Upload folder using huggingface_hub

36_128_e5_3e-5/checkpoint-1100/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

36_128_e5_3e-5/checkpoint-1100/adapter_config.json

@@ -0,0 +1,39 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "ibm-granite/granite-3.3-8b-base",
+  "bias": "none",
+  "corda_config": null,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 256,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 128,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "up_proj",
+    "gate_proj",
+    "q_proj",
+    "v_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

36_128_e5_3e-5/checkpoint-1100/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:17d16c2e2e027d385ad4f59b9f971a9b261fb037ba83e038b79af9fe637ddd84
+size 791751704

36_128_e5_3e-5/checkpoint-1100/latest

@@ -0,0 +1 @@
+global_step1100

36_128_e5_3e-5/checkpoint-1100/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:33778ba3bbccfbd8a264cdec9c785c018607a5d558a786d2e149f8e6456f8b4a
+size 15920

36_128_e5_3e-5/checkpoint-1100/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a772c77578d96aa697c1d834a32bb3a468a76008b1201557e565009c65ac045
+size 15920

36_128_e5_3e-5/checkpoint-1100/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:69815ac8cc3bb1f311cc4427535f3c370ca812857b17ea4d907e40a70fad030d
+size 15920

36_128_e5_3e-5/checkpoint-1100/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1846b9994441b916174f7efcdd6908325bd6e7583c08f082c6d5ce40c59b7b90
+size 15920

36_128_e5_3e-5/checkpoint-1100/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:638ca7fc7c13ad8851f146e1c2a4c59100d507dd1e5f31c86de73e63225ce2b7
+size 15920

36_128_e5_3e-5/checkpoint-1100/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c2cd5e610145c2b3472a817bc439f6abc0ec1bf939eabb884138ff1b085330bb
+size 15920

36_128_e5_3e-5/checkpoint-1100/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6a59452f385e5ccc389e8f3b68999a82c6b5e079263c54166b01330238827054
+size 15920

36_128_e5_3e-5/checkpoint-1100/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b3abe918893d1ea2623f99ca2e45348703cc736f33f820a888f4ab8f63ff10a0
+size 15920

36_128_e5_3e-5/checkpoint-1100/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3185facce666e8bfe5684d71a78e4e9dd0f3082005a66df93d7a69e964012bcf
+size 1064

36_128_e5_3e-5/checkpoint-1100/special_tokens_map.json

@@ -0,0 +1,45 @@
+{
+  "additional_special_tokens": [
+    "<|endoftext|>",
+    "<fim_prefix>",
+    "<fim_middle>",
+    "<fim_suffix>",
+    "<fim_pad>",
+    "<filename>",
+    "<gh_stars>",
+    "<issue_start>",
+    "<issue_comment>",
+    "<issue_closed>",
+    "<jupyter_start>",
+    "<jupyter_text>",
+    "<jupyter_code>",
+    "<jupyter_output>",
+    "<empty_output>",
+    "<commit_before>",
+    "<commit_msg>",
+    "<commit_after>",
+    "<reponame>"
+  ],
+  "bos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "<reponame>",
+  "unk_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

36_128_e5_3e-5/checkpoint-1100/tokenizer_config.json

@@ -0,0 +1,188 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<fim_prefix>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<fim_middle>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<fim_suffix>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<fim_pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "5": {
+      "content": "<filename>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "6": {
+      "content": "<gh_stars>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "7": {
+      "content": "<issue_start>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "8": {
+      "content": "<issue_comment>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "9": {
+      "content": "<issue_closed>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "10": {
+      "content": "<jupyter_start>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "11": {
+      "content": "<jupyter_text>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "12": {
+      "content": "<jupyter_code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "13": {
+      "content": "<jupyter_output>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "14": {
+      "content": "<empty_output>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "15": {
+      "content": "<commit_before>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "16": {
+      "content": "<commit_msg>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "17": {
+      "content": "<commit_after>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "18": {
+      "content": "<reponame>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<|endoftext|>",
+    "<fim_prefix>",
+    "<fim_middle>",
+    "<fim_suffix>",
+    "<fim_pad>",
+    "<filename>",
+    "<gh_stars>",
+    "<issue_start>",
+    "<issue_comment>",
+    "<issue_closed>",
+    "<jupyter_start>",
+    "<jupyter_text>",
+    "<jupyter_code>",
+    "<jupyter_output>",
+    "<empty_output>",
+    "<commit_before>",
+    "<commit_msg>",
+    "<commit_after>",
+    "<reponame>"
+  ],
+  "bos_token": "<|endoftext|>",
+  "clean_up_tokenization_spaces": true,
+  "eos_token": "<|endoftext|>",
+  "extra_special_tokens": {},
+  "model_max_length": 8192,
+  "pad_token": "<reponame>",
+  "padding_side": "left",
+  "tokenizer_class": "GPT2Tokenizer",
+  "unk_token": "<|endoftext|>",
+  "vocab_size": 49152
+}

