Upload folder using huggingface_hub

14_128_e5_3e-5/checkpoint-1135/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

14_128_e5_3e-5/checkpoint-1135/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": [
+    "v_proj",
+    "down_proj",
+    "gate_proj",
+    "o_proj",
+    "k_proj",
+    "q_proj",
+    "up_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

14_128_e5_3e-5/checkpoint-1135/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d057f58209357224907715596cce602c312bf3897d8c6c9f1c7f0f393d0e2af8
+size 791751704

14_128_e5_3e-5/checkpoint-1135/latest

@@ -0,0 +1 @@
+global_step1135

14_128_e5_3e-5/checkpoint-1135/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:09f413d07ae3de701fb5797e469f17bae32c76ff9fee9aac8e9e63c07e791be2
+size 15920

14_128_e5_3e-5/checkpoint-1135/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a1c32d356ac3ffe07048eb59cd8ca2cc00fbb04962aaf852e29f332d9cecbbe
+size 15920

14_128_e5_3e-5/checkpoint-1135/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dcc6fa9036e69969a41efe004f45c4c0a8846d6400d575a3878944414f5377b9
+size 15920

14_128_e5_3e-5/checkpoint-1135/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:785d7de1ad31661ebc86a87a18007e362fa652c614e8a2bf8272bc30f8ab3af1
+size 15920

14_128_e5_3e-5/checkpoint-1135/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:873b73175be9f8e650cc415fadfbe1a0a4e9499d87aae427143aa9f107933d92
+size 15920

14_128_e5_3e-5/checkpoint-1135/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7ff82cd7c85e31b926d60a5749a8616a1f0c119634d103ff4f9e057ee4d88af1
+size 15920

14_128_e5_3e-5/checkpoint-1135/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f973d78740009f107d724f8428ac4b821fa3d82354364106090e66c52572eec7
+size 15920

14_128_e5_3e-5/checkpoint-1135/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:399c4c5fffe3a18f5cf8ecc6e8f7f4a32f7cfad725fbc3ec518b72f93a30875c
+size 15920

14_128_e5_3e-5/checkpoint-1135/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:993567cae0a345bae33668f28d3186a56502e7942402e853636e154873db574b
+size 1064

14_128_e5_3e-5/checkpoint-1135/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
+  }
+}

14_128_e5_3e-5/checkpoint-1135/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
+}

