.9470804929733276,
+      "learning_rate": 2.809826033367235e-05,
+      "loss": 0.5788,
+      "step": 395
+    },
+    {
+      "epoch": 1.0363164721141376,
+      "grad_norm": 1.092249870300293,
+      "learning_rate": 2.803513713874794e-05,
+      "loss": 0.6766,
+      "step": 400
+    },
+    {
+      "epoch": 1.0492866407263295,
+      "grad_norm": 1.013177752494812,
+      "learning_rate": 2.797105668854249e-05,
+      "loss": 0.6742,
+      "step": 405
+    },
+    {
+      "epoch": 1.0622568093385214,
+      "grad_norm": 0.9360693097114563,
+      "learning_rate": 2.790602368890209e-05,
+      "loss": 0.6108,
+      "step": 410
+    },
+    {
+      "epoch": 1.0752269779507133,
+      "grad_norm": 0.9409477710723877,
+      "learning_rate": 2.7840042915624794e-05,
+      "loss": 0.6305,
+      "step": 415
+    },
+    {
+      "epoch": 1.0881971465629052,
+      "grad_norm": 1.3407989740371704,
+      "learning_rate": 2.777311921410983e-05,
+      "loss": 0.5923,
+      "step": 420
+    },
+    {
+      "epoch": 1.1011673151750974,
+      "grad_norm": 1.1192437410354614,
+      "learning_rate": 2.7705257499001837e-05,
+      "loss": 0.5941,
+      "step": 425
+    },
+    {
+      "epoch": 1.1141374837872893,
+      "grad_norm": 1.3298574686050415,
+      "learning_rate": 2.763646275382991e-05,
+      "loss": 0.5569,
+      "step": 430
+    },
+    {
+      "epoch": 1.1271076523994812,
+      "grad_norm": 0.9660617113113403,
+      "learning_rate": 2.7566740030641628e-05,
+      "loss": 0.5473,
+      "step": 435
+    },
+    {
+      "epoch": 1.140077821011673,
+      "grad_norm": 1.0812163352966309,
+      "learning_rate": 2.7496094449632064e-05,
+      "loss": 0.5982,
+      "step": 440
+    },
+    {
+      "epoch": 1.153047989623865,
+      "grad_norm": 1.1607142686843872,
+      "learning_rate": 2.7424531198767783e-05,
+      "loss": 0.6103,
+      "step": 445
+    },
+    {
+      "epoch": 1.1660181582360571,
+      "grad_norm": 1.1823639869689941,
+      "learning_rate": 2.7352055533405828e-05,
+      "loss": 0.5997,
+      "step": 450
+    },
+    {
+      "epoch": 1.178988326848249,
+      "grad_norm": 1.0552186965942383,
+      "learning_rate": 2.72786727759078e-05,
+      "loss": 0.5379,
+      "step": 455
+    },
+    {
+      "epoch": 1.191958495460441,
+      "grad_norm": 1.2875730991363525,
+      "learning_rate": 2.7204388315249007e-05,
+      "loss": 0.5581,
+      "step": 460
+    },
+    {
+      "epoch": 1.2049286640726329,
+      "grad_norm": 1.1362204551696777,
+      "learning_rate": 2.712920760662271e-05,
+      "loss": 0.5421,
+      "step": 465
+    },
+    {
+      "epoch": 1.217898832684825,
+      "grad_norm": 0.9040284156799316,
+      "learning_rate": 2.705313617103951e-05,
+      "loss": 0.5178,
+      "step": 470
+    },
+    {
+      "epoch": 1.230869001297017,
+      "grad_norm": 0.9417663216590881,
+      "learning_rate": 2.6976179594921915e-05,
+      "loss": 0.5913,
+      "step": 475
+    },
+    {
+      "epoch": 1.2438391699092088,
+      "grad_norm": 1.0625457763671875,
+      "learning_rate": 2.6898343529694077e-05,
+      "loss": 0.5811,
+      "step": 480
+    },
+    {
+      "epoch": 1.2568093385214008,
+      "grad_norm": 1.1155017614364624,
+      "learning_rate": 2.6819633691366784e-05,
+      "loss": 0.5369,
+      "step": 485
+    },
+    {
+      "epoch": 1.2697795071335927,
+      "grad_norm": 1.141462802886963,
+      "learning_rate": 2.674005586011769e-05,
+      "loss": 0.5635,
+      "step": 490
+    },
+    {
+      "epoch": 1.2827496757457846,
+      "grad_norm": 1.132550597190857,
+      "learning_rate": 2.6659615879866844e-05,
+      "loss": 0.5135,
+      "step": 495
+    },
+    {
+      "epoch": 1.2957198443579767,
+      "grad_norm": 1.1744695901870728,
+      "learning_rate": 2.6578319657847523e-05,
+      "loss": 0.5586,
+      "step": 500
+    },
+    {
+      "epoch": 1.3086900129701686,
+      "grad_norm": 1.0014809370040894,
+      "learning_rate": 2.649617316417243e-05,
+      "loss": 0.5548,
+      "step": 505
+    },
+    {
+      "epoch": 1.3216601815823605,
+      "grad_norm": 1.0711734294891357,
+      "learning_rate": 2.6413182431395273e-05,
+      "loss": 0.4958,
+      "step": 510
+    },
+    {
+      "epoch": 1.3346303501945525,
+      "grad_norm": 1.3235042095184326,
+      "learning_rate": 2.6329353554067758e-05,
+      "loss": 0.4948,
+      "step": 515
+    },
+    {
+      "epoch": 1.3476005188067446,
+      "grad_norm": 1.1252357959747314,
+      "learning_rate": 2.6244692688292015e-05,
+      "loss": 0.4991,
+      "step": 520
+    },
+    {
+      "epoch": 1.3605706874189365,
+      "grad_norm": 1.1571067571640015,
+      "learning_rate": 2.615920605126852e-05,
+      "loss": 0.5401,
+      "step": 525
+    },
+    {
+      "epoch": 1.3735408560311284,
+      "grad_norm": 1.0673496723175049,
+      "learning_rate": 2.607289992083954e-05,
+      "loss": 0.5342,
+      "step": 530
+    },
+    {
+      "epoch": 1.3865110246433203,
+      "grad_norm": 1.0683461427688599,
+      "learning_rate": 2.598578063502807e-05,
+      "loss": 0.4935,
+      "step": 535
+    },
+    {
+      "epoch": 1.3994811932555122,
+      "grad_norm": 1.0260779857635498,
+      "learning_rate": 2.5897854591572448e-05,
+      "loss": 0.4831,
+      "step": 540
+    },
+    {