36_128_e5_3e-5/checkpoint-1100/trainer_state.json

@@ -0,0 +1,1574 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 1100,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.022779043280182234,
+      "grad_norm": 1.0761960744857788,
+      "learning_rate": 2.181818181818182e-06,
+      "loss": 1.4131,
+      "step": 5
+    },
+    {
+      "epoch": 0.04555808656036447,
+      "grad_norm": 0.7671697735786438,
+      "learning_rate": 4.90909090909091e-06,
+      "loss": 1.3993,
+      "step": 10
+    },
+    {
+      "epoch": 0.0683371298405467,
+      "grad_norm": 0.5325023531913757,
+      "learning_rate": 7.636363636363636e-06,
+      "loss": 1.3575,
+      "step": 15
+    },
+    {
+      "epoch": 0.09111617312072894,
+      "grad_norm": 0.5349851846694946,
+      "learning_rate": 1.0363636363636364e-05,
+      "loss": 1.3727,
+      "step": 20
+    },
+    {
+      "epoch": 0.11389521640091116,
+      "grad_norm": 0.5694962739944458,
+      "learning_rate": 1.309090909090909e-05,
+      "loss": 1.3057,
+      "step": 25
+    },
+    {
+      "epoch": 0.1366742596810934,
+      "grad_norm": 0.541211724281311,
+      "learning_rate": 1.5818181818181818e-05,
+      "loss": 1.2344,
+      "step": 30
+    },
+    {
+      "epoch": 0.15945330296127563,
+      "grad_norm": 0.5115331411361694,
+      "learning_rate": 1.8545454545454545e-05,
+      "loss": 1.3365,
+      "step": 35
+    },
+    {
+      "epoch": 0.18223234624145787,
+      "grad_norm": 0.46041619777679443,
+      "learning_rate": 2.1272727272727273e-05,
+      "loss": 1.2517,
+      "step": 40
+    },
+    {
+      "epoch": 0.20501138952164008,
+      "grad_norm": 0.4588950276374817,
+      "learning_rate": 2.4e-05,
+      "loss": 1.233,
+      "step": 45
+    },
+    {
+      "epoch": 0.22779043280182232,
+      "grad_norm": 0.571067750453949,
+      "learning_rate": 2.6727272727272728e-05,
+      "loss": 1.1614,
+      "step": 50
+    },
+    {
+      "epoch": 0.2505694760820046,
+      "grad_norm": 0.47105151414871216,
+      "learning_rate": 2.9454545454545456e-05,
+      "loss": 1.1789,
+      "step": 55
+    },
+    {
+      "epoch": 0.2733485193621868,
+      "grad_norm": 0.5167170763015747,
+      "learning_rate": 2.9998915465986464e-05,
+      "loss": 1.1674,
+      "step": 60
+    },
+    {
+      "epoch": 0.296127562642369,
+      "grad_norm": 0.5363845229148865,
+      "learning_rate": 2.999450981533838e-05,
+      "loss": 1.1646,
+      "step": 65
+    },
+    {
+      "epoch": 0.31890660592255127,
+      "grad_norm": 0.5026624798774719,
+      "learning_rate": 2.9986716259340288e-05,
+      "loss": 1.1268,
+      "step": 70
+    },
+    {
+      "epoch": 0.3416856492027335,
+      "grad_norm": 0.46400442719459534,
+      "learning_rate": 2.9975536558892034e-05,
+      "loss": 1.0501,
+      "step": 75
+    },
+    {
+      "epoch": 0.36446469248291574,
+      "grad_norm": 0.6948593258857727,
+      "learning_rate": 2.9960973239969295e-05,
+      "loss": 1.0567,
+      "step": 80
+    },
+    {
+      "epoch": 0.38724373576309795,
+      "grad_norm": 0.6658697724342346,
+      "learning_rate": 2.9943029593052822e-05,
+      "loss": 1.1199,
+      "step": 85
+    },
+    {
+      "epoch": 0.41002277904328016,
+      "grad_norm": 0.5796182155609131,
+      "learning_rate": 2.992170967238502e-05,
+      "loss": 1.0407,
+      "step": 90
+    },
+    {
+      "epoch": 0.4328018223234624,
+      "grad_norm": 0.5862748622894287,
+      "learning_rate": 2.9897018295053883e-05,
+      "loss": 1.0524,
+      "step": 95
+    },
+    {
+      "epoch": 0.45558086560364464,
+      "grad_norm": 0.714149534702301,
+      "learning_rate": 2.9868961039904628e-05,
+      "loss": 1.0367,
+      "step": 100
+    },
+    {
+      "epoch": 0.4783599088838269,
+      "grad_norm": 0.7176491618156433,
+      "learning_rate": 2.983754424627919e-05,
+      "loss": 1.0013,
+      "step": 105
+    },
+    {
+      "epoch": 0.5011389521640092,
+      "grad_norm": 0.6981073617935181,
+      "learning_rate": 2.9802775012583884e-05,
+      "loss": 1.0079,
+      "step": 110
+    },
+    {
+      "epoch": 0.5239179954441914,
+      "grad_norm": 0.5984258651733398,
+      "learning_rate": 2.9764661194685583e-05,
+      "loss": 0.975,
+      "step": 115
+    },
+    {
+      "epoch": 0.5466970387243736,
+      "grad_norm": 0.7986994981765747,
+      "learning_rate": 2.972321140413672e-05,
+      "loss": 0.9417,
+      "step": 120
+    },
+    {
+      "epoch": 0.5694760820045558,
+      "grad_norm": 0.6552079916000366,
+      "learning_rate": 2.9678435006229585e-05,
+      "loss": 0.9208,
+      "step": 125
+    },
+    {
+      "epoch": 0.592255125284738,
+      "grad_norm": 0.7615071535110474,
+      "learning_rate": 2.9630342117880293e-05,
+      "loss": 0.9208,
+      "step": 130
+    },
+    {
+      "epoch": 0.6150341685649203,
+      "grad_norm": 0.73598313331604,
+      "learning_rate": 2.957894360534295e-05,
+      "loss": 0.9459,
+      "step": 135
+    },
+    {
+      "epoch": 0.6378132118451025,
+      "grad_norm": 0.9029515981674194,
+      "learning_rate": 2.9524251081754475e-05,
+      "loss": 0.9338,
+      "step": 140
+    },
+    {
+      "epoch": 0.6605922551252847,
+      "grad_norm": 0.8563511967658997,
+      "learning_rate": 2.9466276904510713e-05,
+      "loss": 0.8437,
+      "step": 145
+    },
+    {
+      "epoch": 0.683371298405467,
+      "grad_norm": 0.8966438174247742,
+      "learning_rate": 2.9405034172474363e-05,
+      "loss": 0.8995,
+      "step": 150
+    },
+    {
+      "epoch": 0.7061503416856492,
+      "grad_norm": 0.9085351228713989,
+      "learning_rate": 2.9340536723015367e-05,
+      "loss": 0.8988,
+      "step": 155
+    },
+    {
+      "epoch": 0.7289293849658315,
+      "grad_norm": 0.920749843120575,
+      "learning_rate": 2.927279912888447e-05,
+      "loss": 0.8457,