14_128_e5_3e-5/checkpoint-1135/trainer_state.json

@@ -0,0 +1,1623 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 1135,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.022026431718061675,
+      "grad_norm": 1.2465029954910278,
+      "learning_rate": 2.1052631578947366e-06,
+      "loss": 1.3594,
+      "step": 5
+    },
+    {
+      "epoch": 0.04405286343612335,
+      "grad_norm": 0.7338076233863831,
+      "learning_rate": 4.736842105263158e-06,
+      "loss": 1.3607,
+      "step": 10
+    },
+    {
+      "epoch": 0.06607929515418502,
+      "grad_norm": 0.5697624087333679,
+      "learning_rate": 7.3684210526315784e-06,
+      "loss": 1.3328,
+      "step": 15
+    },
+    {
+      "epoch": 0.0881057268722467,
+      "grad_norm": 0.661296010017395,
+      "learning_rate": 9.999999999999999e-06,
+      "loss": 1.3393,
+      "step": 20
+    },
+    {
+      "epoch": 0.11013215859030837,
+      "grad_norm": 0.6124210357666016,
+      "learning_rate": 1.263157894736842e-05,
+      "loss": 1.2621,
+      "step": 25
+    },
+    {
+      "epoch": 0.13215859030837004,
+      "grad_norm": 0.5665815472602844,
+      "learning_rate": 1.5263157894736842e-05,
+      "loss": 1.2624,
+      "step": 30
+    },
+    {
+      "epoch": 0.15418502202643172,
+      "grad_norm": 0.5574439167976379,
+      "learning_rate": 1.7894736842105264e-05,
+      "loss": 1.2487,
+      "step": 35
+    },
+    {
+      "epoch": 0.1762114537444934,
+      "grad_norm": 0.4665200412273407,
+      "learning_rate": 2.0526315789473685e-05,
+      "loss": 1.2335,
+      "step": 40
+    },
+    {
+      "epoch": 0.19823788546255505,
+      "grad_norm": 0.4646995961666107,
+      "learning_rate": 2.3157894736842103e-05,
+      "loss": 1.1568,
+      "step": 45
+    },
+    {
+      "epoch": 0.22026431718061673,
+      "grad_norm": 0.5178118348121643,
+      "learning_rate": 2.578947368421053e-05,
+      "loss": 1.214,
+      "step": 50
+    },
+    {
+      "epoch": 0.2422907488986784,
+      "grad_norm": 0.41648760437965393,
+      "learning_rate": 2.8421052631578946e-05,
+      "loss": 1.151,
+      "step": 55
+    },
+    {
+      "epoch": 0.2643171806167401,
+      "grad_norm": 0.5988641977310181,
+      "learning_rate": 2.9999745210076202e-05,
+      "loss": 1.1945,
+      "step": 60
+    },
+    {
+      "epoch": 0.28634361233480177,
+      "grad_norm": 0.4767996370792389,
+      "learning_rate": 2.9996878922838097e-05,
+      "loss": 1.0975,
+      "step": 65
+    },
+    {
+      "epoch": 0.30837004405286345,
+      "grad_norm": 0.547401487827301,
+      "learning_rate": 2.9990828471561044e-05,
+      "loss": 1.1063,
+      "step": 70
+    },
+    {
+      "epoch": 0.3303964757709251,
+      "grad_norm": 0.5365089178085327,
+      "learning_rate": 2.998159514088762e-05,
+      "loss": 1.1153,
+      "step": 75
+    },
+    {
+      "epoch": 0.3524229074889868,
+      "grad_norm": 0.5236433148384094,
+      "learning_rate": 2.9969180891255046e-05,
+      "loss": 1.088,
+      "step": 80
+    },
+    {
+      "epoch": 0.3744493392070485,
+      "grad_norm": 0.6282631754875183,
+      "learning_rate": 2.995358835847891e-05,
+      "loss": 1.0695,
+      "step": 85
+    },
+    {
+      "epoch": 0.3964757709251101,
+      "grad_norm": 0.6258669495582581,
+      "learning_rate": 2.9934820853193538e-05,
+      "loss": 1.0281,
+      "step": 90
+    },
+    {
+      "epoch": 0.4185022026431718,
+      "grad_norm": 0.6035173535346985,
+      "learning_rate": 2.991288236014907e-05,
+      "loss": 1.0368,
+      "step": 95
+    },
+    {
+      "epoch": 0.44052863436123346,
+      "grad_norm": 0.553928554058075,
+      "learning_rate": 2.9887777537365416e-05,
+      "loss": 0.9775,
+      "step": 100
+    },
+    {
+      "epoch": 0.46255506607929514,
+      "grad_norm": 0.6714206337928772,
+      "learning_rate": 2.985951171514326e-05,
+      "loss": 0.9667,
+      "step": 105
+    },
+    {
+      "epoch": 0.4845814977973568,
+      "grad_norm": 0.640594482421875,
+      "learning_rate": 2.982809089493231e-05,
+      "loss": 0.9989,
+      "step": 110
+    },
+    {
+      "epoch": 0.5066079295154186,
+      "grad_norm": 0.6225477457046509,
+      "learning_rate": 2.9793521748057065e-05,
+      "loss": 1.011,
+      "step": 115
+    },
+    {
+      "epoch": 0.5286343612334802,
+      "grad_norm": 0.6979171633720398,
+      "learning_rate": 2.975581161430035e-05,
+      "loss": 0.9919,
+      "step": 120
+    },
+    {
+      "epoch": 0.5506607929515418,
+      "grad_norm": 0.7126419544219971,
+      "learning_rate": 2.971496850034492e-05,
+      "loss": 0.9232,
+      "step": 125
+    },
+    {
+      "epoch": 0.5726872246696035,
+      "grad_norm": 0.8123069405555725,
+      "learning_rate": 2.9671001078073453e-05,
+      "loss": 0.932,
+      "step": 130
+    },
+    {
+      "epoch": 0.5947136563876652,
+      "grad_norm": 0.8383358120918274,
+      "learning_rate": 2.9623918682727355e-05,
+      "loss": 0.9127,
+      "step": 135
+    },
+    {
+      "epoch": 0.6167400881057269,
+      "grad_norm": 0.7429929971694946,
+      "learning_rate": 2.957373131092464e-05,
+      "loss": 0.9187,
+      "step": 140
+    },
+    {
+      "epoch": 0.6387665198237885,
+      "grad_norm": 0.9482147693634033,
+      "learning_rate": 2.9520449618537465e-05,
+      "loss": 0.8848,
+      "step": 145
+    },
+    {
+      "epoch": 0.6607929515418502,
+      "grad_norm": 0.8874338865280151,
+      "learning_rate": 2.946408491842964e-05,
+      "loss": 0.8802,
+      "step": 150
+    },
+    {
+      "epoch": 0.6828193832599119,
+      "grad_norm": 0.7921388149261475,
+      "learning_rate": 2.940464917805466e-05,
+      "loss": 0.8105,
+      "step": 155
+    },
+    {
+      "epoch": 0.7048458149779736,
+      "grad_norm": 0.823284924030304,
+      "learning_rate": 2.9342155016914772e-05,
+      "loss": 0.8335,
+      "step": 160
+    },
+    {
+      "epoch": 0.7268722466960352,
+      "grad_norm": 0.8587915301322937,
+      "learning_rate": 2.927661570388155e-05,
+      "loss": 0.8236,
+      "step": 165
+    },
+    {
+      "epoch": 0.748898678414097,
+      "grad_norm": 0.9209868311882019,
+      "learning_rate": 2.920804515437865e-05,
+      "loss": 0.7313,
+      "step": 170
+    },
+    {
+      "epoch": 0.7709251101321586,
+      "grad_norm": 0.9900104999542236,
+      "learning_rate": 2.9136457927427254e-05,
+      "loss": 0.8445,
+      "step": 175
+    },
+    {
+      "epoch": 0.7929515418502202,
+      "grad_norm": 1.1005882024765015,
+      "learning_rate": 2.9061869222554863e-05,
+      "loss": 0.7258,
+      "step": 180
+    },
+    {
+      "epoch": 0.8149779735682819,
+      "grad_norm": 0.7985159754753113,
+      "learning_rate": 2.898429487656813e-05,
+      "loss": 0.7554,
+      "step": 185
+    },
+    {
+      "epoch": 0.8370044052863436,
+      "grad_norm": 1.0106185674667358,
+      "learning_rate": 2.8903751360190327e-05,
+      "loss": 0.7773,
+      "step": 190
+    },
+    {
+      "epoch": 0.8590308370044053,
+      "grad_norm": 1.0603998899459839,
+      "learning_rate": 2.8820255774564287e-05,
+      "loss": 0.8034,
+      "step": 195
+    },
+    {
+      "epoch": 0.8810572687224669,
+      "grad_norm": 0.9879089593887329,
+      "learning_rate": 2.8733825847621436e-05,
+      "loss": 0.71,
+      "step": 200
+    },
+    {
+      "epoch": 0.9030837004405287,
+      "grad_norm": 1.0008033514022827,
+      "learning_rate": 2.8644479930317776e-05,
+      "loss": 0.7837,
+      "step": 205
+    },
+    {
+      "epoch": 0.9251101321585903,
+      "grad_norm": 1.0469297170639038,
+      "learning_rate": 2.8552236992737572e-05,
+      "loss": 0.7139,
+      "step": 210
+    },
+    {
+      "epoch": 0.947136563876652,
+      "grad_norm": 0.9759002327919006,
+      "learning_rate": 2.8457116620065596e-05,
+      "loss": 0.7207,
+      "step": 215
+    },
+    {
+      "epoch": 0.9691629955947136,
+      "grad_norm": 0.907706618309021,
+      "learning_rate": 2.8359139008428758e-05,
+      "loss": 0.7134,
+      "step": 220
+    },
+    {
+      "epoch": 0.9911894273127754,
+      "grad_norm": 1.022333025932312,
+      "learning_rate": 2.8258324960608043e-05,
+      "loss": 0.669,