+      "epoch": 1.4124513618677041,
+      "grad_norm": 0.9995431900024414,
+      "learning_rate": 2.5809128247456478e-05,
+      "loss": 0.5238,
+      "step": 545
+    },
+    {
+      "epoch": 1.4254215304798963,
+      "grad_norm": 1.1451547145843506,
+      "learning_rate": 2.5719608118435266e-05,
+      "loss": 0.5088,
+      "step": 550
+    },
+    {
+      "epoch": 1.4383916990920882,
+      "grad_norm": 1.133154273033142,
+      "learning_rate": 2.5629300778556746e-05,
+      "loss": 0.4793,
+      "step": 555
+    },
+    {
+      "epoch": 1.45136186770428,
+      "grad_norm": 1.1944438219070435,
+      "learning_rate": 2.5538212859678876e-05,
+      "loss": 0.5004,
+      "step": 560
+    },
+    {
+      "epoch": 1.4643320363164722,
+      "grad_norm": 1.150473713874817,
+      "learning_rate": 2.544635105098264e-05,
+      "loss": 0.5104,
+      "step": 565
+    },
+    {
+      "epoch": 1.4773022049286642,
+      "grad_norm": 1.1394140720367432,
+      "learning_rate": 2.5353722098480788e-05,
+      "loss": 0.5064,
+      "step": 570
+    },
+    {
+      "epoch": 1.490272373540856,
+      "grad_norm": 1.1548528671264648,
+      "learning_rate": 2.5260332804522478e-05,
+      "loss": 0.5266,
+      "step": 575
+    },
+    {
+      "epoch": 1.503242542153048,
+      "grad_norm": 1.0762923955917358,
+      "learning_rate": 2.5166190027293684e-05,
+      "loss": 0.446,
+      "step": 580
+    },
+    {
+      "epoch": 1.51621271076524,
+      "grad_norm": 1.1425952911376953,
+      "learning_rate": 2.50713006803136e-05,
+      "loss": 0.4396,
+      "step": 585
+    },
+    {
+      "epoch": 1.5291828793774318,
+      "grad_norm": 1.0265759229660034,
+      "learning_rate": 2.4975671731926892e-05,
+      "loss": 0.4541,
+      "step": 590
+    },
+    {
+      "epoch": 1.5421530479896237,
+      "grad_norm": 1.0789086818695068,
+      "learning_rate": 2.4879310204792012e-05,
+      "loss": 0.4674,
+      "step": 595
+    },
+    {
+      "epoch": 1.5551232166018158,
+      "grad_norm": 1.0244364738464355,
+      "learning_rate": 2.478222317536544e-05,
+      "loss": 0.4654,
+      "step": 600
+    },
+    {
+      "epoch": 1.5680933852140078,
+      "grad_norm": 1.1058456897735596,
+      "learning_rate": 2.4684417773382037e-05,
+      "loss": 0.4599,
+      "step": 605
+    },
+    {
+      "epoch": 1.5810635538262,
+      "grad_norm": 1.031064510345459,
+      "learning_rate": 2.458590118133145e-05,
+      "loss": 0.417,
+      "step": 610
+    },
+    {
+      "epoch": 1.5940337224383918,
+      "grad_norm": 1.0740469694137573,
+      "learning_rate": 2.4486680633930658e-05,
+      "loss": 0.4712,
+      "step": 615
+    },
+    {
+      "epoch": 1.6070038910505837,
+      "grad_norm": 1.1247283220291138,
+      "learning_rate": 2.4386763417592688e-05,
+      "loss": 0.5005,
+      "step": 620
+    },
+    {
+      "epoch": 1.6199740596627756,
+      "grad_norm": 1.0776716470718384,
+      "learning_rate": 2.428615686989151e-05,
+      "loss": 0.4437,
+      "step": 625
+    },
+    {
+      "epoch": 1.6329442282749675,
+      "grad_norm": 1.0497078895568848,
+      "learning_rate": 2.4184868379023204e-05,
+      "loss": 0.436,
+      "step": 630
+    },
+    {
+      "epoch": 1.6459143968871595,
+      "grad_norm": 1.2037216424942017,
+      "learning_rate": 2.4082905383263387e-05,
+      "loss": 0.4378,
+      "step": 635
+    },
+    {
+      "epoch": 1.6588845654993514,
+      "grad_norm": 1.0372422933578491,
+      "learning_rate": 2.3980275370420986e-05,
+      "loss": 0.4469,
+      "step": 640
+    },
+    {
+      "epoch": 1.6718547341115433,
+      "grad_norm": 1.1579840183258057,
+      "learning_rate": 2.387698587728834e-05,
+      "loss": 0.434,
+      "step": 645
+    },
+    {
+      "epoch": 1.6848249027237354,
+      "grad_norm": 1.1483240127563477,
+      "learning_rate": 2.377304448908775e-05,
+      "loss": 0.4355,
+      "step": 650
+    },
+    {
+      "epoch": 1.6977950713359273,
+      "grad_norm": 1.0725358724594116,
+      "learning_rate": 2.3668458838914398e-05,
+      "loss": 0.4579,
+      "step": 655
+    },
+    {
+      "epoch": 1.7107652399481195,
+      "grad_norm": 1.1938557624816895,
+      "learning_rate": 2.3563236607175877e-05,
+      "loss": 0.4397,
+      "step": 660
+    },
+    {
+      "epoch": 1.7237354085603114,
+      "grad_norm": 1.0592831373214722,
+      "learning_rate": 2.3457385521028085e-05,
+      "loss": 0.4342,
+      "step": 665
+    },
+    {
+      "epoch": 1.7367055771725033,
+      "grad_norm": 1.191778302192688,
+      "learning_rate": 2.3350913353807826e-05,
+      "loss": 0.4108,
+      "step": 670
+    },
+    {
+      "epoch": 1.7496757457846952,
+      "grad_norm": 1.1862207651138306,
+      "learning_rate": 2.324382792446194e-05,
+      "loss": 0.4462,
+      "step": 675
+    },
+    {
+      "epoch": 1.7626459143968871,
+      "grad_norm": 1.0156222581863403,
+      "learning_rate": 2.3136137096973124e-05,
+      "loss": 0.4245,
+      "step": 680
+    },
+    {
+      "epoch": 1.775616083009079,
+      "grad_norm": 1.2515448331832886,
+      "learning_rate": 2.3027848779782388e-05,
+      "loss": 0.3718,
+      "step": 685
+    },
+    {
+      "epoch": 1.788586251621271,
+      "grad_norm": 1.0088294744491577,
+      "learning_rate": 2.291897092520834e-05,
+      "loss": 0.41,