+      "step": 160
+    },
+    {
+      "epoch": 0.7517084282460137,
+      "grad_norm": 0.844048023223877,
+      "learning_rate": 2.920183669492061e-05,
+      "loss": 0.8699,
+      "step": 165
+    },
+    {
+      "epoch": 0.7744874715261959,
+      "grad_norm": 1.1362037658691406,
+      "learning_rate": 2.9127665454592872e-05,
+      "loss": 0.8561,
+      "step": 170
+    },
+    {
+      "epoch": 0.7972665148063781,
+      "grad_norm": 0.9490585923194885,
+      "learning_rate": 2.9050302166377858e-05,
+      "loss": 0.8016,
+      "step": 175
+    },
+    {
+      "epoch": 0.8200455580865603,
+      "grad_norm": 1.047890543937683,
+      "learning_rate": 2.896976430997323e-05,
+      "loss": 0.8047,
+      "step": 180
+    },
+    {
+      "epoch": 0.8428246013667426,
+      "grad_norm": 1.252584457397461,
+      "learning_rate": 2.8886070082348268e-05,
+      "loss": 0.8087,
+      "step": 185
+    },
+    {
+      "epoch": 0.8656036446469249,
+      "grad_norm": 1.0744786262512207,
+      "learning_rate": 2.879923839363242e-05,
+      "loss": 0.7823,
+      "step": 190
+    },
+    {
+      "epoch": 0.8883826879271071,
+      "grad_norm": 0.9743952751159668,
+      "learning_rate": 2.870928886284267e-05,
+      "loss": 0.727,
+      "step": 195
+    },
+    {
+      "epoch": 0.9111617312072893,
+      "grad_norm": 1.029543399810791,
+      "learning_rate": 2.8616241813450755e-05,
+      "loss": 0.7648,
+      "step": 200
+    },
+    {
+      "epoch": 0.9339407744874715,
+      "grad_norm": 1.031134009361267,
+      "learning_rate": 2.852011826879125e-05,
+      "loss": 0.7444,
+      "step": 205
+    },
+    {
+      "epoch": 0.9567198177676538,
+      "grad_norm": 1.1572555303573608,
+      "learning_rate": 2.8420939947311454e-05,
+      "loss": 0.7408,
+      "step": 210
+    },
+    {
+      "epoch": 0.979498861047836,
+      "grad_norm": 1.0397342443466187,
+      "learning_rate": 2.8318729257664265e-05,
+      "loss": 0.7015,
+      "step": 215
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.4852145910263062,
+      "learning_rate": 2.821350929364512e-05,
+      "loss": 0.6917,
+      "step": 220
+    },
+    {
+      "epoch": 1.0227790432801822,
+      "grad_norm": 1.1299735307693481,
+      "learning_rate": 2.8105303828974095e-05,
+      "loss": 0.6275,
+      "step": 225
+    },
+    {
+      "epoch": 1.0455580865603644,
+      "grad_norm": 1.2456783056259155,
+      "learning_rate": 2.79941373119244e-05,
+      "loss": 0.5556,
+      "step": 230
+    },
+    {
+      "epoch": 1.0683371298405466,
+      "grad_norm": 1.192712664604187,
+      "learning_rate": 2.7880034859798476e-05,
+      "loss": 0.5791,
+      "step": 235
+    },
+    {
+      "epoch": 1.0911161731207288,
+      "grad_norm": 1.3306063413619995,
+      "learning_rate": 2.7763022253252882e-05,
+      "loss": 0.5883,
+      "step": 240
+    },
+    {
+      "epoch": 1.1138952164009113,
+      "grad_norm": 1.1254067420959473,
+      "learning_rate": 2.7643125930473375e-05,
+      "loss": 0.5807,
+      "step": 245
+    },
+    {
+      "epoch": 1.1366742596810935,
+      "grad_norm": 1.2082189321517944,
+      "learning_rate": 2.7520372981201362e-05,
+      "loss": 0.5561,
+      "step": 250
+    },
+    {
+      "epoch": 1.1594533029612757,
+      "grad_norm": 1.1813552379608154,
+      "learning_rate": 2.739479114061316e-05,
+      "loss": 0.5774,
+      "step": 255
+    },
+    {
+      "epoch": 1.182232346241458,
+      "grad_norm": 1.1329344511032104,
+      "learning_rate": 2.7266408783053437e-05,
+      "loss": 0.5704,
+      "step": 260
+    },
+    {
+      "epoch": 1.20501138952164,
+      "grad_norm": 1.1925603151321411,
+      "learning_rate": 2.7135254915624213e-05,
+      "loss": 0.5447,
+      "step": 265
+    },
+    {
+      "epoch": 1.2277904328018223,
+      "grad_norm": 1.3497655391693115,
+      "learning_rate": 2.700135917163092e-05,
+      "loss": 0.5564,
+      "step": 270
+    },
+    {
+      "epoch": 1.2505694760820045,
+      "grad_norm": 1.0640569925308228,
+      "learning_rate": 2.6864751803886963e-05,
+      "loss": 0.5631,
+      "step": 275
+    },
+    {
+      "epoch": 1.2733485193621867,
+      "grad_norm": 1.0790766477584839,
+      "learning_rate": 2.672546367787828e-05,
+      "loss": 0.5358,
+      "step": 280
+    },
+    {
+      "epoch": 1.296127562642369,
+      "grad_norm": 1.149834394454956,
+      "learning_rate": 2.6583526264789538e-05,
+      "loss": 0.5354,
+      "step": 285
+    },
+    {
+      "epoch": 1.3189066059225514,
+      "grad_norm": 1.3402076959609985,
+      "learning_rate": 2.6438971634393416e-05,
+      "loss": 0.526,
+      "step": 290
+    },
+    {
+      "epoch": 1.3416856492027334,
+      "grad_norm": 1.1796164512634277,
+      "learning_rate": 2.6291832447804658e-05,
+      "loss": 0.5534,
+      "step": 295
+    },
+    {
+      "epoch": 1.3644646924829158,
+      "grad_norm": 1.2765209674835205,
+      "learning_rate": 2.6142141950100533e-05,
+      "loss": 0.5322,
+      "step": 300
+    },
+    {
+      "epoch": 1.387243735763098,
+      "grad_norm": 1.3256808519363403,
+      "learning_rate": 2.598993396280934e-05,
+      "loss": 0.463,
+      "step": 305
+    },
+    {
+      "epoch": 1.4100227790432802,
+      "grad_norm": 1.1312872171401978,
+      "learning_rate": 2.5835242876268667e-05,
+      "loss": 0.5104,
+      "step": 310
+    },
+    {
+      "epoch": 1.4328018223234624,
+      "grad_norm": 1.1364809274673462,
+      "learning_rate": 2.5678103641855122e-05,
+      "loss": 0.5196,
+      "step": 315
+    },
+    {
+      "epoch": 1.4555808656036446,
+      "grad_norm": 1.1903729438781738,
+      "learning_rate": 2.5518551764087326e-05,
+      "loss": 0.4946,
+      "step": 320
+    },
+    {
+      "epoch": 1.4783599088838268,
+      "grad_norm": 1.1419878005981445,