+      "step": 225
+    },
+    {
+      "epoch": 1.013215859030837,
+      "grad_norm": 1.021018385887146,
+      "learning_rate": 2.815469588162161e-05,
+      "loss": 0.6179,
+      "step": 230
+    },
+    {
+      "epoch": 1.0352422907488987,
+      "grad_norm": 1.1175833940505981,
+      "learning_rate": 2.8048273774180043e-05,
+      "loss": 0.5932,
+      "step": 235
+    },
+    {
+      "epoch": 1.0572687224669604,
+      "grad_norm": 0.9679650068283081,
+      "learning_rate": 2.7939081234014708e-05,
+      "loss": 0.5915,
+      "step": 240
+    },
+    {
+      "epoch": 1.079295154185022,
+      "grad_norm": 1.1451464891433716,
+      "learning_rate": 2.7827141445080196e-05,
+      "loss": 0.5768,
+      "step": 245
+    },
+    {
+      "epoch": 1.1013215859030836,
+      "grad_norm": 1.274778127670288,
+      "learning_rate": 2.7712478174631813e-05,
+      "loss": 0.5711,
+      "step": 250
+    },
+    {
+      "epoch": 1.1233480176211454,
+      "grad_norm": 1.4071186780929565,
+      "learning_rate": 2.759511576817934e-05,
+      "loss": 0.5991,
+      "step": 255
+    },
+    {
+      "epoch": 1.145374449339207,
+      "grad_norm": 1.2345290184020996,
+      "learning_rate": 2.747507914431791e-05,
+      "loss": 0.5184,
+      "step": 260
+    },
+    {
+      "epoch": 1.1674008810572687,
+      "grad_norm": 1.345145344734192,
+      "learning_rate": 2.7352393789437258e-05,
+      "loss": 0.5362,
+      "step": 265
+    },
+    {
+      "epoch": 1.1894273127753303,
+      "grad_norm": 1.1527823209762573,
+      "learning_rate": 2.7227085752310413e-05,
+      "loss": 0.543,
+      "step": 270
+    },
+    {
+      "epoch": 1.2114537444933922,
+      "grad_norm": 1.1420563459396362,
+      "learning_rate": 2.709918163856295e-05,
+      "loss": 0.5326,
+      "step": 275
+    },
+    {
+      "epoch": 1.2334801762114538,
+      "grad_norm": 1.0940356254577637,
+      "learning_rate": 2.696870860502408e-05,
+      "loss": 0.5188,
+      "step": 280
+    },
+    {
+      "epoch": 1.2555066079295154,
+      "grad_norm": 1.2769014835357666,
+      "learning_rate": 2.6835694353960623e-05,
+      "loss": 0.5029,
+      "step": 285
+    },
+    {
+      "epoch": 1.277533039647577,
+      "grad_norm": 1.3038066625595093,
+      "learning_rate": 2.6700167127195233e-05,
+      "loss": 0.5492,
+      "step": 290
+    },
+    {
+      "epoch": 1.2995594713656389,
+      "grad_norm": 1.0858526229858398,
+      "learning_rate": 2.6562155700110046e-05,
+      "loss": 0.5278,
+      "step": 295
+    },
+    {
+      "epoch": 1.3215859030837005,
+      "grad_norm": 1.1577568054199219,
+      "learning_rate": 2.6421689375537015e-05,
+      "loss": 0.5314,
+      "step": 300
+    },
+    {
+      "epoch": 1.3436123348017621,
+      "grad_norm": 1.1254470348358154,
+      "learning_rate": 2.6278797977536325e-05,
+      "loss": 0.466,
+      "step": 305
+    },
+    {
+      "epoch": 1.3656387665198237,
+      "grad_norm": 1.2429901361465454,
+      "learning_rate": 2.613351184506405e-05,
+      "loss": 0.5185,
+      "step": 310
+    },
+    {
+      "epoch": 1.3876651982378854,
+      "grad_norm": 1.3175548315048218,
+      "learning_rate": 2.598586182553056e-05,
+      "loss": 0.5077,
+      "step": 315
+    },
+    {
+      "epoch": 1.4096916299559472,
+      "grad_norm": 1.3169467449188232,
+      "learning_rate": 2.5835879268250934e-05,
+      "loss": 0.5124,
+      "step": 320
+    },
+    {
+      "epoch": 1.4317180616740088,
+      "grad_norm": 1.2650501728057861,
+      "learning_rate": 2.568359601778881e-05,
+      "loss": 0.4961,
+      "step": 325
+    },
+    {
+      "epoch": 1.4537444933920705,
+      "grad_norm": 1.1624271869659424,
+      "learning_rate": 2.5529044407195127e-05,
+      "loss": 0.4552,
+      "step": 330
+    },
+    {
+      "epoch": 1.475770925110132,
+      "grad_norm": 1.1693929433822632,
+      "learning_rate": 2.5372257251143056e-05,
+      "loss": 0.4668,
+      "step": 335
+    },
+    {
+      "epoch": 1.497797356828194,
+      "grad_norm": 1.105682134628296,
+      "learning_rate": 2.5213267838960772e-05,
+      "loss": 0.4485,
+      "step": 340
+    },
+    {
+      "epoch": 1.5198237885462555,
+      "grad_norm": 1.365013599395752,
+      "learning_rate": 2.5052109927563393e-05,
+      "loss": 0.4641,
+      "step": 345
+    },
+    {
+      "epoch": 1.5418502202643172,
+      "grad_norm": 1.2356834411621094,
+      "learning_rate": 2.4888817734285657e-05,
+      "loss": 0.4388,
+      "step": 350
+    },
+    {
+      "epoch": 1.5638766519823788,
+      "grad_norm": 1.8782477378845215,
+      "learning_rate": 2.472342592961683e-05,
+      "loss": 0.4556,
+      "step": 355
+    },
+    {
+      "epoch": 1.5859030837004404,
+      "grad_norm": 1.1085964441299438,
+      "learning_rate": 2.4555969629839393e-05,
+      "loss": 0.4284,
+      "step": 360
+    },
+    {
+      "epoch": 1.607929515418502,
+      "grad_norm": 1.3663532733917236,
+      "learning_rate": 2.4386484389573126e-05,
+      "loss": 0.4322,
+      "step": 365
+    },
+    {
+      "epoch": 1.6299559471365639,
+      "grad_norm": 1.307347297668457,
+      "learning_rate": 2.421500619422606e-05,
+      "loss": 0.4378,
+      "step": 370
+    },
+    {
+      "epoch": 1.6519823788546255,
+      "grad_norm": 1.0821038484573364,
+      "learning_rate": 2.4041571452353982e-05,
+      "loss": 0.4003,
+      "step": 375
+    },
+    {
+      "epoch": 1.6740088105726874,
+      "grad_norm": 1.056179404258728,
+      "learning_rate": 2.386621698793015e-05,
+      "loss": 0.4161,
+      "step": 380
+    },
+    {
+      "epoch": 1.696035242290749,
+      "grad_norm": 1.2139962911605835,
+      "learning_rate": 2.3688980032526707e-05,
+      "loss": 0.3905,
+      "step": 385
+    },
+    {
+      "epoch": 1.7180616740088106,
+      "grad_norm": 1.0942909717559814,
+      "learning_rate": 2.3509898217409645e-05,
+      "loss": 0.4268,
+      "step": 390
+    },
+    {
+      "epoch": 1.7400881057268722,
+      "grad_norm": 1.277398705482483,
+      "learning_rate": 2.3329009565548857e-05,
+      "loss": 0.4007,
+      "step": 395
+    },
+    {
+      "epoch": 1.7621145374449338,
+      "grad_norm": 1.1945730447769165,
+      "learning_rate": 2.3146352483545026e-05,
+      "loss": 0.3755,
+      "step": 400
+    },
+    {
+      "epoch": 1.7841409691629955,
+      "grad_norm": 1.1625131368637085,
+      "learning_rate": 2.2961965753475074e-05,
+      "loss": 0.4081,
+      "step": 405
+    },
+    {
+      "epoch": 1.8061674008810573,
+      "grad_norm": 1.1497328281402588,
+      "learning_rate": 2.277588852465788e-05,
+      "loss": 0.3711,
+      "step": 410
+    },
+    {
+      "epoch": 1.828193832599119,
+      "grad_norm": 1.1712104082107544,
+      "learning_rate": 2.2588160305342024e-05,
+      "loss": 0.4283,
+      "step": 415
+    },
+    {
+      "epoch": 1.8502202643171806,
+      "grad_norm": 1.3465914726257324,
+      "learning_rate": 2.2398820954317342e-05,
+      "loss": 0.3948,
+      "step": 420
+    },
+    {
+      "epoch": 1.8722466960352424,
+      "grad_norm": 1.3581944704055786,
+      "learning_rate": 2.220791067245201e-05,
+      "loss": 0.4051,
+      "step": 425
+    },
+    {
+      "epoch": 1.894273127753304,
+      "grad_norm": 1.656079649925232,
+      "learning_rate": 2.201546999415704e-05,
+      "loss": 0.3842,
+      "step": 430
+    },
+    {
+      "epoch": 1.9162995594713657,
+      "grad_norm": 1.3536030054092407,
+      "learning_rate": 2.182153977877994e-05,
+      "loss": 0.3618,
+      "step": 435
+    },
+    {
+      "epoch": 1.9383259911894273,
+      "grad_norm": 1.2806636095046997,
+      "learning_rate": 2.162616120192939e-05,
+      "loss": 0.362,
+      "step": 440
+    },
+    {
+      "epoch": 1.960352422907489,
+      "grad_norm": 1.1593424081802368,
+      "learning_rate": 2.142937574673275e-05,
+      "loss": 0.3802,
+      "step": 445
+    },
+    {
+      "epoch": 1.9823788546255505,
+      "grad_norm": 1.5513302087783813,
+      "learning_rate": 2.12312251950283e-05,
+      "loss": 0.36,
+      "step": 450
+    },
+    {
+      "epoch": 2.004405286343612,
+      "grad_norm": 1.4302456378936768,
+      "learning_rate": 2.103175161849403e-05,