+      "step": 690
+    },
+    {
+      "epoch": 1.801556420233463,
+      "grad_norm": 1.4383920431137085,
+      "learning_rate": 2.280951152886316e-05,
+      "loss": 0.4075,
+      "step": 695
+    },
+    {
+      "epoch": 1.814526588845655,
+      "grad_norm": 1.1313135623931885,
+      "learning_rate": 2.269947862906544e-05,
+      "loss": 0.4139,
+      "step": 700
+    },
+    {
+      "epoch": 1.827496757457847,
+      "grad_norm": 1.1307212114334106,
+      "learning_rate": 2.2588880306249873e-05,
+      "loss": 0.3916,
+      "step": 705
+    },
+    {
+      "epoch": 1.840466926070039,
+      "grad_norm": 1.10785710811615,
+      "learning_rate": 2.2477724682373868e-05,
+      "loss": 0.4109,
+      "step": 710
+    },
+    {
+      "epoch": 1.853437094682231,
+      "grad_norm": 1.2819660902023315,
+      "learning_rate": 2.236601992032108e-05,
+      "loss": 0.3913,
+      "step": 715
+    },
+    {
+      "epoch": 1.8664072632944229,
+      "grad_norm": 1.2175217866897583,
+      "learning_rate": 2.2253774223301984e-05,
+      "loss": 0.3831,
+      "step": 720
+    },
+    {
+      "epoch": 1.8793774319066148,
+      "grad_norm": 1.0192928314208984,
+      "learning_rate": 2.2140995834251427e-05,
+      "loss": 0.3897,
+      "step": 725
+    },
+    {
+      "epoch": 1.8923476005188067,
+      "grad_norm": 1.0274885892868042,
+      "learning_rate": 2.2027693035223323e-05,
+      "loss": 0.4076,
+      "step": 730
+    },
+    {
+      "epoch": 1.9053177691309986,
+      "grad_norm": 1.0526772737503052,
+      "learning_rate": 2.1913874146782417e-05,
+      "loss": 0.3463,
+      "step": 735
+    },
+    {
+      "epoch": 1.9182879377431905,
+      "grad_norm": 1.1391632556915283,
+      "learning_rate": 2.1799547527393313e-05,
+      "loss": 0.3822,
+      "step": 740
+    },
+    {
+      "epoch": 1.9312581063553826,
+      "grad_norm": 1.1479544639587402,
+      "learning_rate": 2.1684721572806573e-05,
+      "loss": 0.361,
+      "step": 745
+    },
+    {
+      "epoch": 1.9442282749675746,
+      "grad_norm": 1.2531360387802124,
+      "learning_rate": 2.1569404715442223e-05,
+      "loss": 0.3892,
+      "step": 750
+    },
+    {
+      "epoch": 1.9571984435797667,
+      "grad_norm": 1.17800772190094,
+      "learning_rate": 2.145360542377049e-05,
+      "loss": 0.4036,
+      "step": 755
+    },
+    {
+      "epoch": 1.9701686121919586,
+      "grad_norm": 1.2836965322494507,
+      "learning_rate": 2.1337332201689896e-05,
+      "loss": 0.3661,
+      "step": 760
+    },
+    {
+      "epoch": 1.9831387808041505,
+      "grad_norm": 1.1780281066894531,
+      "learning_rate": 2.1220593587902772e-05,
+      "loss": 0.3602,
+      "step": 765
+    },
+    {
+      "epoch": 1.9961089494163424,
+      "grad_norm": 1.1417920589447021,
+      "learning_rate": 2.1103398155288207e-05,
+      "loss": 0.3628,
+      "step": 770
+    },
+    {
+      "epoch": 2.007782101167315,
+      "grad_norm": 1.0694905519485474,
+      "learning_rate": 2.0985754510272474e-05,
+      "loss": 0.3022,
+      "step": 775
+    },
+    {
+      "epoch": 2.020752269779507,
+      "grad_norm": 1.2128925323486328,
+      "learning_rate": 2.0867671292197013e-05,
+      "loss": 0.3188,
+      "step": 780
+    },
+    {
+      "epoch": 2.033722438391699,
+      "grad_norm": 1.137680172920227,
+      "learning_rate": 2.0749157172683994e-05,
+      "loss": 0.3077,
+      "step": 785
+    },
+    {
+      "epoch": 2.046692607003891,
+      "grad_norm": 1.3190040588378906,
+      "learning_rate": 2.063022085499948e-05,
+      "loss": 0.292,
+      "step": 790
+    },
+    {
+      "epoch": 2.059662775616083,
+      "grad_norm": 1.097852110862732,
+      "learning_rate": 2.0510871073414315e-05,
+      "loss": 0.2982,
+      "step": 795
+    },
+    {
+      "epoch": 2.072632944228275,
+      "grad_norm": 1.1585649251937866,
+      "learning_rate": 2.039111659256269e-05,
+      "loss": 0.2592,
+      "step": 800
+    },
+    {
+      "epoch": 2.085603112840467,
+      "grad_norm": 1.065697193145752,
+      "learning_rate": 2.0270966206798523e-05,
+      "loss": 0.2759,
+      "step": 805
+    },
+    {
+      "epoch": 2.098573281452659,
+      "grad_norm": 1.011765718460083,
+      "learning_rate": 2.0150428739549595e-05,
+      "loss": 0.2843,
+      "step": 810
+    },
+    {
+      "epoch": 2.111543450064851,
+      "grad_norm": 1.0232386589050293,
+      "learning_rate": 2.002951304266962e-05,
+      "loss": 0.3209,
+      "step": 815
+    },
+    {
+      "epoch": 2.124513618677043,
+      "grad_norm": 1.1222460269927979,
+      "learning_rate": 1.990822799578819e-05,
+      "loss": 0.2832,
+      "step": 820
+    },
+    {
+      "epoch": 2.1374837872892347,
+      "grad_norm": 1.1336065530776978,
+      "learning_rate": 1.9786582505658668e-05,
+      "loss": 0.2981,
+      "step": 825
+    },
+    {
+      "epoch": 2.1504539559014266,
+      "grad_norm": 1.2448867559432983,
+      "learning_rate": 1.9664585505504134e-05,
+      "loss": 0.2898,
+      "step": 830
+    },
+    {
+      "epoch": 2.1634241245136185,
+      "grad_norm": 1.1029194593429565,
+      "learning_rate": 1.9542245954361333e-05,
+      "loss": 0.2748,
+      "step": 835
+    },
+    {
+      "epoch": 2.1763942931258105,
+      "grad_norm": 1.2409709692001343,
+      "learning_rate": 1.941957283642277e-05,