+      "learning_rate": 2.5356623292603923e-05,
+      "loss": 0.4853,
+      "step": 325
+    },
+    {
+      "epoch": 1.501138952164009,
+      "grad_norm": 1.103503704071045,
+      "learning_rate": 2.519235481401837e-05,
+      "loss": 0.4929,
+      "step": 330
+    },
+    {
+      "epoch": 1.5239179954441915,
+      "grad_norm": 1.1578071117401123,
+      "learning_rate": 2.5025783443652504e-05,
+      "loss": 0.4715,
+      "step": 335
+    },
+    {
+      "epoch": 1.5466970387243735,
+      "grad_norm": 1.2676241397857666,
+      "learning_rate": 2.4856946817150555e-05,
+      "loss": 0.4388,
+      "step": 340
+    },
+    {
+      "epoch": 1.569476082004556,
+      "grad_norm": 1.409696340560913,
+      "learning_rate": 2.4685883081975646e-05,
+      "loss": 0.4604,
+      "step": 345
+    },
+    {
+      "epoch": 1.592255125284738,
+      "grad_norm": 1.1401517391204834,
+      "learning_rate": 2.4512630888790623e-05,
+      "loss": 0.5073,
+      "step": 350
+    },
+    {
+      "epoch": 1.6150341685649203,
+      "grad_norm": 1.2104820013046265,
+      "learning_rate": 2.4337229382725245e-05,
+      "loss": 0.4442,
+      "step": 355
+    },
+    {
+      "epoch": 1.6378132118451025,
+      "grad_norm": 1.1961909532546997,
+      "learning_rate": 2.4159718194531573e-05,
+      "loss": 0.4318,
+      "step": 360
+    },
+    {
+      "epoch": 1.6605922551252847,
+      "grad_norm": 1.214699625968933,
+      "learning_rate": 2.398013743162969e-05,
+      "loss": 0.4273,
+      "step": 365
+    },
+    {
+      "epoch": 1.683371298405467,
+      "grad_norm": 1.2664371728897095,
+      "learning_rate": 2.3798527669045714e-05,
+      "loss": 0.4125,
+      "step": 370
+    },
+    {
+      "epoch": 1.7061503416856492,
+      "grad_norm": 1.263482928276062,
+      "learning_rate": 2.3614929940244155e-05,
+      "loss": 0.4843,
+      "step": 375
+    },
+    {
+      "epoch": 1.7289293849658316,
+      "grad_norm": 1.2268201112747192,
+      "learning_rate": 2.3429385727856674e-05,
+      "loss": 0.4274,
+      "step": 380
+    },
+    {
+      "epoch": 1.7517084282460136,
+      "grad_norm": 1.437774419784546,
+      "learning_rate": 2.3241936954309382e-05,
+      "loss": 0.4274,
+      "step": 385
+    },
+    {
+      "epoch": 1.774487471526196,
+      "grad_norm": 1.389108657836914,
+      "learning_rate": 2.305262597235076e-05,
+      "loss": 0.397,
+      "step": 390
+    },
+    {
+      "epoch": 1.797266514806378,
+      "grad_norm": 1.2691073417663574,
+      "learning_rate": 2.2861495555482337e-05,
+      "loss": 0.4511,
+      "step": 395
+    },
+    {
+      "epoch": 1.8200455580865604,
+      "grad_norm": 1.2614039182662964,
+      "learning_rate": 2.266858888829433e-05,
+      "loss": 0.4209,
+      "step": 400
+    },
+    {
+      "epoch": 1.8428246013667426,
+      "grad_norm": 1.3508285284042358,
+      "learning_rate": 2.247394955670838e-05,
+      "loss": 0.3974,
+      "step": 405
+    },
+    {
+      "epoch": 1.8656036446469249,
+      "grad_norm": 1.3065261840820312,
+      "learning_rate": 2.2277621538129612e-05,
+      "loss": 0.4346,
+      "step": 410
+    },
+    {
+      "epoch": 1.888382687927107,
+      "grad_norm": 1.1292978525161743,
+      "learning_rate": 2.2079649191510248e-05,
+      "loss": 0.4045,
+      "step": 415
+    },
+    {
+      "epoch": 1.9111617312072893,
+      "grad_norm": 1.2417933940887451,
+      "learning_rate": 2.1880077247327008e-05,
+      "loss": 0.3923,
+      "step": 420
+    },
+    {
+      "epoch": 1.9339407744874715,
+      "grad_norm": 1.2816452980041504,
+      "learning_rate": 2.1678950797474578e-05,
+      "loss": 0.3782,
+      "step": 425
+    },
+    {
+      "epoch": 1.9567198177676537,
+      "grad_norm": 1.464908242225647,
+      "learning_rate": 2.1476315285077393e-05,
+      "loss": 0.3885,
+      "step": 430
+    },
+    {
+      "epoch": 1.9794988610478361,
+      "grad_norm": 1.3482881784439087,
+      "learning_rate": 2.1272216494222108e-05,
+      "loss": 0.3641,
+      "step": 435
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.697672724723816,
+      "learning_rate": 2.1066700539612964e-05,
+      "loss": 0.3719,
+      "step": 440
+    },
+    {
+      "epoch": 2.0227790432801824,
+      "grad_norm": 1.3298288583755493,
+      "learning_rate": 2.0859813856152513e-05,
+      "loss": 0.3461,
+      "step": 445
+    },
+    {
+      "epoch": 2.0455580865603644,
+      "grad_norm": 1.277348279953003,
+      "learning_rate": 2.0651603188449976e-05,
+      "loss": 0.2962,
+      "step": 450
+    },
+    {
+      "epoch": 2.068337129840547,
+      "grad_norm": 1.368998408317566,
+      "learning_rate": 2.0442115580259615e-05,
+      "loss": 0.2816,
+      "step": 455
+    },
+    {
+      "epoch": 2.091116173120729,
+      "grad_norm": 1.5888235569000244,
+      "learning_rate": 2.023139836385153e-05,
+      "loss": 0.3188,
+      "step": 460
+    },
+    {
+      "epoch": 2.1138952164009113,
+      "grad_norm": 1.3389489650726318,
+      "learning_rate": 2.0019499149317252e-05,
+      "loss": 0.288,
+      "step": 465
+    },
+    {
+      "epoch": 2.1366742596810933,
+      "grad_norm": 1.2392019033432007,
+      "learning_rate": 1.980646581381258e-05,
+      "loss": 0.294,
+      "step": 470
+    },
+    {
+      "epoch": 2.1594533029612757,
+      "grad_norm": 1.3480933904647827,
+      "learning_rate": 1.959234649074007e-05,
+      "loss": 0.2866,
+      "step": 475
+    },
+    {
+      "epoch": 2.1822323462414577,
+      "grad_norm": 1.2456738948822021,
+      "learning_rate": 1.9377189558873634e-05,
+      "loss": 0.3042,
+      "step": 480
+    },
+    {
+      "epoch": 2.20501138952164,
+      "grad_norm": 1.3260343074798584,
+      "learning_rate": 1.916104363142767e-05,
+      "loss": 0.2762,
+      "step": 485
+    },
+    {
+      "epoch": 2.2277904328018225,