+      "loss": 0.3098,
+      "step": 455
+    },
+    {
+      "epoch": 2.026431718061674,
+      "grad_norm": 1.7442538738250732,
+      "learning_rate": 2.0830997369714936e-05,
+      "loss": 0.2725,
+      "step": 460
+    },
+    {
+      "epoch": 2.048458149779736,
+      "grad_norm": 1.14962899684906,
+      "learning_rate": 2.0629005073190615e-05,
+      "loss": 0.2694,
+      "step": 465
+    },
+    {
+      "epoch": 2.0704845814977975,
+      "grad_norm": 1.1875709295272827,
+      "learning_rate": 2.0425817616285164e-05,
+      "loss": 0.248,
+      "step": 470
+    },
+    {
+      "epoch": 2.092511013215859,
+      "grad_norm": 1.1875449419021606,
+      "learning_rate": 2.0221478140121268e-05,
+      "loss": 0.2886,
+      "step": 475
+    },
+    {
+      "epoch": 2.1145374449339207,
+      "grad_norm": 1.168346643447876,
+      "learning_rate": 2.001603003042039e-05,
+      "loss": 0.2601,
+      "step": 480
+    },
+    {
+      "epoch": 2.1365638766519823,
+      "grad_norm": 1.2313181161880493,
+      "learning_rate": 1.9809516908291032e-05,
+      "loss": 0.3034,
+      "step": 485
+    },
+    {
+      "epoch": 2.158590308370044,
+      "grad_norm": 1.40427827835083,
+      "learning_rate": 1.9601982620967047e-05,
+      "loss": 0.2919,
+      "step": 490
+    },
+    {
+      "epoch": 2.1806167400881056,
+      "grad_norm": 1.6281092166900635,
+      "learning_rate": 1.9393471232497883e-05,
+      "loss": 0.282,
+      "step": 495
+    },
+    {
+      "epoch": 2.202643171806167,
+      "grad_norm": 1.186156988143921,
+      "learning_rate": 1.9184027014392857e-05,
+      "loss": 0.2727,
+      "step": 500
+    },
+    {
+      "epoch": 2.2246696035242293,
+      "grad_norm": 1.2361948490142822,
+      "learning_rate": 1.8973694436221322e-05,
+      "loss": 0.269,
+      "step": 505
+    },
+    {
+      "epoch": 2.246696035242291,
+      "grad_norm": 1.0598633289337158,
+      "learning_rate": 1.876251815617084e-05,
+      "loss": 0.2589,
+      "step": 510
+    },
+    {
+      "epoch": 2.2687224669603525,
+      "grad_norm": 1.190380573272705,
+      "learning_rate": 1.8550543011565245e-05,
+      "loss": 0.2317,
+      "step": 515
+    },
+    {
+      "epoch": 2.290748898678414,
+      "grad_norm": 1.2344335317611694,
+      "learning_rate": 1.8337814009344716e-05,
+      "loss": 0.2843,
+      "step": 520
+    },
+    {
+      "epoch": 2.3127753303964758,
+      "grad_norm": 1.1089266538619995,
+      "learning_rate": 1.8124376316509832e-05,
+      "loss": 0.2688,
+      "step": 525
+    },
+    {
+      "epoch": 2.3348017621145374,
+      "grad_norm": 1.2801549434661865,
+      "learning_rate": 1.7910275250531612e-05,
+      "loss": 0.2554,
+      "step": 530
+    },
+    {
+      "epoch": 2.356828193832599,
+      "grad_norm": 1.2876077890396118,
+      "learning_rate": 1.769555626972967e-05,
+      "loss": 0.2512,
+      "step": 535
+    },
+    {
+      "epoch": 2.3788546255506606,
+      "grad_norm": 1.1712980270385742,
+      "learning_rate": 1.748026496362035e-05,
+      "loss": 0.2688,
+      "step": 540
+    },
+    {
+      "epoch": 2.4008810572687223,
+      "grad_norm": 1.2399232387542725,
+      "learning_rate": 1.7264447043237165e-05,
+      "loss": 0.2333,
+      "step": 545
+    },
+    {
+      "epoch": 2.4229074889867843,
+      "grad_norm": 1.134351134300232,
+      "learning_rate": 1.704814833142529e-05,
+      "loss": 0.2152,
+      "step": 550
+    },
+    {
+      "epoch": 2.444933920704846,
+      "grad_norm": 1.1559290885925293,
+      "learning_rate": 1.68314147531124e-05,
+      "loss": 0.2972,
+      "step": 555
+    },
+    {
+      "epoch": 2.4669603524229076,
+      "grad_norm": 1.2436527013778687,
+      "learning_rate": 1.661429232555779e-05,
+      "loss": 0.2352,
+      "step": 560
+    },
+    {
+      "epoch": 2.488986784140969,
+      "grad_norm": 1.2720732688903809,
+      "learning_rate": 1.639682714858193e-05,
+      "loss": 0.2527,
+      "step": 565
+    },
+    {
+      "epoch": 2.511013215859031,
+      "grad_norm": 1.3503328561782837,
+      "learning_rate": 1.6179065394778445e-05,
+      "loss": 0.2403,
+      "step": 570
+    },
+    {
+      "epoch": 2.5330396475770924,
+      "grad_norm": 1.28740656375885,
+      "learning_rate": 1.5961053299710696e-05,
+      "loss": 0.2761,
+      "step": 575
+    },
+    {
+      "epoch": 2.555066079295154,
+      "grad_norm": 1.5024586915969849,
+      "learning_rate": 1.5742837152094932e-05,
+      "loss": 0.2269,
+      "step": 580
+    },
+    {
+      "epoch": 2.577092511013216,
+      "grad_norm": 1.1579198837280273,
+      "learning_rate": 1.552446328397221e-05,
+      "loss": 0.2502,
+      "step": 585
+    },
+    {
+      "epoch": 2.5991189427312777,
+      "grad_norm": 1.0729395151138306,
+      "learning_rate": 1.5305978060871083e-05,
+      "loss": 0.2265,
+      "step": 590
+    },
+    {
+      "epoch": 2.6211453744493394,
+      "grad_norm": 1.1695958375930786,
+      "learning_rate": 1.5087427871963195e-05,
+      "loss": 0.2492,
+      "step": 595
+    },
+    {
+      "epoch": 2.643171806167401,
+      "grad_norm": 1.2385801076889038,
+      "learning_rate": 1.4868859120213842e-05,
+      "loss": 0.1856,
+      "step": 600
+    },
+    {
+      "epoch": 2.6651982378854626,
+      "grad_norm": 1.270134687423706,
+      "learning_rate": 1.4650318212529613e-05,
+      "loss": 0.2223,
+      "step": 605
+    },
+    {
+      "epoch": 2.6872246696035242,
+      "grad_norm": 1.183335542678833,
+      "learning_rate": 1.443185154990521e-05,
+      "loss": 0.2215,
+      "step": 610
+    },
+    {
+      "epoch": 2.709251101321586,
+      "grad_norm": 1.1907864809036255,
+      "learning_rate": 1.4213505517571477e-05,
+      "loss": 0.1935,
+      "step": 615
+    },
+    {
+      "epoch": 2.7312775330396475,
+      "grad_norm": 1.272538661956787,
+      "learning_rate": 1.3995326475146819e-05,
+      "loss": 0.2444,
+      "step": 620
+    },
+    {
+      "epoch": 2.753303964757709,
+      "grad_norm": 1.1249198913574219,
+      "learning_rate": 1.3777360746794052e-05,
+      "loss": 0.1985,
+      "step": 625
+    },
+    {
+      "epoch": 2.7753303964757707,
+      "grad_norm": 1.0267751216888428,
+      "learning_rate": 1.3559654611384774e-05,
+      "loss": 0.1852,
+      "step": 630
+    },
+    {
+      "epoch": 2.7973568281938324,
+      "grad_norm": 1.3851100206375122,
+      "learning_rate": 1.334225429267334e-05,
+      "loss": 0.2149,
+      "step": 635
+    },
+    {
+      "epoch": 2.8193832599118944,
+      "grad_norm": 1.3569889068603516,
+      "learning_rate": 1.3125205949482551e-05,
+      "loss": 0.1972,
+      "step": 640
+    },
+    {
+      "epoch": 2.841409691629956,
+      "grad_norm": 1.1331690549850464,
+      "learning_rate": 1.2908555665903151e-05,
+      "loss": 0.1995,
+      "step": 645
+    },
+    {
+      "epoch": 2.8634361233480177,
+      "grad_norm": 1.1397310495376587,
+      "learning_rate": 1.269234944150916e-05,
+      "loss": 0.198,
+      "step": 650
+    },
+    {
+      "epoch": 2.8854625550660793,
+      "grad_norm": 1.380522608757019,
+      "learning_rate": 1.247663318159115e-05,
+      "loss": 0.1623,
+      "step": 655
+    },
+    {
+      "epoch": 2.907488986784141,
+      "grad_norm": 1.1323232650756836,
+      "learning_rate": 1.2261452687409576e-05,
+      "loss": 0.1748,
+      "step": 660
+    },
+    {
+      "epoch": 2.9295154185022025,
+      "grad_norm": 1.1639511585235596,
+      "learning_rate": 1.2046853646470144e-05,
+      "loss": 0.1747,
+      "step": 665
+    },
+    {
+      "epoch": 2.951541850220264,
+      "grad_norm": 1.2315850257873535,
+      "learning_rate": 1.1832881622823376e-05,
+      "loss": 0.1897,
+      "step": 670
+    },
+    {
+      "epoch": 2.9735682819383262,
+      "grad_norm": 1.0831575393676758,
+      "learning_rate": 1.1619582047390377e-05,
+      "loss": 0.1836,
+      "step": 675
+    },
+    {
+      "epoch": 2.995594713656388,
+      "grad_norm": 1.153684377670288,
+      "learning_rate": 1.140700020831684e-05,
+      "loss": 0.174,
+      "step": 680
+    },
+    {
+      "epoch": 3.0176211453744495,
+      "grad_norm": 1.1012849807739258,