+      "loss": 0.2791,
+      "step": 840
+    },
+    {
+      "epoch": 2.189364461738003,
+      "grad_norm": 1.0632833242416382,
+      "learning_rate": 1.9296575160376953e-05,
+      "loss": 0.2774,
+      "step": 845
+    },
+    {
+      "epoch": 2.2023346303501947,
+      "grad_norm": 1.166345238685608,
+      "learning_rate": 1.9173261958746793e-05,
+      "loss": 0.2695,
+      "step": 850
+    },
+    {
+      "epoch": 2.2153047989623866,
+      "grad_norm": 1.0866575241088867,
+      "learning_rate": 1.9049642287226307e-05,
+      "loss": 0.2739,
+      "step": 855
+    },
+    {
+      "epoch": 2.2282749675745785,
+      "grad_norm": 1.0431710481643677,
+      "learning_rate": 1.8925725224015616e-05,
+      "loss": 0.256,
+      "step": 860
+    },
+    {
+      "epoch": 2.2412451361867705,
+      "grad_norm": 1.2127219438552856,
+      "learning_rate": 1.8801519869154237e-05,
+      "loss": 0.2638,
+      "step": 865
+    },
+    {
+      "epoch": 2.2542153047989624,
+      "grad_norm": 1.1814619302749634,
+      "learning_rate": 1.867703534385284e-05,
+      "loss": 0.2781,
+      "step": 870
+    },
+    {
+      "epoch": 2.2671854734111543,
+      "grad_norm": 1.1074808835983276,
+      "learning_rate": 1.8552280789823394e-05,
+      "loss": 0.2788,
+      "step": 875
+    },
+    {
+      "epoch": 2.280155642023346,
+      "grad_norm": 1.226494312286377,
+      "learning_rate": 1.8427265368607844e-05,
+      "loss": 0.2707,
+      "step": 880
+    },
+    {
+      "epoch": 2.293125810635538,
+      "grad_norm": 1.1561022996902466,
+      "learning_rate": 1.830199826090532e-05,
+      "loss": 0.2879,
+      "step": 885
+    },
+    {
+      "epoch": 2.30609597924773,
+      "grad_norm": 1.1817694902420044,
+      "learning_rate": 1.8176488665897945e-05,
+      "loss": 0.277,
+      "step": 890
+    },
+    {
+      "epoch": 2.319066147859922,
+      "grad_norm": 1.3267537355422974,
+      "learning_rate": 1.805074580057524e-05,
+      "loss": 0.2832,
+      "step": 895
+    },
+    {
+      "epoch": 2.3320363164721143,
+      "grad_norm": 1.1436495780944824,
+      "learning_rate": 1.7924778899057316e-05,
+      "loss": 0.2822,
+      "step": 900
+    },
+    {
+      "epoch": 2.345006485084306,
+      "grad_norm": 1.1241085529327393,
+      "learning_rate": 1.779859721191671e-05,
+      "loss": 0.29,
+      "step": 905
+    },
+    {
+      "epoch": 2.357976653696498,
+      "grad_norm": 1.5335941314697266,
+      "learning_rate": 1.7672210005499088e-05,
+      "loss": 0.2511,
+      "step": 910
+    },
+    {
+      "epoch": 2.37094682230869,
+      "grad_norm": 1.233126163482666,
+      "learning_rate": 1.754562656124272e-05,
+      "loss": 0.2595,
+      "step": 915
+    },
+    {
+      "epoch": 2.383916990920882,
+      "grad_norm": 1.1877148151397705,
+      "learning_rate": 1.7418856174996915e-05,
+      "loss": 0.2536,
+      "step": 920
+    },
+    {
+      "epoch": 2.396887159533074,
+      "grad_norm": 1.3099548816680908,
+      "learning_rate": 1.729190815633935e-05,
+      "loss": 0.2719,
+      "step": 925
+    },
+    {
+      "epoch": 2.4098573281452658,
+      "grad_norm": 1.1045476198196411,
+      "learning_rate": 1.71647918278924e-05,
+      "loss": 0.2438,
+      "step": 930
+    },
+    {
+      "epoch": 2.4228274967574577,
+      "grad_norm": 1.1141847372055054,
+      "learning_rate": 1.7037516524638548e-05,
+      "loss": 0.2439,
+      "step": 935
+    },
+    {
+      "epoch": 2.43579766536965,
+      "grad_norm": 1.0792944431304932,
+      "learning_rate": 1.6910091593234823e-05,
+      "loss": 0.2265,
+      "step": 940
+    },
+    {
+      "epoch": 2.448767833981842,
+      "grad_norm": 1.1306582689285278,
+      "learning_rate": 1.6782526391326425e-05,
+      "loss": 0.2525,
+      "step": 945
+    },
+    {
+      "epoch": 2.461738002594034,
+      "grad_norm": 1.1637719869613647,
+      "learning_rate": 1.6654830286859533e-05,
+      "loss": 0.2327,
+      "step": 950
+    },
+    {
+      "epoch": 2.4747081712062258,
+      "grad_norm": 1.0156742334365845,
+      "learning_rate": 1.6527012657393372e-05,
+      "loss": 0.2229,
+      "step": 955
+    },
+    {
+      "epoch": 2.4876783398184177,
+      "grad_norm": 1.0547713041305542,
+      "learning_rate": 1.6399082889411525e-05,
+      "loss": 0.2477,
+      "step": 960
+    },
+    {
+      "epoch": 2.5006485084306096,
+      "grad_norm": 1.1955467462539673,
+      "learning_rate": 1.627105037763264e-05,
+      "loss": 0.2333,
+      "step": 965
+    },
+    {
+      "epoch": 2.5136186770428015,
+      "grad_norm": 1.1450884342193604,
+      "learning_rate": 1.614292452432052e-05,
+      "loss": 0.2504,
+      "step": 970
+    },
+    {
+      "epoch": 2.5265888456549934,
+      "grad_norm": 1.1756523847579956,
+      "learning_rate": 1.6014714738593637e-05,
+      "loss": 0.2415,
+      "step": 975
+    },
+    {
+      "epoch": 2.5395590142671853,
+      "grad_norm": 1.1718183755874634,
+      "learning_rate": 1.588643043573419e-05,
+      "loss": 0.2289,
+      "step": 980
+    },
+    {
+      "epoch": 2.5525291828793772,
+      "grad_norm": 1.236280918121338,
+      "learning_rate": 1.575808103649663e-05,
+      "loss": 0.2355,
+      "step": 985
+    },
+    {
+      "epoch": 2.565499351491569,
+      "grad_norm": 1.082622766494751,