+      "grad_norm": 1.2444536685943604,
+      "learning_rate": 1.8943957545073273e-05,
+      "loss": 0.2668,
+      "step": 490
+    },
+    {
+      "epoch": 2.2505694760820045,
+      "grad_norm": 1.2988715171813965,
+      "learning_rate": 1.8725980348903898e-05,
+      "loss": 0.2667,
+      "step": 495
+    },
+    {
+      "epoch": 2.273348519362187,
+      "grad_norm": 1.2734081745147705,
+      "learning_rate": 1.850716129335309e-05,
+      "loss": 0.2921,
+      "step": 500
+    },
+    {
+      "epoch": 2.296127562642369,
+      "grad_norm": 1.2905198335647583,
+      "learning_rate": 1.8287549819066665e-05,
+      "loss": 0.2674,
+      "step": 505
+    },
+    {
+      "epoch": 2.3189066059225514,
+      "grad_norm": 1.3245160579681396,
+      "learning_rate": 1.8067195545731942e-05,
+      "loss": 0.2727,
+      "step": 510
+    },
+    {
+      "epoch": 2.3416856492027334,
+      "grad_norm": 1.383800983428955,
+      "learning_rate": 1.784614826086655e-05,
+      "loss": 0.2591,
+      "step": 515
+    },
+    {
+      "epoch": 2.364464692482916,
+      "grad_norm": 1.4593446254730225,
+      "learning_rate": 1.7624457908569226e-05,
+      "loss": 0.2517,
+      "step": 520
+    },
+    {
+      "epoch": 2.387243735763098,
+      "grad_norm": 1.3672267198562622,
+      "learning_rate": 1.7402174578235356e-05,
+      "loss": 0.265,
+      "step": 525
+    },
+    {
+      "epoch": 2.41002277904328,
+      "grad_norm": 1.3659502267837524,
+      "learning_rate": 1.7179348493239582e-05,
+      "loss": 0.2567,
+      "step": 530
+    },
+    {
+      "epoch": 2.432801822323462,
+      "grad_norm": 1.3636661767959595,
+      "learning_rate": 1.695602999958819e-05,
+      "loss": 0.247,
+      "step": 535
+    },
+    {
+      "epoch": 2.4555808656036446,
+      "grad_norm": 1.3314388990402222,
+      "learning_rate": 1.6732269554543794e-05,
+      "loss": 0.2442,
+      "step": 540
+    },
+    {
+      "epoch": 2.478359908883827,
+      "grad_norm": 1.3202639818191528,
+      "learning_rate": 1.6508117715224834e-05,
+      "loss": 0.2259,
+      "step": 545
+    },
+    {
+      "epoch": 2.501138952164009,
+      "grad_norm": 1.1319786310195923,
+      "learning_rate": 1.6283625127182596e-05,
+      "loss": 0.2525,
+      "step": 550
+    },
+    {
+      "epoch": 2.5239179954441915,
+      "grad_norm": 1.2015621662139893,
+      "learning_rate": 1.605884251295815e-05,
+      "loss": 0.2403,
+      "step": 555
+    },
+    {
+      "epoch": 2.5466970387243735,
+      "grad_norm": 1.2344118356704712,
+      "learning_rate": 1.583382066062198e-05,
+      "loss": 0.2558,
+      "step": 560
+    },
+    {
+      "epoch": 2.569476082004556,
+      "grad_norm": 1.3637142181396484,
+      "learning_rate": 1.5608610412298758e-05,
+      "loss": 0.2385,
+      "step": 565
+    },
+    {
+      "epoch": 2.592255125284738,
+      "grad_norm": 1.2858781814575195,
+      "learning_rate": 1.5383262652679914e-05,
+      "loss": 0.2283,
+      "step": 570
+    },
+    {
+      "epoch": 2.6150341685649203,
+      "grad_norm": 1.1508971452713013,
+      "learning_rate": 1.51578282975266e-05,
+      "loss": 0.2213,
+      "step": 575
+    },
+    {
+      "epoch": 2.6378132118451028,
+      "grad_norm": 1.2581448554992676,
+      "learning_rate": 1.4932358282165646e-05,
+      "loss": 0.2368,
+      "step": 580
+    },
+    {
+      "epoch": 2.6605922551252847,
+      "grad_norm": 1.173695683479309,
+      "learning_rate": 1.4706903549981035e-05,
+      "loss": 0.2353,
+      "step": 585
+    },
+    {
+      "epoch": 2.6833712984054667,
+      "grad_norm": 1.4867786169052124,
+      "learning_rate": 1.4481515040903639e-05,
+      "loss": 0.2278,
+      "step": 590
+    },
+    {
+      "epoch": 2.706150341685649,
+      "grad_norm": 1.411230206489563,
+      "learning_rate": 1.4256243679901665e-05,
+      "loss": 0.2214,
+      "step": 595
+    },
+    {
+      "epoch": 2.7289293849658316,
+      "grad_norm": 1.1627558469772339,
+      "learning_rate": 1.4031140365474543e-05,
+      "loss": 0.2095,
+      "step": 600
+    },
+    {
+      "epoch": 2.7517084282460136,
+      "grad_norm": 1.2212531566619873,
+      "learning_rate": 1.3806255958152725e-05,
+      "loss": 0.2208,
+      "step": 605
+    },
+    {
+      "epoch": 2.774487471526196,
+      "grad_norm": 1.4650417566299438,
+      "learning_rate": 1.3581641269006142e-05,
+      "loss": 0.2072,
+      "step": 610
+    },
+    {
+      "epoch": 2.797266514806378,
+      "grad_norm": 1.1035012006759644,
+      "learning_rate": 1.3357347048163767e-05,
+      "loss": 0.2158,
+      "step": 615
+    },
+    {
+      "epoch": 2.8200455580865604,
+      "grad_norm": 1.2653112411499023,
+      "learning_rate": 1.3133423973346996e-05,
+      "loss": 0.2226,
+      "step": 620
+    },
+    {
+      "epoch": 2.8428246013667424,
+      "grad_norm": 1.2133084535598755,
+      "learning_rate": 1.2909922638419336e-05,
+      "loss": 0.2191,
+      "step": 625
+    },
+    {
+      "epoch": 2.865603644646925,
+      "grad_norm": 1.3063286542892456,
+      "learning_rate": 1.2686893541955113e-05,
+      "loss": 0.2295,
+      "step": 630
+    },
+    {
+      "epoch": 2.8883826879271073,
+      "grad_norm": 1.2617278099060059,
+      "learning_rate": 1.2464387075829654e-05,
+      "loss": 0.2168,
+      "step": 635
+    },
+    {
+      "epoch": 2.9111617312072893,
+      "grad_norm": 1.2661510705947876,
+      "learning_rate": 1.2242453513833584e-05,
+      "loss": 0.2109,
+      "step": 640
+    },
+    {
+      "epoch": 2.9339407744874713,
+      "grad_norm": 1.5437356233596802,
+      "learning_rate": 1.2021143000313822e-05,
+      "loss": 0.1772,
+      "step": 645
+    },
+    {
+      "epoch": 2.9567198177676537,
+      "grad_norm": 1.3134946823120117,
+      "learning_rate": 1.18005055388438e-05,
+      "loss": 0.2097,