+      "learning_rate": 1.1195181241357392e-05,
+      "loss": 0.1605,
+      "step": 685
+    },
+    {
+      "epoch": 3.039647577092511,
+      "grad_norm": 1.1384843587875366,
+      "learning_rate": 1.0984170120292299e-05,
+      "loss": 0.1424,
+      "step": 690
+    },
+    {
+      "epoch": 3.0616740088105727,
+      "grad_norm": 1.1249345541000366,
+      "learning_rate": 1.0774011647378554e-05,
+      "loss": 0.1541,
+      "step": 695
+    },
+    {
+      "epoch": 3.0837004405286343,
+      "grad_norm": 1.0345319509506226,
+      "learning_rate": 1.0564750443837374e-05,
+      "loss": 0.1347,
+      "step": 700
+    },
+    {
+      "epoch": 3.105726872246696,
+      "grad_norm": 1.0952510833740234,
+      "learning_rate": 1.0356430940380132e-05,
+      "loss": 0.1327,
+      "step": 705
+    },
+    {
+      "epoch": 3.1277533039647576,
+      "grad_norm": 1.3496341705322266,
+      "learning_rate": 1.0149097367774775e-05,
+      "loss": 0.1296,
+      "step": 710
+    },
+    {
+      "epoch": 3.149779735682819,
+      "grad_norm": 1.2481356859207153,
+      "learning_rate": 9.942793747454654e-06,
+      "loss": 0.1312,
+      "step": 715
+    },
+    {
+      "epoch": 3.171806167400881,
+      "grad_norm": 1.3235247135162354,
+      "learning_rate": 9.737563882171816e-06,
+      "loss": 0.1249,
+      "step": 720
+    },
+    {
+      "epoch": 3.193832599118943,
+      "grad_norm": 1.1433985233306885,
+      "learning_rate": 9.533451346696733e-06,
+      "loss": 0.1247,
+      "step": 725
+    },
+    {
+      "epoch": 3.2158590308370045,
+      "grad_norm": 1.016939401626587,
+      "learning_rate": 9.330499478566404e-06,
+      "loss": 0.1371,
+      "step": 730
+    },
+    {
+      "epoch": 3.237885462555066,
+      "grad_norm": 1.2426491975784302,
+      "learning_rate": 9.128751368882883e-06,
+      "loss": 0.1473,
+      "step": 735
+    },
+    {
+      "epoch": 3.2599118942731278,
+      "grad_norm": 1.182470440864563,
+      "learning_rate": 8.928249853164093e-06,
+      "loss": 0.141,
+      "step": 740
+    },
+    {
+      "epoch": 3.2819383259911894,
+      "grad_norm": 0.9752205014228821,
+      "learning_rate": 8.729037502248901e-06,
+      "loss": 0.1157,
+      "step": 745
+    },
+    {
+      "epoch": 3.303964757709251,
+      "grad_norm": 1.0723822116851807,
+      "learning_rate": 8.531156613258386e-06,
+      "loss": 0.1342,
+      "step": 750
+    },
+    {
+      "epoch": 3.3259911894273126,
+      "grad_norm": 1.183722734451294,
+      "learning_rate": 8.334649200615302e-06,
+      "loss": 0.1274,
+      "step": 755
+    },
+    {
+      "epoch": 3.3480176211453743,
+      "grad_norm": 1.0636101961135864,
+      "learning_rate": 8.139556987123429e-06,
+      "loss": 0.1253,
+      "step": 760
+    },
+    {
+      "epoch": 3.3700440528634363,
+      "grad_norm": 1.1327389478683472,
+      "learning_rate": 7.945921395108949e-06,
+      "loss": 0.1346,
+      "step": 765
+    },
+    {
+      "epoch": 3.392070484581498,
+      "grad_norm": 1.0662906169891357,
+      "learning_rate": 7.753783537625581e-06,
+      "loss": 0.1543,
+      "step": 770
+    },
+    {
+      "epoch": 3.4140969162995596,
+      "grad_norm": 1.048923373222351,
+      "learning_rate": 7.56318420972538e-06,
+      "loss": 0.1388,
+      "step": 775
+    },
+    {
+      "epoch": 3.436123348017621,
+      "grad_norm": 1.3158599138259888,
+      "learning_rate": 7.37416387979706e-06,
+      "loss": 0.1416,
+      "step": 780
+    },
+    {
+      "epoch": 3.458149779735683,
+      "grad_norm": 1.0559903383255005,
+      "learning_rate": 7.186762680973674e-06,
+      "loss": 0.1256,
+      "step": 785
+    },
+    {
+      "epoch": 3.4801762114537445,
+      "grad_norm": 1.036526083946228,
+      "learning_rate": 7.001020402611475e-06,
+      "loss": 0.1368,
+      "step": 790
+    },
+    {
+      "epoch": 3.502202643171806,
+      "grad_norm": 1.054312825202942,
+      "learning_rate": 6.816976481841771e-06,
+      "loss": 0.1294,
+      "step": 795
+    },
+    {
+      "epoch": 3.5242290748898677,
+      "grad_norm": 1.1312750577926636,
+      "learning_rate": 6.634669995197561e-06,
+      "loss": 0.1129,
+      "step": 800
+    },
+    {
+      "epoch": 3.5462555066079293,
+      "grad_norm": 0.9955660700798035,
+      "learning_rate": 6.454139650316735e-06,
+      "loss": 0.1372,
+      "step": 805
+    },
+    {
+      "epoch": 3.568281938325991,
+      "grad_norm": 1.1197795867919922,
+      "learning_rate": 6.275423777723599e-06,
+      "loss": 0.1327,
+      "step": 810
+    },
+    {
+      "epoch": 3.590308370044053,
+      "grad_norm": 1.1895395517349243,
+      "learning_rate": 6.0985603226904794e-06,
+      "loss": 0.1233,
+      "step": 815
+    },
+    {
+      "epoch": 3.6123348017621146,
+      "grad_norm": 1.023464322090149,
+      "learning_rate": 5.9235868371811e-06,
+      "loss": 0.1105,
+      "step": 820
+    },
+    {
+      "epoch": 3.6343612334801763,
+      "grad_norm": 1.1077297925949097,
+      "learning_rate": 5.750540471877487e-06,
+      "loss": 0.1223,
+      "step": 825
+    },
+    {
+      "epoch": 3.656387665198238,
+      "grad_norm": 0.9085190892219543,
+      "learning_rate": 5.579457968292091e-06,
+      "loss": 0.1218,
+      "step": 830
+    },
+    {
+      "epoch": 3.6784140969162995,
+      "grad_norm": 1.0670191049575806,
+      "learning_rate": 5.410375650966732e-06,
+      "loss": 0.135,
+      "step": 835
+    },
+    {
+      "epoch": 3.700440528634361,
+      "grad_norm": 1.2323230504989624,
+      "learning_rate": 5.243329419760109e-06,
+      "loss": 0.1383,
+      "step": 840
+    },
+    {
+      "epoch": 3.7224669603524227,
+      "grad_norm": 0.9415250420570374,
+      "learning_rate": 5.078354742225505e-06,
+      "loss": 0.1049,
+      "step": 845
+    },
+    {
+      "epoch": 3.744493392070485,
+      "grad_norm": 1.076449990272522,
+      "learning_rate": 4.915486646080247e-06,
+      "loss": 0.1015,
+      "step": 850
+    },
+    {
+      "epoch": 3.7665198237885464,
+      "grad_norm": 1.5556225776672363,
+      "learning_rate": 4.754759711768541e-06,
+      "loss": 0.1216,
+      "step": 855
+    },
+    {
+      "epoch": 3.788546255506608,
+      "grad_norm": 1.0354886054992676,
+      "learning_rate": 4.596208065119285e-06,
+      "loss": 0.1319,
+      "step": 860
+    },
+    {
+      "epoch": 3.8105726872246697,
+      "grad_norm": 1.2755696773529053,
+      "learning_rate": 4.439865370100445e-06,
+      "loss": 0.1176,
+      "step": 865
+    },
+    {
+      "epoch": 3.8325991189427313,
+      "grad_norm": 0.968616247177124,
+      "learning_rate": 4.285764821671446e-06,
+      "loss": 0.1177,
+      "step": 870
+    },
+    {
+      "epoch": 3.854625550660793,
+      "grad_norm": 0.9090492129325867,
+      "learning_rate": 4.133939138735143e-06,
+      "loss": 0.1205,
+      "step": 875
+    },
+    {
+      "epoch": 3.8766519823788546,
+      "grad_norm": 0.9453270435333252,
+      "learning_rate": 3.9844205571909046e-06,
+      "loss": 0.1046,
+      "step": 880
+    },
+    {
+      "epoch": 3.898678414096916,
+      "grad_norm": 0.8504048585891724,
+      "learning_rate": 3.837240823090225e-06,
+      "loss": 0.1073,
+      "step": 885
+    },
+    {
+      "epoch": 3.920704845814978,
+      "grad_norm": 1.0514414310455322,
+      "learning_rate": 3.6924311858963395e-06,
+      "loss": 0.1268,
+      "step": 890
+    },
+    {
+      "epoch": 3.9427312775330394,
+      "grad_norm": 0.9555254578590393,
+      "learning_rate": 3.550022391849292e-06,
+      "loss": 0.1038,
+      "step": 895
+    },
+    {
+      "epoch": 3.964757709251101,
+      "grad_norm": 0.9492741823196411,
+      "learning_rate": 3.4100446774378336e-06,
+      "loss": 0.1113,
+      "step": 900
+    },
+    {
+      "epoch": 3.986784140969163,
+      "grad_norm": 0.8036649227142334,
+      "learning_rate": 3.272527762979553e-06,
+      "loss": 0.088,
+      "step": 905
+    },
+    {
+      "epoch": 4.008810572687224,
+      "grad_norm": 0.8136736750602722,
+      "learning_rate": 3.137500846310619e-06,
+      "loss": 0.1109,
+      "step": 910
+    },
+    {
+      "epoch": 4.030837004405286,
+      "grad_norm": 0.6706807017326355,