+      "learning_rate": 1.5629675966415877e-05,
+      "loss": 0.2434,
+      "step": 990
+    },
+    {
+      "epoch": 2.5784695201037615,
+      "grad_norm": 1.3068712949752808,
+      "learning_rate": 1.5501224655115118e-05,
+      "loss": 0.2186,
+      "step": 995
+    },
+    {
+      "epoch": 2.5914396887159534,
+      "grad_norm": 1.0971678495407104,
+      "learning_rate": 1.5372736535613347e-05,
+      "loss": 0.2442,
+      "step": 1000
+    },
+    {
+      "epoch": 2.6044098573281453,
+      "grad_norm": 1.1903417110443115,
+      "learning_rate": 1.5244221043632605e-05,
+      "loss": 0.2455,
+      "step": 1005
+    },
+    {
+      "epoch": 2.6173800259403373,
+      "grad_norm": 1.1289457082748413,
+      "learning_rate": 1.511568761690509e-05,
+      "loss": 0.2531,
+      "step": 1010
+    },
+    {
+      "epoch": 2.630350194552529,
+      "grad_norm": 1.1403872966766357,
+      "learning_rate": 1.4987145694480054e-05,
+      "loss": 0.2355,
+      "step": 1015
+    },
+    {
+      "epoch": 2.643320363164721,
+      "grad_norm": 1.3029553890228271,
+      "learning_rate": 1.4858604716030655e-05,
+      "loss": 0.2496,
+      "step": 1020
+    },
+    {
+      "epoch": 2.656290531776913,
+      "grad_norm": 0.9976205825805664,
+      "learning_rate": 1.4730074121160718e-05,
+      "loss": 0.2097,
+      "step": 1025
+    },
+    {
+      "epoch": 2.669260700389105,
+      "grad_norm": 1.191940426826477,
+      "learning_rate": 1.4601563348711543e-05,
+      "loss": 0.2129,
+      "step": 1030
+    },
+    {
+      "epoch": 2.6822308690012973,
+      "grad_norm": 1.2161098718643188,
+      "learning_rate": 1.4473081836068728e-05,
+      "loss": 0.2274,
+      "step": 1035
+    },
+    {
+      "epoch": 2.695201037613489,
+      "grad_norm": 1.191472053527832,
+      "learning_rate": 1.4344639018469145e-05,
+      "loss": 0.2332,
+      "step": 1040
+    },
+    {
+      "epoch": 2.708171206225681,
+      "grad_norm": 1.1076041460037231,
+      "learning_rate": 1.4216244328308029e-05,
+      "loss": 0.2337,
+      "step": 1045
+    },
+    {
+      "epoch": 2.721141374837873,
+      "grad_norm": 1.1517136096954346,
+      "learning_rate": 1.4087907194446307e-05,
+      "loss": 0.227,
+      "step": 1050
+    },
+    {
+      "epoch": 2.734111543450065,
+      "grad_norm": 1.2073043584823608,
+      "learning_rate": 1.395963704151817e-05,
+      "loss": 0.2006,
+      "step": 1055
+    },
+    {
+      "epoch": 2.747081712062257,
+      "grad_norm": 1.2052357196807861,
+      "learning_rate": 1.3831443289238951e-05,
+      "loss": 0.2119,
+      "step": 1060
+    },
+    {
+      "epoch": 2.7600518806744487,
+      "grad_norm": 1.129930853843689,
+      "learning_rate": 1.3703335351713403e-05,
+      "loss": 0.1809,
+      "step": 1065
+    },
+    {
+      "epoch": 2.7730220492866406,
+      "grad_norm": 1.2224830389022827,
+      "learning_rate": 1.3575322636744325e-05,
+      "loss": 0.2393,
+      "step": 1070
+    },
+    {
+      "epoch": 2.7859922178988326,
+      "grad_norm": 1.2128647565841675,
+      "learning_rate": 1.3447414545141712e-05,
+      "loss": 0.2124,
+      "step": 1075
+    },
+    {
+      "epoch": 2.7989623865110245,
+      "grad_norm": 1.084707260131836,
+      "learning_rate": 1.3319620470032379e-05,
+      "loss": 0.2222,
+      "step": 1080
+    },
+    {
+      "epoch": 2.8119325551232164,
+      "grad_norm": 1.155920147895813,
+      "learning_rate": 1.3191949796170156e-05,
+      "loss": 0.2154,
+      "step": 1085
+    },
+    {
+      "epoch": 2.8249027237354083,
+      "grad_norm": 1.206000566482544,
+      "learning_rate": 1.3064411899246728e-05,
+      "loss": 0.1929,
+      "step": 1090
+    },
+    {
+      "epoch": 2.8378728923476007,
+      "grad_norm": 1.216287612915039,
+      "learning_rate": 1.293701614520309e-05,
+      "loss": 0.2301,
+      "step": 1095
+    },
+    {
+      "epoch": 2.8508430609597926,
+      "grad_norm": 1.1954846382141113,
+      "learning_rate": 1.280977188954176e-05,
+      "loss": 0.2108,
+      "step": 1100
+    },
+    {
+      "epoch": 2.8638132295719845,
+      "grad_norm": 1.132566213607788,
+      "learning_rate": 1.2682688476639753e-05,
+      "loss": 0.2138,
+      "step": 1105
+    },
+    {
+      "epoch": 2.8767833981841764,
+      "grad_norm": 1.0993660688400269,
+      "learning_rate": 1.2555775239062323e-05,
+      "loss": 0.2069,
+      "step": 1110
+    },
+    {
+      "epoch": 2.8897535667963683,
+      "grad_norm": 1.035963535308838,
+      "learning_rate": 1.242904149687766e-05,
+      "loss": 0.2059,
+      "step": 1115
+    },
+    {
+      "epoch": 2.90272373540856,
+      "grad_norm": 1.0626498460769653,
+      "learning_rate": 1.2302496556972424e-05,
+      "loss": 0.2147,
+      "step": 1120
+    },
+    {
+      "epoch": 2.915693904020752,
+      "grad_norm": 1.0666407346725464,
+      "learning_rate": 1.2176149712368298e-05,
+      "loss": 0.1955,
+      "step": 1125
+    },
+    {
+      "epoch": 2.9286640726329445,
+      "grad_norm": 1.1655992269515991,
+      "learning_rate": 1.2050010241539524e-05,
+      "loss": 0.2158,
+      "step": 1130
+    },
+    {
+      "epoch": 2.9416342412451364,
+      "grad_norm": 1.0219794511795044,
+      "learning_rate": 1.1924087407731552e-05,
+      "loss": 0.2001,
+      "step": 1135
+    },
+    {