+      "step": 650
+    },
+    {
+      "epoch": 2.979498861047836,
+      "grad_norm": 1.4121174812316895,
+      "learning_rate": 1.1580590980925527e-05,
+      "loss": 0.2145,
+      "step": 655
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 1.8628265857696533,
+      "learning_rate": 1.1361449014725974e-05,
+      "loss": 0.212,
+      "step": 660
+    },
+    {
+      "epoch": 3.0227790432801824,
+      "grad_norm": 1.1067537069320679,
+      "learning_rate": 1.1143129153850385e-05,
+      "loss": 0.1637,
+      "step": 665
+    },
+    {
+      "epoch": 3.0455580865603644,
+      "grad_norm": 1.003252387046814,
+      "learning_rate": 1.0925680726154994e-05,
+      "loss": 0.1557,
+      "step": 670
+    },
+    {
+      "epoch": 3.068337129840547,
+      "grad_norm": 1.0778887271881104,
+      "learning_rate": 1.0709152862601775e-05,
+      "loss": 0.1517,
+      "step": 675
+    },
+    {
+      "epoch": 3.091116173120729,
+      "grad_norm": 1.4502981901168823,
+      "learning_rate": 1.049359448615763e-05,
+      "loss": 0.165,
+      "step": 680
+    },
+    {
+      "epoch": 3.1138952164009113,
+      "grad_norm": 1.1783398389816284,
+      "learning_rate": 1.027905430074058e-05,
+      "loss": 0.1593,
+      "step": 685
+    },
+    {
+      "epoch": 3.1366742596810933,
+      "grad_norm": 0.985548198223114,
+      "learning_rate": 1.0065580780215457e-05,
+      "loss": 0.1401,
+      "step": 690
+    },
+    {
+      "epoch": 3.1594533029612757,
+      "grad_norm": 1.2806044816970825,
+      "learning_rate": 9.853222157441562e-06,
+      "loss": 0.1588,
+      "step": 695
+    },
+    {
+      "epoch": 3.1822323462414577,
+      "grad_norm": 1.209817886352539,
+      "learning_rate": 9.642026413374792e-06,
+      "loss": 0.1473,
+      "step": 700
+    },
+    {
+      "epoch": 3.20501138952164,
+      "grad_norm": 1.0814965963363647,
+      "learning_rate": 9.432041266226686e-06,
+      "loss": 0.1494,
+      "step": 705
+    },
+    {
+      "epoch": 3.2277904328018225,
+      "grad_norm": 1.0251121520996094,
+      "learning_rate": 9.223314160682821e-06,
+      "loss": 0.1426,
+      "step": 710
+    },
+    {
+      "epoch": 3.2505694760820045,
+      "grad_norm": 1.1800068616867065,
+      "learning_rate": 9.015892257182988e-06,
+      "loss": 0.1242,
+      "step": 715
+    },
+    {
+      "epoch": 3.273348519362187,
+      "grad_norm": 1.143933653831482,
+      "learning_rate": 8.809822421265665e-06,
+      "loss": 0.1483,
+      "step": 720
+    },
+    {
+      "epoch": 3.296127562642369,
+      "grad_norm": 1.2569365501403809,
+      "learning_rate": 8.605151212979033e-06,
+      "loss": 0.1376,
+      "step": 725
+    },
+    {
+      "epoch": 3.3189066059225514,
+      "grad_norm": 1.1757892370224,
+      "learning_rate": 8.401924876361101e-06,
+      "loss": 0.1316,
+      "step": 730
+    },
+    {
+      "epoch": 3.3416856492027334,
+      "grad_norm": 1.0660637617111206,
+      "learning_rate": 8.200189328991125e-06,
+      "loss": 0.144,
+      "step": 735
+    },
+    {
+      "epoch": 3.364464692482916,
+      "grad_norm": 1.2305912971496582,
+      "learning_rate": 7.999990151614895e-06,
+      "loss": 0.1518,
+      "step": 740
+    },
+    {
+      "epoch": 3.387243735763098,
+      "grad_norm": 1.0165345668792725,
+      "learning_rate": 7.801372577846091e-06,
+      "loss": 0.1302,
+      "step": 745
+    },
+    {
+      "epoch": 3.41002277904328,
+      "grad_norm": 1.1290851831436157,
+      "learning_rate": 7.604381483946028e-06,
+      "loss": 0.1397,
+      "step": 750
+    },
+    {
+      "epoch": 3.432801822323462,
+      "grad_norm": 1.2893080711364746,
+      "learning_rate": 7.409061378684187e-06,
+      "loss": 0.1416,
+      "step": 755
+    },
+    {
+      "epoch": 3.4555808656036446,
+      "grad_norm": 1.0431509017944336,
+      "learning_rate": 7.215456393281777e-06,
+      "loss": 0.1244,
+      "step": 760
+    },
+    {
+      "epoch": 3.478359908883827,
+      "grad_norm": 1.3273895978927612,
+      "learning_rate": 7.023610271440583e-06,
+      "loss": 0.1293,
+      "step": 765
+    },
+    {
+      "epoch": 3.501138952164009,
+      "grad_norm": 1.1418635845184326,
+      "learning_rate": 6.833566359459399e-06,
+      "loss": 0.132,
+      "step": 770
+    },
+    {
+      "epoch": 3.5239179954441915,
+      "grad_norm": 1.0589687824249268,
+      "learning_rate": 6.645367596440202e-06,
+      "loss": 0.1257,
+      "step": 775
+    },
+    {
+      "epoch": 3.5466970387243735,
+      "grad_norm": 1.180116057395935,
+      "learning_rate": 6.459056504586382e-06,
+      "loss": 0.1361,
+      "step": 780
+    },
+    {
+      "epoch": 3.569476082004556,
+      "grad_norm": 1.1112143993377686,
+      "learning_rate": 6.274675179595138e-06,
+      "loss": 0.119,
+      "step": 785
+    },
+    {
+      "epoch": 3.592255125284738,
+      "grad_norm": 1.2985987663269043,
+      "learning_rate": 6.092265281146247e-06,
+      "loss": 0.1291,
+      "step": 790
+    },
+    {
+      "epoch": 3.6150341685649203,
+      "grad_norm": 1.1494839191436768,
+      "learning_rate": 5.911868023489367e-06,
+      "loss": 0.1256,
+      "step": 795
+    },
+    {
+      "epoch": 3.6378132118451028,
+      "grad_norm": 1.0739927291870117,
+      "learning_rate": 5.733524166131927e-06,
+      "loss": 0.1417,
+      "step": 800
+    },
+    {
+      "epoch": 3.6605922551252847,
+      "grad_norm": 1.0789777040481567,
+      "learning_rate": 5.557274004629831e-06,
+      "loss": 0.1249,
+      "step": 805
+    },
+    {
+      "epoch": 3.6833712984054667,
+      "grad_norm": 1.1680610179901123,
+      "learning_rate": 5.3831573614829445e-06,
+      "loss": 0.1203,
+      "step": 810
+    },
+    {
+      "epoch": 3.706150341685649,
+      "grad_norm": 1.1551055908203125,
+      "learning_rate": 5.21121357713747e-06,