+      "learning_rate": 3.00499259658643e-06,
+      "loss": 0.0855,
+      "step": 915
+    },
+    {
+      "epoch": 4.052863436123348,
+      "grad_norm": 0.8383581638336182,
+      "learning_rate": 2.875031148194532e-06,
+      "loss": 0.0749,
+      "step": 920
+    },
+    {
+      "epoch": 4.07488986784141,
+      "grad_norm": 0.9318509101867676,
+      "learning_rate": 2.7476440947810828e-06,
+      "loss": 0.1103,
+      "step": 925
+    },
+    {
+      "epoch": 4.096916299559472,
+      "grad_norm": 0.7745897769927979,
+      "learning_rate": 2.6228584833921087e-06,
+      "loss": 0.0867,
+      "step": 930
+    },
+    {
+      "epoch": 4.118942731277533,
+      "grad_norm": 0.8513121604919434,
+      "learning_rate": 2.5007008087308366e-06,
+      "loss": 0.0813,
+      "step": 935
+    },
+    {
+      "epoch": 4.140969162995595,
+      "grad_norm": 0.8945024609565735,
+      "learning_rate": 2.38119700753228e-06,
+      "loss": 0.0869,
+      "step": 940
+    },
+    {
+      "epoch": 4.1629955947136565,
+      "grad_norm": 0.8105630278587341,
+      "learning_rate": 2.264372453056332e-06,
+      "loss": 0.0809,
+      "step": 945
+    },
+    {
+      "epoch": 4.185022026431718,
+      "grad_norm": 1.0123265981674194,
+      "learning_rate": 2.1502519497004243e-06,
+      "loss": 0.089,
+      "step": 950
+    },
+    {
+      "epoch": 4.20704845814978,
+      "grad_norm": 0.7849066257476807,
+      "learning_rate": 2.038859727733057e-06,
+      "loss": 0.0926,
+      "step": 955
+    },
+    {
+      "epoch": 4.229074889867841,
+      "grad_norm": 0.8726152181625366,
+      "learning_rate": 1.930219438149159e-06,
+      "loss": 0.0993,
+      "step": 960
+    },
+    {
+      "epoch": 4.251101321585903,
+      "grad_norm": 0.9060913324356079,
+      "learning_rate": 1.8243541476484876e-06,
+      "loss": 0.0856,
+      "step": 965
+    },
+    {
+      "epoch": 4.273127753303965,
+      "grad_norm": 1.004108190536499,
+      "learning_rate": 1.7212863337380353e-06,
+      "loss": 0.109,
+      "step": 970
+    },
+    {
+      "epoch": 4.295154185022026,
+      "grad_norm": 0.8456615209579468,
+      "learning_rate": 1.621037879959592e-06,
+      "loss": 0.076,
+      "step": 975
+    },
+    {
+      "epoch": 4.317180616740088,
+      "grad_norm": 0.7427312135696411,
+      "learning_rate": 1.5236300712433899e-06,
+      "loss": 0.0828,
+      "step": 980
+    },
+    {
+      "epoch": 4.3392070484581495,
+      "grad_norm": 0.7335637807846069,
+      "learning_rate": 1.4290835893888365e-06,
+      "loss": 0.0856,
+      "step": 985
+    },
+    {
+      "epoch": 4.361233480176211,
+      "grad_norm": 0.7588874697685242,
+      "learning_rate": 1.337418508673321e-06,
+      "loss": 0.0859,
+      "step": 990
+    },
+    {
+      "epoch": 4.383259911894273,
+      "grad_norm": 0.7946503758430481,
+      "learning_rate": 1.2486542915900134e-06,
+      "loss": 0.0902,
+      "step": 995
+    },
+    {
+      "epoch": 4.405286343612334,
+      "grad_norm": 0.7260093092918396,
+      "learning_rate": 1.16280978471556e-06,
+      "loss": 0.0764,
+      "step": 1000
+    },
+    {
+      "epoch": 4.427312775330396,
+      "grad_norm": 0.893501877784729,
+      "learning_rate": 1.079903214708528e-06,
+      "loss": 0.0866,
+      "step": 1005
+    },
+    {
+      "epoch": 4.4493392070484585,
+      "grad_norm": 0.7707322835922241,
+      "learning_rate": 9.999521844394989e-07,
+      "loss": 0.0778,
+      "step": 1010
+    },
+    {
+      "epoch": 4.47136563876652,
+      "grad_norm": 0.8184295296669006,
+      "learning_rate": 9.229736692535934e-07,
+      "loss": 0.076,
+      "step": 1015
+    },
+    {
+      "epoch": 4.493392070484582,
+      "grad_norm": 0.8709239959716797,
+      "learning_rate": 8.489840133662397e-07,
+      "loss": 0.093,
+      "step": 1020
+    },
+    {
+      "epoch": 4.515418502202643,
+      "grad_norm": 0.8765608668327332,
+      "learning_rate": 7.779989263929327e-07,
+      "loss": 0.0839,
+      "step": 1025
+    },
+    {
+      "epoch": 4.537444933920705,
+      "grad_norm": 0.7354438304901123,
+      "learning_rate": 7.100334800137465e-07,
+      "loss": 0.0748,
+      "step": 1030
+    },
+    {
+      "epoch": 4.559471365638767,
+      "grad_norm": 0.809737503528595,
+      "learning_rate": 6.451021047732869e-07,
+      "loss": 0.0733,
+      "step": 1035
+    },
+    {
+      "epoch": 4.581497797356828,
+      "grad_norm": 0.9064865708351135,
+      "learning_rate": 5.832185870167761e-07,
+      "loss": 0.1069,
+      "step": 1040
+    },
+    {
+      "epoch": 4.60352422907489,
+      "grad_norm": 0.8218539953231812,
+      "learning_rate": 5.243960659629021e-07,
+      "loss": 0.0871,
+      "step": 1045
+    },
+    {
+      "epoch": 4.6255506607929515,
+      "grad_norm": 0.7513611912727356,
+      "learning_rate": 4.686470309140839e-07,
+      "loss": 0.0798,
+      "step": 1050
+    },
+    {
+      "epoch": 4.647577092511013,
+      "grad_norm": 0.7547969818115234,
+      "learning_rate": 4.1598331860472405e-07,
+      "loss": 0.081,
+      "step": 1055
+    },
+    {
+      "epoch": 4.669603524229075,
+      "grad_norm": 0.7942334413528442,
+      "learning_rate": 3.664161106879993e-07,
+      "loss": 0.093,
+      "step": 1060
+    },
+    {
+      "epoch": 4.691629955947136,
+      "grad_norm": 0.7672529220581055,
+      "learning_rate": 3.199559313617612e-07,
+      "loss": 0.0736,
+      "step": 1065
+    },
+    {
+      "epoch": 4.713656387665198,
+      "grad_norm": 0.7307041883468628,
+      "learning_rate": 2.766126451340184e-07,
+      "loss": 0.0721,
+      "step": 1070
+    },
+    {
+      "epoch": 4.73568281938326,
+      "grad_norm": 0.8792541027069092,
+      "learning_rate": 2.363954547284941e-07,
+      "loss": 0.0893,
+      "step": 1075
+    },
+    {
+      "epoch": 4.757709251101321,
+      "grad_norm": 0.7935943007469177,
+      "learning_rate": 1.9931289913066697e-07,
+      "loss": 0.0654,
+      "step": 1080
+    },
+    {
+      "epoch": 4.779735682819383,
+      "grad_norm": 0.843226969242096,
+      "learning_rate": 1.6537285177477823e-07,
+      "loss": 0.0918,
+      "step": 1085
+    },
+    {
+      "epoch": 4.8017621145374445,
+      "grad_norm": 0.7649157047271729,
+      "learning_rate": 1.3458251887212226e-07,
+      "loss": 0.0878,
+      "step": 1090
+    },
+    {
+      "epoch": 4.823788546255507,
+      "grad_norm": 0.9087342619895935,
+      "learning_rate": 1.0694843788101372e-07,
+      "loss": 0.0728,
+      "step": 1095
+    },
+    {
+      "epoch": 4.845814977973569,
+      "grad_norm": 0.8982771635055542,
+      "learning_rate": 8.247647611873632e-08,
+      "loss": 0.0783,
+      "step": 1100
+    },
+    {
+      "epoch": 4.86784140969163,
+      "grad_norm": 0.805414080619812,
+      "learning_rate": 6.117182951578925e-08,
+      "loss": 0.085,
+      "step": 1105
+    },
+    {
+      "epoch": 4.889867841409692,
+      "grad_norm": 0.8038418292999268,
+      "learning_rate": 4.303902151268135e-08,
+      "loss": 0.0794,
+      "step": 1110
+    },
+    {
+      "epoch": 4.9118942731277535,
+      "grad_norm": 0.6947316527366638,
+      "learning_rate": 2.8081902099496015e-08,
+      "loss": 0.0878,
+      "step": 1115
+    },
+    {
+      "epoch": 4.933920704845815,
+      "grad_norm": 0.6632384657859802,
+      "learning_rate": 1.6303646998465095e-08,
+      "loss": 0.0809,
+      "step": 1120
+    },
+    {
+      "epoch": 4.955947136563877,
+      "grad_norm": 0.7563242316246033,
+      "learning_rate": 7.706756989681596e-09,
+      "loss": 0.0897,
+      "step": 1125
+    },
+    {
+      "epoch": 4.977973568281938,
+      "grad_norm": 0.7428744435310364,
+      "learning_rate": 2.2930573801444032e-09,
+      "loss": 0.0843,
+      "step": 1130
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.7959029078483582,
+      "learning_rate": 6.369761619495762e-11,
+      "loss": 0.0764,
+      "step": 1135
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1135,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 2000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.6419735946552934e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