+      "epoch": 2.9546044098573283,
+      "grad_norm": 1.1955561637878418,
+      "learning_rate": 1.1798390458280721e-05,
+      "loss": 0.2017,
+      "step": 1140
+    },
+    {
+      "epoch": 2.96757457846952,
+      "grad_norm": 1.2346644401550293,
+      "learning_rate": 1.1672928623935246e-05,
+      "loss": 0.1987,
+      "step": 1145
+    },
+    {
+      "epoch": 2.980544747081712,
+      "grad_norm": 1.2184339761734009,
+      "learning_rate": 1.1547711118177288e-05,
+      "loss": 0.185,
+      "step": 1150
+    },
+    {
+      "epoch": 2.993514915693904,
+      "grad_norm": 1.226433277130127,
+      "learning_rate": 1.1422747136546358e-05,
+      "loss": 0.1663,
+      "step": 1155
+    },
+    {
+      "epoch": 3.0051880674448768,
+      "grad_norm": 1.0118134021759033,
+      "learning_rate": 1.1298045855964048e-05,
+      "loss": 0.1612,
+      "step": 1160
+    },
+    {
+      "epoch": 3.0181582360570687,
+      "grad_norm": 1.0349138975143433,
+      "learning_rate": 1.1173616434060079e-05,
+      "loss": 0.1566,
+      "step": 1165
+    },
+    {
+      "epoch": 3.0311284046692606,
+      "grad_norm": 1.2392154932022095,
+      "learning_rate": 1.1049468008499827e-05,
+      "loss": 0.1645,
+      "step": 1170
+    },
+    {
+      "epoch": 3.0440985732814525,
+      "grad_norm": 1.0370545387268066,
+      "learning_rate": 1.092560969631326e-05,
+      "loss": 0.1424,
+      "step": 1175
+    },
+    {
+      "epoch": 3.0570687418936444,
+      "grad_norm": 1.106817603111267,
+      "learning_rate": 1.0802050593225438e-05,
+      "loss": 0.1613,
+      "step": 1180
+    },
+    {
+      "epoch": 3.0700389105058368,
+      "grad_norm": 1.019032597541809,
+      "learning_rate": 1.0678799772988535e-05,
+      "loss": 0.1474,
+      "step": 1185
+    },
+    {
+      "epoch": 3.0830090791180287,
+      "grad_norm": 1.195290207862854,
+      "learning_rate": 1.0555866286715493e-05,
+      "loss": 0.1538,
+      "step": 1190
+    },
+    {
+      "epoch": 3.0959792477302206,
+      "grad_norm": 1.1602429151535034,
+      "learning_rate": 1.0433259162215361e-05,
+      "loss": 0.1376,
+      "step": 1195
+    },
+    {
+      "epoch": 3.1089494163424125,
+      "grad_norm": 0.8880344033241272,
+      "learning_rate": 1.031098740333031e-05,
+      "loss": 0.1495,
+      "step": 1200
+    },
+    {
+      "epoch": 3.1219195849546044,
+      "grad_norm": 1.0927754640579224,
+      "learning_rate": 1.0189059989274412e-05,
+      "loss": 0.1605,
+      "step": 1205
+    },
+    {
+      "epoch": 3.1348897535667963,
+      "grad_norm": 0.9393549561500549,
+      "learning_rate": 1.0067485873974262e-05,
+      "loss": 0.1277,
+      "step": 1210
+    },
+    {
+      "epoch": 3.1478599221789882,
+      "grad_norm": 0.9691026210784912,
+      "learning_rate": 9.946273985411406e-06,
+      "loss": 0.142,
+      "step": 1215
+    },
+    {
+      "epoch": 3.16083009079118,
+      "grad_norm": 0.9570019245147705,
+      "learning_rate": 9.825433224966724e-06,
+      "loss": 0.1317,
+      "step": 1220
+    },
+    {
+      "epoch": 3.173800259403372,
+      "grad_norm": 0.9395800232887268,
+      "learning_rate": 9.70497246676673e-06,
+      "loss": 0.1428,
+      "step": 1225
+    },
+    {
+      "epoch": 3.1867704280155644,
+      "grad_norm": 1.0369064807891846,
+      "learning_rate": 9.58490055703189e-06,
+      "loss": 0.1429,
+      "step": 1230
+    },
+    {
+      "epoch": 3.1997405966277563,
+      "grad_norm": 1.1036053895950317,
+      "learning_rate": 9.465226313426992e-06,
+      "loss": 0.1584,
+      "step": 1235
+    },
+    {
+      "epoch": 3.2127107652399483,
+      "grad_norm": 1.1598259210586548,
+      "learning_rate": 9.345958524413598e-06,
+      "loss": 0.1557,
+      "step": 1240
+    },
+    {
+      "epoch": 3.22568093385214,
+      "grad_norm": 1.0509331226348877,
+      "learning_rate": 9.227105948604643e-06,
+      "loss": 0.161,
+      "step": 1245
+    },
+    {
+      "epoch": 3.238651102464332,
+      "grad_norm": 0.964561402797699,
+      "learning_rate": 9.108677314121266e-06,
+      "loss": 0.1309,
+      "step": 1250
+    },
+    {
+      "epoch": 3.251621271076524,
+      "grad_norm": 1.1135774850845337,
+      "learning_rate": 8.990681317951806e-06,
+      "loss": 0.1424,
+      "step": 1255
+    },
+    {
+      "epoch": 3.264591439688716,
+      "grad_norm": 1.0265324115753174,
+      "learning_rate": 8.873126625313157e-06,
+      "loss": 0.1428,
+      "step": 1260
+    },
+    {
+      "epoch": 3.277561608300908,
+      "grad_norm": 1.0568013191223145,
+      "learning_rate": 8.756021869014407e-06,
+      "loss": 0.1369,
+      "step": 1265
+    },
+    {
+      "epoch": 3.2905317769130997,
+      "grad_norm": 1.1729710102081299,
+      "learning_rate": 8.639375648822866e-06,
+      "loss": 0.1572,
+      "step": 1270
+    },
+    {
+      "epoch": 3.3035019455252916,
+      "grad_norm": 1.1114223003387451,
+      "learning_rate": 8.523196530832569e-06,
+      "loss": 0.134,
+      "step": 1275
+    },
+    {
+      "epoch": 3.3164721141374836,
+      "grad_norm": 1.010528564453125,
+      "learning_rate": 8.407493046835155e-06,
+      "loss": 0.1334,
+      "step": 1280
+    },
+    {
+      "epoch": 3.329442282749676,
+      "grad_norm": 1.123506784439087,
+      "learning_rate": 8.292273693693375e-06,
+      "loss": 0.1504,