+      "loss": 0.1169,
+      "step": 815
+    },
+    {
+      "epoch": 3.7289293849658316,
+      "grad_norm": 1.0219001770019531,
+      "learning_rate": 5.04148150109728e-06,
+      "loss": 0.1292,
+      "step": 820
+    },
+    {
+      "epoch": 3.7517084282460136,
+      "grad_norm": 1.1913859844207764,
+      "learning_rate": 4.8739994831460795e-06,
+      "loss": 0.1242,
+      "step": 825
+    },
+    {
+      "epoch": 3.774487471526196,
+      "grad_norm": 1.3804059028625488,
+      "learning_rate": 4.708805364682613e-06,
+      "loss": 0.1089,
+      "step": 830
+    },
+    {
+      "epoch": 3.797266514806378,
+      "grad_norm": 1.1627036333084106,
+      "learning_rate": 4.5459364701706e-06,
+      "loss": 0.1262,
+      "step": 835
+    },
+    {
+      "epoch": 3.8200455580865604,
+      "grad_norm": 1.069296956062317,
+      "learning_rate": 4.385429598705566e-06,
+      "loss": 0.1094,
+      "step": 840
+    },
+    {
+      "epoch": 3.8428246013667424,
+      "grad_norm": 1.0232250690460205,
+      "learning_rate": 4.227321015700348e-06,
+      "loss": 0.1258,
+      "step": 845
+    },
+    {
+      "epoch": 3.865603644646925,
+      "grad_norm": 0.9653981924057007,
+      "learning_rate": 4.07164644469113e-06,
+      "loss": 0.1082,
+      "step": 850
+    },
+    {
+      "epoch": 3.8883826879271073,
+      "grad_norm": 1.1044456958770752,
+      "learning_rate": 3.918441059266026e-06,
+      "loss": 0.1053,
+      "step": 855
+    },
+    {
+      "epoch": 3.9111617312072893,
+      "grad_norm": 0.9676677584648132,
+      "learning_rate": 3.7677394751177957e-06,
+      "loss": 0.1271,
+      "step": 860
+    },
+    {
+      "epoch": 3.9339407744874713,
+      "grad_norm": 1.0099689960479736,
+      "learning_rate": 3.619575742222695e-06,
+      "loss": 0.1132,
+      "step": 865
+    },
+    {
+      "epoch": 3.9567198177676537,
+      "grad_norm": 0.9933716058731079,
+      "learning_rate": 3.473983337147118e-06,
+      "loss": 0.1183,
+      "step": 870
+    },
+    {
+      "epoch": 3.979498861047836,
+      "grad_norm": 0.9863209128379822,
+      "learning_rate": 3.3309951554837882e-06,
+      "loss": 0.1159,
+      "step": 875
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 1.268900990486145,
+      "learning_rate": 3.1906435044192545e-06,
+      "loss": 0.1049,
+      "step": 880
+    },
+    {
+      "epoch": 4.022779043280182,
+      "grad_norm": 0.895099401473999,
+      "learning_rate": 3.0529600954342717e-06,
+      "loss": 0.0875,
+      "step": 885
+    },
+    {
+      "epoch": 4.045558086560365,
+      "grad_norm": 0.8261404037475586,
+      "learning_rate": 2.9179760371388347e-06,
+      "loss": 0.0897,
+      "step": 890
+    },
+    {
+      "epoch": 4.068337129840547,
+      "grad_norm": 0.902686595916748,
+      "learning_rate": 2.785721828243399e-06,
+      "loss": 0.0908,
+      "step": 895
+    },
+    {
+      "epoch": 4.091116173120729,
+      "grad_norm": 0.9131365418434143,
+      "learning_rate": 2.6562273506679007e-06,
+      "loss": 0.0959,
+      "step": 900
+    },
+    {
+      "epoch": 4.113895216400911,
+      "grad_norm": 1.048699975013733,
+      "learning_rate": 2.5295218627901695e-06,
+      "loss": 0.1006,
+      "step": 905
+    },
+    {
+      "epoch": 4.136674259681094,
+      "grad_norm": 0.8155301213264465,
+      "learning_rate": 2.405633992835164e-06,
+      "loss": 0.0904,
+      "step": 910
+    },
+    {
+      "epoch": 4.159453302961276,
+      "grad_norm": 0.8433247804641724,
+      "learning_rate": 2.2845917324066522e-06,
+      "loss": 0.0831,
+      "step": 915
+    },
+    {
+      "epoch": 4.182232346241458,
+      "grad_norm": 0.812890887260437,
+      "learning_rate": 2.166422430162693e-06,
+      "loss": 0.0822,
+      "step": 920
+    },
+    {
+      "epoch": 4.20501138952164,
+      "grad_norm": 0.9599792957305908,
+      "learning_rate": 2.0511527856363916e-06,
+      "loss": 0.0946,
+      "step": 925
+    },
+    {
+      "epoch": 4.2277904328018225,
+      "grad_norm": 1.0594674348831177,
+      "learning_rate": 1.9388088432033446e-06,
+      "loss": 0.0904,
+      "step": 930
+    },
+    {
+      "epoch": 4.2505694760820045,
+      "grad_norm": 1.017570972442627,
+      "learning_rate": 1.8294159861970605e-06,
+      "loss": 0.0881,
+      "step": 935
+    },
+    {
+      "epoch": 4.2733485193621865,
+      "grad_norm": 0.9285730719566345,
+      "learning_rate": 1.7229989311738142e-06,
+      "loss": 0.0871,
+      "step": 940
+    },
+    {
+      "epoch": 4.296127562642369,
+      "grad_norm": 0.9436463117599487,
+      "learning_rate": 1.619581722328073e-06,
+      "loss": 0.097,
+      "step": 945
+    },
+    {
+      "epoch": 4.318906605922551,
+      "grad_norm": 0.8113735914230347,
+      "learning_rate": 1.519187726059903e-06,
+      "loss": 0.0943,
+      "step": 950
+    },
+    {
+      "epoch": 4.341685649202733,
+      "grad_norm": 0.9037413001060486,
+      "learning_rate": 1.421839625695492e-06,
+      "loss": 0.0929,
+      "step": 955
+    },
+    {
+      "epoch": 4.364464692482915,
+      "grad_norm": 0.8155743479728699,
+      "learning_rate": 1.3275594163619875e-06,
+      "loss": 0.0927,
+      "step": 960
+    },
+    {
+      "epoch": 4.387243735763098,
+      "grad_norm": 0.9952108860015869,
+      "learning_rate": 1.2363684000178788e-06,
+      "loss": 0.0951,
+      "step": 965
+    },
+    {
+      "epoch": 4.41002277904328,
+      "grad_norm": 0.8636078834533691,
+      "learning_rate": 1.148287180639952e-06,
+      "loss": 0.1034,
+      "step": 970
+    },
+    {
+      "epoch": 4.432801822323462,
+      "grad_norm": 0.8305910229682922,
+      "learning_rate": 1.0633356595679683e-06,
+      "loss": 0.0812,
+      "step": 975
+    },
+    {
+      "epoch": 4.455580865603645,
+      "grad_norm": 0.9145544767379761,