14_128_e5_3e-5/checkpoint-1135/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28b08ed8a6d7b069ef0067d87b8e8cc95aa65498f52b2ba0437d8dc8b44282b7
+size 7736

14_128_e5_3e-5/checkpoint-1135/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)

14_128_e5_3e-5/checkpoint-227/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

14_128_e5_3e-5/checkpoint-227/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": [
+    "v_proj",
+    "down_proj",
+    "gate_proj",
+    "o_proj",
+    "k_proj",
+    "q_proj",
+    "up_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

14_128_e5_3e-5/checkpoint-227/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8a22c1e0cd5d6506f1ccdc80e1783f6d949fdd2db7bf2819dcc1a061ef33be91
+size 791751704

14_128_e5_3e-5/checkpoint-227/latest

@@ -0,0 +1 @@
+global_step227

14_128_e5_3e-5/checkpoint-227/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a9e1138e32e14c7f507ac596939a2707fd27f91d203f766813a241eca8450298
+size 15920

14_128_e5_3e-5/checkpoint-227/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7c2d934212e10e9f1d41f99cee70e41c57f7512ff6458ce54ffd32bdbc81ab80
+size 15920

14_128_e5_3e-5/checkpoint-227/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9bf9aaac40711cc553c30c1dd835cac81045d22c66103f128618edbd8998ba93
+size 15920

14_128_e5_3e-5/checkpoint-227/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:826b0306f95e0856a90ee3887667a72618bd481867e7836f237ce75e0b1e0872
+size 15920

14_128_e5_3e-5/checkpoint-227/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d575793975cad01052b1a0131f4802af9b7a99e27dca49f0ddb5431be51fce41
+size 15920

14_128_e5_3e-5/checkpoint-227/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:68b52e08a8de8a1bb096455259638f52466860b45c214ef17e6ac2c7550e55e4
+size 15920

14_128_e5_3e-5/checkpoint-227/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:37cf1f8a118c79b7b9df12b51bf4841cdb46375a782e130095b1340b926cd710
+size 15920

14_128_e5_3e-5/checkpoint-227/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a610194b1bb039c248be8bdbc131db4c8e672335389bbb5539e9b77eb8375f65
+size 15920

14_128_e5_3e-5/checkpoint-227/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8841fc08f8c5cb2ed95c1807e6abfa5355d1dc51d467d6148b0e92f474b95d80
+size 1064

14_128_e5_3e-5/checkpoint-227/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
+  }
+}

14_128_e5_3e-5/checkpoint-227/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
+}