+      "step": 1285
+    },
+    {
+      "epoch": 3.342412451361868,
+      "grad_norm": 1.037865161895752,
+      "learning_rate": 8.177546932717088e-06,
+      "loss": 0.1429,
+      "step": 1290
+    },
+    {
+      "epoch": 3.3553826199740597,
+      "grad_norm": 1.2145488262176514,
+      "learning_rate": 8.063321189041864e-06,
+      "loss": 0.1259,
+      "step": 1295
+    },
+    {
+      "epoch": 3.3683527885862516,
+      "grad_norm": 1.0262635946273804,
+      "learning_rate": 7.949604851010328e-06,
+      "loss": 0.1319,
+      "step": 1300
+    },
+    {
+      "epoch": 3.3813229571984436,
+      "grad_norm": 1.0105372667312622,
+      "learning_rate": 7.836406269556116e-06,
+      "loss": 0.1352,
+      "step": 1305
+    },
+    {
+      "epoch": 3.3942931258106355,
+      "grad_norm": 1.5561460256576538,
+      "learning_rate": 7.723733757590605e-06,
+      "loss": 0.1477,
+      "step": 1310
+    },
+    {
+      "epoch": 3.4072632944228274,
+      "grad_norm": 1.1214747428894043,
+      "learning_rate": 7.611595589392466e-06,
+      "loss": 0.1365,
+      "step": 1315
+    },
+    {
+      "epoch": 3.4202334630350193,
+      "grad_norm": 1.2157924175262451,
+      "learning_rate": 7.500000000000004e-06,
+      "loss": 0.1444,
+      "step": 1320
+    },
+    {
+      "epoch": 3.4332036316472117,
+      "grad_norm": 1.1298187971115112,
+      "learning_rate": 7.388955184606416e-06,
+      "loss": 0.1433,
+      "step": 1325
+    },
+    {
+      "epoch": 3.4461738002594036,
+      "grad_norm": 1.063402533531189,
+      "learning_rate": 7.2784692979579856e-06,
+      "loss": 0.1361,
+      "step": 1330
+    },
+    {
+      "epoch": 3.4591439688715955,
+      "grad_norm": 1.2517809867858887,
+      "learning_rate": 7.168550453755184e-06,
+      "loss": 0.1374,
+      "step": 1335
+    },
+    {
+      "epoch": 3.4721141374837874,
+      "grad_norm": 1.1007767915725708,
+      "learning_rate": 7.05920672405687e-06,
+      "loss": 0.1303,
+      "step": 1340
+    },
+    {
+      "epoch": 3.4850843060959793,
+      "grad_norm": 1.0745270252227783,
+      "learning_rate": 6.950446138687472e-06,
+      "loss": 0.138,
+      "step": 1345
+    },
+    {
+      "epoch": 3.498054474708171,
+      "grad_norm": 1.012318730354309,
+      "learning_rate": 6.842276684647323e-06,
+      "loss": 0.1525,
+      "step": 1350
+    },
+    {
+      "epoch": 3.511024643320363,
+      "grad_norm": 1.0598697662353516,
+      "learning_rate": 6.734706305526136e-06,
+      "loss": 0.1195,
+      "step": 1355
+    },
+    {
+      "epoch": 3.523994811932555,
+      "grad_norm": 1.050558090209961,
+      "learning_rate": 6.627742900919611e-06,
+      "loss": 0.1305,
+      "step": 1360
+    },
+    {
+      "epoch": 3.536964980544747,
+      "grad_norm": 0.9080147743225098,
+      "learning_rate": 6.521394325849376e-06,
+      "loss": 0.1117,
+      "step": 1365
+    },
+    {
+      "epoch": 3.549935149156939,
+      "grad_norm": 0.913749098777771,
+      "learning_rate": 6.4156683901860865e-06,
+      "loss": 0.1324,
+      "step": 1370
+    },
+    {
+      "epoch": 3.5629053177691308,
+      "grad_norm": 1.009790301322937,
+      "learning_rate": 6.3105728580759185e-06,
+      "loss": 0.1217,
+      "step": 1375
+    },
+    {
+      "epoch": 3.5758754863813227,
+      "grad_norm": 0.9976526498794556,
+      "learning_rate": 6.206115447370415e-06,
+      "loss": 0.125,
+      "step": 1380
+    },
+    {
+      "epoch": 3.588845654993515,
+      "grad_norm": 0.9304535984992981,
+      "learning_rate": 6.102303829059672e-06,
+      "loss": 0.1182,
+      "step": 1385
+    },
+    {
+      "epoch": 3.601815823605707,
+      "grad_norm": 1.0728025436401367,
+      "learning_rate": 5.9991456267090504e-06,
+      "loss": 0.1104,
+      "step": 1390
+    },
+    {
+      "epoch": 3.614785992217899,
+      "grad_norm": 0.9775097370147705,
+      "learning_rate": 5.896648415899307e-06,
+      "loss": 0.1181,
+      "step": 1395
+    },
+    {
+      "epoch": 3.627756160830091,
+      "grad_norm": 0.8441417217254639,
+      "learning_rate": 5.7948197236702475e-06,
+      "loss": 0.111,
+      "step": 1400
+    },
+    {
+      "epoch": 3.6407263294422827,
+      "grad_norm": 1.0574108362197876,
+      "learning_rate": 5.693667027968016e-06,
+      "loss": 0.1254,
+      "step": 1405
+    },
+    {
+      "epoch": 3.6536964980544746,
+      "grad_norm": 1.1657330989837646,
+      "learning_rate": 5.593197757095912e-06,
+      "loss": 0.1262,
+      "step": 1410
+    },
+    {
+      "epoch": 3.6666666666666665,
+      "grad_norm": 0.9888033270835876,
+      "learning_rate": 5.493419289168871e-06,
+      "loss": 0.1134,
+      "step": 1415
+    },
+    {
+      "epoch": 3.679636835278859,
+      "grad_norm": 1.235443115234375,
+      "learning_rate": 5.394338951571674e-06,
+      "loss": 0.1193,
+      "step": 1420
+    },
+    {
+      "epoch": 3.692607003891051,
+      "grad_norm": 0.9676278233528137,
+      "learning_rate": 5.295964020420815e-06,
+      "loss": 0.1208,
+      "step": 1425
+    },
+    {
+      "epoch": 3.7055771725032427,
+      "grad_norm": 0.8817445039749146,
+      "learning_rate": 5.198301720030186e-06,
+      "loss": 0.108,
+      "step": 1430
+    },
+    {
+      "epoch": 3.7185473411154346,
+      "grad_norm": 1.0504984855651855,