14_128_e5_3e-5/checkpoint-227/trainer_state.json

@@ -0,0 +1,349 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 227,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.022026431718061675,
+      "grad_norm": 1.2465029954910278,
+      "learning_rate": 2.1052631578947366e-06,
+      "loss": 1.3594,
+      "step": 5
+    },
+    {
+      "epoch": 0.04405286343612335,
+      "grad_norm": 0.7338076233863831,
+      "learning_rate": 4.736842105263158e-06,
+      "loss": 1.3607,
+      "step": 10
+    },
+    {
+      "epoch": 0.06607929515418502,
+      "grad_norm": 0.5697624087333679,
+      "learning_rate": 7.3684210526315784e-06,
+      "loss": 1.3328,
+      "step": 15
+    },
+    {
+      "epoch": 0.0881057268722467,
+      "grad_norm": 0.661296010017395,
+      "learning_rate": 9.999999999999999e-06,
+      "loss": 1.3393,
+      "step": 20
+    },
+    {
+      "epoch": 0.11013215859030837,
+      "grad_norm": 0.6124210357666016,
+      "learning_rate": 1.263157894736842e-05,
+      "loss": 1.2621,
+      "step": 25
+    },
+    {
+      "epoch": 0.13215859030837004,
+      "grad_norm": 0.5665815472602844,
+      "learning_rate": 1.5263157894736842e-05,
+      "loss": 1.2624,
+      "step": 30
+    },
+    {
+      "epoch": 0.15418502202643172,
+      "grad_norm": 0.5574439167976379,
+      "learning_rate": 1.7894736842105264e-05,
+      "loss": 1.2487,
+      "step": 35
+    },
+    {
+      "epoch": 0.1762114537444934,
+      "grad_norm": 0.4665200412273407,
+      "learning_rate": 2.0526315789473685e-05,
+      "loss": 1.2335,
+      "step": 40
+    },
+    {
+      "epoch": 0.19823788546255505,
+      "grad_norm": 0.4646995961666107,
+      "learning_rate": 2.3157894736842103e-05,
+      "loss": 1.1568,
+      "step": 45
+    },
+    {
+      "epoch": 0.22026431718061673,
+      "grad_norm": 0.5178118348121643,
+      "learning_rate": 2.578947368421053e-05,
+      "loss": 1.214,
+      "step": 50
+    },
+    {
+      "epoch": 0.2422907488986784,
+      "grad_norm": 0.41648760437965393,
+      "learning_rate": 2.8421052631578946e-05,
+      "loss": 1.151,
+      "step": 55
+    },
+    {
+      "epoch": 0.2643171806167401,
+      "grad_norm": 0.5988641977310181,
+      "learning_rate": 2.9999745210076202e-05,
+      "loss": 1.1945,
+      "step": 60
+    },
+    {
+      "epoch": 0.28634361233480177,
+      "grad_norm": 0.4767996370792389,
+      "learning_rate": 2.9996878922838097e-05,
+      "loss": 1.0975,
+      "step": 65
+    },
+    {
+      "epoch": 0.30837004405286345,
+      "grad_norm": 0.547401487827301,
+      "learning_rate": 2.9990828471561044e-05,
+      "loss": 1.1063,
+      "step": 70
+    },
+    {
+      "epoch": 0.3303964757709251,
+      "grad_norm": 0.5365089178085327,
+      "learning_rate": 2.998159514088762e-05,
+      "loss": 1.1153,
+      "step": 75
+    },
+    {
+      "epoch": 0.3524229074889868,
+      "grad_norm": 0.5236433148384094,
+      "learning_rate": 2.9969180891255046e-05,
+      "loss": 1.088,
+      "step": 80
+    },
+    {
+      "epoch": 0.3744493392070485,
+      "grad_norm": 0.6282631754875183,
+      "learning_rate": 2.995358835847891e-05,
+      "loss": 1.0695,
+      "step": 85
+    },
+    {
+      "epoch": 0.3964757709251101,
+      "grad_norm": 0.6258669495582581,
+      "learning_rate": 2.9934820853193538e-05,
+      "loss": 1.0281,
+      "step": 90
+    },
+    {
+      "epoch": 0.4185022026431718,
+      "grad_norm": 0.6035173535346985,
+      "learning_rate": 2.991288236014907e-05,
+      "loss": 1.0368,
+      "step": 95
+    },
+    {
+      "epoch": 0.44052863436123346,
+      "grad_norm": 0.553928554058075,
+      "learning_rate": 2.9887777537365416e-05,
+      "loss": 0.9775,
+      "step": 100
+    },
+    {
+      "epoch": 0.46255506607929514,
+      "grad_norm": 0.6714206337928772,
+      "learning_rate": 2.985951171514326e-05,
+      "loss": 0.9667,
+      "step": 105
+    },
+    {
+      "epoch": 0.4845814977973568,
+      "grad_norm": 0.640594482421875,
+      "learning_rate": 2.982809089493231e-05,
+      "loss": 0.9989,
+      "step": 110
+    },
+    {
+      "epoch": 0.5066079295154186,
+      "grad_norm": 0.6225477457046509,
+      "learning_rate": 2.9793521748057065e-05,
+      "loss": 1.011,
+      "step": 115
+    },
+    {
+      "epoch": 0.5286343612334802,
+      "grad_norm": 0.6979171633720398,
+      "learning_rate": 2.975581161430035e-05,
+      "loss": 0.9919,
+      "step": 120
+    },
+    {
+      "epoch": 0.5506607929515418,
+      "grad_norm": 0.7126419544219971,
+      "learning_rate": 2.971496850034492e-05,
+      "loss": 0.9232,
+      "step": 125
+    },
+    {
+      "epoch": 0.5726872246696035,
+      "grad_norm": 0.8123069405555725,
+      "learning_rate": 2.9671001078073453e-05,
+      "loss": 0.932,
+      "step": 130
+    },
+    {
+      "epoch": 0.5947136563876652,
+      "grad_norm": 0.8383358120918274,
+      "learning_rate": 2.9623918682727355e-05,
+      "loss": 0.9127,
+      "step": 135
+    },
+    {
+      "epoch": 0.6167400881057269,
+      "grad_norm": 0.7429929971694946,
+      "learning_rate": 2.957373131092464e-05,
+      "loss": 0.9187,
+      "step": 140
+    },
+    {
+      "epoch": 0.6387665198237885,
+      "grad_norm": 0.9482147693634033,
+      "learning_rate": 2.9520449618537465e-05,
+      "loss": 0.8848,
+      "step": 145
+    },
+    {
+      "epoch": 0.6607929515418502,
+      "grad_norm": 0.8874338865280151,
+      "learning_rate": 2.946408491842964e-05,
+      "loss": 0.8802,
+      "step": 150
+    },
+    {
+      "epoch": 0.6828193832599119,
+      "grad_norm": 0.7921388149261475,
+      "learning_rate": 2.940464917805466e-05,
+      "loss": 0.8105,
+      "step": 155
+    },
+    {
+      "epoch": 0.7048458149779736,
+      "grad_norm": 0.823284924030304,
+      "learning_rate": 2.9342155016914772e-05,
+      "loss": 0.8335,
+      "step": 160
+    },
+    {
+      "epoch": 0.7268722466960352,
+      "grad_norm": 0.8587915301322937,
+      "learning_rate": 2.927661570388155e-05,
+      "loss": 0.8236,
+      "step": 165
+    },
+    {
+      "epoch": 0.748898678414097,
+      "grad_norm": 0.9209868311882019,
+      "learning_rate": 2.920804515437865e-05,
+      "loss": 0.7313,
+      "step": 170
+    },
+    {
+      "epoch": 0.7709251101321586,
+      "grad_norm": 0.9900104999542236,
+      "learning_rate": 2.9136457927427254e-05,
+      "loss": 0.8445,
+      "step": 175
+    },
+    {
+      "epoch": 0.7929515418502202,
+      "grad_norm": 1.1005882024765015,
+      "learning_rate": 2.9061869222554863e-05,
+      "loss": 0.7258,
+      "step": 180
+    },
+    {
+      "epoch": 0.8149779735682819,
+      "grad_norm": 0.7985159754753113,
+      "learning_rate": 2.898429487656813e-05,
+      "loss": 0.7554,
+      "step": 185
+    },
+    {
+      "epoch": 0.8370044052863436,
+      "grad_norm": 1.0106185674667358,
+      "learning_rate": 2.8903751360190327e-05,
+      "loss": 0.7773,
+      "step": 190
+    },
+    {
+      "epoch": 0.8590308370044053,
+      "grad_norm": 1.0603998899459839,
+      "learning_rate": 2.8820255774564287e-05,
+      "loss": 0.8034,
+      "step": 195
+    },
+    {
+      "epoch": 0.8810572687224669,
+      "grad_norm": 0.9879089593887329,
+      "learning_rate": 2.8733825847621436e-05,
+      "loss": 0.71,
+      "step": 200
+    },
+    {
+      "epoch": 0.9030837004405287,
+      "grad_norm": 1.0008033514022827,
+      "learning_rate": 2.8644479930317776e-05,
+      "loss": 0.7837,
+      "step": 205
+    },
+    {
+      "epoch": 0.9251101321585903,
+      "grad_norm": 1.0469297170639038,
+      "learning_rate": 2.8552236992737572e-05,
+      "loss": 0.7139,
+      "step": 210
+    },
+    {
+      "epoch": 0.947136563876652,
+      "grad_norm": 0.9759002327919006,
+      "learning_rate": 2.8457116620065596e-05,
+      "loss": 0.7207,
+      "step": 215
+    },
+    {
+      "epoch": 0.9691629955947136,
+      "grad_norm": 0.907706618309021,
+      "learning_rate": 2.8359139008428758e-05,
+      "loss": 0.7134,
+      "step": 220
+    },
+    {
+      "epoch": 0.9911894273127754,
+      "grad_norm": 1.022333025932312,
+      "learning_rate": 2.8258324960608043e-05,
+      "loss": 0.669,
+      "step": 225
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1135,
+  "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": 3.282421561897779e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

14_128_e5_3e-5/checkpoint-227/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28b08ed8a6d7b069ef0067d87b8e8cc95aa65498f52b2ba0437d8dc8b44282b7
+size 7736

14_128_e5_3e-5/checkpoint-227/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)

14_128_e5_3e-5/checkpoint-454/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

14_128_e5_3e-5/checkpoint-454/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": [
+    "v_proj",
+    "down_proj",
+    "gate_proj",
+    "o_proj",
+    "k_proj",
+    "q_proj",
+    "up_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

14_128_e5_3e-5/checkpoint-454/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:347e6c758331741c67ef6980f8a5945ea601edb714a55cc3cd2a60c3159ef349
+size 791751704

14_128_e5_3e-5/checkpoint-454/latest

@@ -0,0 +1 @@
+global_step454

14_128_e5_3e-5/checkpoint-454/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:52413e79052a2f1c2c53396d524e56899d2519040b4023b44bcd5b8bff7eed2d
+size 15920

14_128_e5_3e-5/checkpoint-454/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:288cd46fd98e2349c24411371f9000214b52d594a6bf90a1314118b427d406c0
+size 15920

14_128_e5_3e-5/checkpoint-454/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8663e94a0b4deb65bdcb260c305d362bbc66404907d5fceeab586043db18b103
+size 15920