+      "learning_rate": 5.10135922238056e-06,
+      "loss": 0.122,
+      "step": 1435
+    },
+    {
+      "epoch": 3.7315175097276265,
+      "grad_norm": 0.9180973768234253,
+      "learning_rate": 5.005143646592872e-06,
+      "loss": 0.1109,
+      "step": 1440
+    },
+    {
+      "epoch": 3.7444876783398184,
+      "grad_norm": 0.9016922116279602,
+      "learning_rate": 4.909662058405456e-06,
+      "loss": 0.1225,
+      "step": 1445
+    },
+    {
+      "epoch": 3.7574578469520103,
+      "grad_norm": 0.9249210357666016,
+      "learning_rate": 4.814921469655128e-06,
+      "loss": 0.1295,
+      "step": 1450
+    },
+    {
+      "epoch": 3.7704280155642023,
+      "grad_norm": 0.8599777817726135,
+      "learning_rate": 4.720928837762271e-06,
+      "loss": 0.1376,
+      "step": 1455
+    },
+    {
+      "epoch": 3.783398184176394,
+      "grad_norm": 1.2021362781524658,
+      "learning_rate": 4.627691065219927e-06,
+      "loss": 0.1221,
+      "step": 1460
+    },
+    {
+      "epoch": 3.796368352788586,
+      "grad_norm": 1.0530153512954712,
+      "learning_rate": 4.5352149990868625e-06,
+      "loss": 0.1076,
+      "step": 1465
+    },
+    {
+      "epoch": 3.809338521400778,
+      "grad_norm": 1.143200159072876,
+      "learning_rate": 4.443507430484787e-06,
+      "loss": 0.1391,
+      "step": 1470
+    },
+    {
+      "epoch": 3.82230869001297,
+      "grad_norm": 0.861686110496521,
+      "learning_rate": 4.352575094099602e-06,
+      "loss": 0.1231,
+      "step": 1475
+    },
+    {
+      "epoch": 3.8352788586251623,
+      "grad_norm": 1.0032557249069214,
+      "learning_rate": 4.262424667686841e-06,
+      "loss": 0.1137,
+      "step": 1480
+    },
+    {
+      "epoch": 3.848249027237354,
+      "grad_norm": 0.9715796113014221,
+      "learning_rate": 4.173062771581291e-06,
+      "loss": 0.1135,
+      "step": 1485
+    },
+    {
+      "epoch": 3.861219195849546,
+      "grad_norm": 1.0717939138412476,
+      "learning_rate": 4.084495968210794e-06,
+      "loss": 0.1208,
+      "step": 1490
+    },
+    {
+      "epoch": 3.874189364461738,
+      "grad_norm": 0.9647429585456848,
+      "learning_rate": 3.99673076161435e-06,
+      "loss": 0.1068,
+      "step": 1495
+    },
+    {
+      "epoch": 3.88715953307393,
+      "grad_norm": 0.982612133026123,
+      "learning_rate": 3.909773596964468e-06,
+      "loss": 0.1375,
+      "step": 1500
+    },
+    {
+      "epoch": 3.900129701686122,
+      "grad_norm": 0.8596996665000916,
+      "learning_rate": 3.823630860093853e-06,
+      "loss": 0.1152,
+      "step": 1505
+    },
+    {
+      "epoch": 3.9130998702983137,
+      "grad_norm": 1.1000279188156128,
+      "learning_rate": 3.738308877026459e-06,
+      "loss": 0.1147,
+      "step": 1510
+    },
+    {
+      "epoch": 3.926070038910506,
+      "grad_norm": 0.9659633040428162,
+      "learning_rate": 3.6538139135129213e-06,
+      "loss": 0.1019,
+      "step": 1515
+    },
+    {
+      "epoch": 3.939040207522698,
+      "grad_norm": 0.8772383332252502,
+      "learning_rate": 3.570152174570438e-06,
+      "loss": 0.0971,
+      "step": 1520
+    },
+    {
+      "epoch": 3.95201037613489,
+      "grad_norm": 0.9329488277435303,
+      "learning_rate": 3.4873298040270807e-06,
+      "loss": 0.1127,
+      "step": 1525
+    },
+    {
+      "epoch": 3.964980544747082,
+      "grad_norm": 0.9905430674552917,
+      "learning_rate": 3.4053528840706076e-06,
+      "loss": 0.1263,
+      "step": 1530
+    },
+    {
+      "epoch": 3.9779507133592737,
+      "grad_norm": 0.9845380783081055,
+      "learning_rate": 3.324227434801817e-06,
+      "loss": 0.1157,
+      "step": 1535
+    },
+    {
+      "epoch": 3.9909208819714657,
+      "grad_norm": 0.8644956350326538,
+      "learning_rate": 3.243959413792466e-06,
+      "loss": 0.1166,
+      "step": 1540
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1930,
+  "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": 2.3448237423554724e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

32_128_e5_3e-5/checkpoint-1544/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b54d15b6a19c1b0185a13af26c266453a69d60551948a83f1872822676a12e8
+size 7736

32_128_e5_3e-5/checkpoint-1544/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)

32_128_e5_3e-5/checkpoint-1930/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

32_128_e5_3e-5/checkpoint-1930/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": [
+    "up_proj",
+    "gate_proj",
+    "down_proj",
+    "q_proj",
+    "v_proj",
+    "k_proj",
+    "o_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

32_128_e5_3e-5/checkpoint-1930/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a3098c6cef92d01887bfc1b641e5cc5101149edb4597a0d5e1abe61411b11930
+size 791751704

32_128_e5_3e-5/checkpoint-1930/latest

@@ -0,0 +1 @@
+global_step1930

32_128_e5_3e-5/checkpoint-1930/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:66444374e6488352be4b24604721b9f9ea2dfaa1a2245db26f10f7ede74bbab3
+size 15920

32_128_e5_3e-5/checkpoint-1930/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:901ec0011d27c70f356b4ad3a73eb35e6f759abb84c3a7aa88a0a2db2898c9c5
+size 15920

32_128_e5_3e-5/checkpoint-1930/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b8253bca7cdc0beafbdd287f63a806f9b44d8aebd91e94c620196e2f64666b22
+size 15920