+      "learning_rate": 9.815330310080889e-07,
+      "loss": 0.0843,
+      "step": 980
+    },
+    {
+      "epoch": 4.478359908883827,
+      "grad_norm": 0.8118730187416077,
+      "learning_rate": 9.028977776960873e-07,
+      "loss": 0.0806,
+      "step": 985
+    },
+    {
+      "epoch": 4.501138952164009,
+      "grad_norm": 0.8564863204956055,
+      "learning_rate": 8.274476667212933e-07,
+      "loss": 0.0953,
+      "step": 990
+    },
+    {
+      "epoch": 4.523917995444191,
+      "grad_norm": 0.9333673119544983,
+      "learning_rate": 7.551997455122573e-07,
+      "loss": 0.0828,
+      "step": 995
+    },
+    {
+      "epoch": 4.546697038724374,
+      "grad_norm": 0.9160373210906982,
+      "learning_rate": 6.861703379849815e-07,
+      "loss": 0.082,
+      "step": 1000
+    },
+    {
+      "epoch": 4.569476082004556,
+      "grad_norm": 0.7882616519927979,
+      "learning_rate": 6.203750408546638e-07,
+      "loss": 0.0871,
+      "step": 1005
+    },
+    {
+      "epoch": 4.592255125284738,
+      "grad_norm": 0.9012757539749146,
+      "learning_rate": 5.578287201117127e-07,
+      "loss": 0.0922,
+      "step": 1010
+    },
+    {
+      "epoch": 4.61503416856492,
+      "grad_norm": 0.9675836563110352,
+      "learning_rate": 4.985455076628915e-07,
+      "loss": 0.0933,
+      "step": 1015
+    },
+    {
+      "epoch": 4.637813211845103,
+      "grad_norm": 0.8786028027534485,
+      "learning_rate": 4.4253879813831166e-07,
+      "loss": 0.0835,
+      "step": 1020
+    },
+    {
+      "epoch": 4.660592255125285,
+      "grad_norm": 0.852175235748291,
+      "learning_rate": 3.8982124586499804e-07,
+      "loss": 0.0872,
+      "step": 1025
+    },
+    {
+      "epoch": 4.683371298405467,
+      "grad_norm": 0.7953546643257141,
+      "learning_rate": 3.404047620077433e-07,
+      "loss": 0.094,
+      "step": 1030
+    },
+    {
+      "epoch": 4.70615034168565,
+      "grad_norm": 0.748761773109436,
+      "learning_rate": 2.943005118778597e-07,
+      "loss": 0.0867,
+      "step": 1035
+    },
+    {
+      "epoch": 4.728929384965832,
+      "grad_norm": 0.8128888607025146,
+      "learning_rate": 2.5151891241045e-07,
+      "loss": 0.0842,
+      "step": 1040
+    },
+    {
+      "epoch": 4.751708428246014,
+      "grad_norm": 0.9451525211334229,
+      "learning_rate": 2.1206962981078216e-07,
+      "loss": 0.0939,
+      "step": 1045
+    },
+    {
+      "epoch": 4.774487471526196,
+      "grad_norm": 0.8104263544082642,
+      "learning_rate": 1.759615773702644e-07,
+      "loss": 0.0757,
+      "step": 1050
+    },
+    {
+      "epoch": 4.7972665148063784,
+      "grad_norm": 0.8320966958999634,
+      "learning_rate": 1.4320291345256308e-07,
+      "loss": 0.0858,
+      "step": 1055
+    },
+    {
+      "epoch": 4.82004555808656,
+      "grad_norm": 0.7932759523391724,
+      "learning_rate": 1.1380103965026867e-07,
+      "loss": 0.0845,
+      "step": 1060
+    },
+    {
+      "epoch": 4.842824601366742,
+      "grad_norm": 0.8291938900947571,
+      "learning_rate": 8.776259911255857e-08,
+      "loss": 0.0768,
+      "step": 1065
+    },
+    {
+      "epoch": 4.865603644646924,
+      "grad_norm": 0.8188297748565674,
+      "learning_rate": 6.509347504422613e-08,
+      "loss": 0.0884,
+      "step": 1070
+    },
+    {
+      "epoch": 4.888382687927107,
+      "grad_norm": 0.8020090460777283,
+      "learning_rate": 4.579878937640725e-08,
+      "loss": 0.0891,
+      "step": 1075
+    },
+    {
+      "epoch": 4.911161731207289,
+      "grad_norm": 0.8145992755889893,
+      "learning_rate": 2.9882901609321054e-08,
+      "loss": 0.0985,
+      "step": 1080
+    },
+    {
+      "epoch": 4.933940774487471,
+      "grad_norm": 0.791728675365448,
+      "learning_rate": 1.7349407827264574e-08,
+      "loss": 0.0818,
+      "step": 1085
+    },
+    {
+      "epoch": 4.956719817767654,
+      "grad_norm": 0.871566116809845,
+      "learning_rate": 8.201139886109265e-09,
+      "loss": 0.085,
+      "step": 1090
+    },
+    {
+      "epoch": 4.979498861047836,
+      "grad_norm": 0.8973204493522644,
+      "learning_rate": 2.440164773457876e-09,
+      "loss": 0.0755,
+      "step": 1095
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 1.0401437282562256,
+      "learning_rate": 6.778414162467162e-11,
+      "loss": 0.079,
+      "step": 1100
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1100,
+  "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": 1.589753555701465e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

36_128_e5_3e-5/checkpoint-1100/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eef332d20abd54b13c49f72e87471df99ec0c4e4488c78179cdbc413b0c8caa1
+size 7736

36_128_e5_3e-5/checkpoint-1100/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)

36_128_e5_3e-5/checkpoint-220/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

36_128_e5_3e-5/checkpoint-220/adapter_config.json

@@ -0,0 +1,39 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "ibm-granite/granite-3.3-8b-base",
+  "bias": "none",
+  "corda_config": null,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 256,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 128,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "up_proj",
+    "gate_proj",
+    "q_proj",
+    "v_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

36_128_e5_3e-5/checkpoint-220/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:048f50c54ec20c10cbc04855d298353267d25258ff86fdddbbb6879a30a05f8e
+size 791751704

36_128_e5_3e-5/checkpoint-220/latest

@@ -0,0 +1 @@
+global_step220

36_128_e5_3e-5/checkpoint-220/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8a824ba00ede0024e51452ccd5bcac2a264625d6be457ae58c76c2673c5c04c1
+size 15920

36_128_e5_3e-5/checkpoint-220/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d73bc7f0181681ee09bf57074ee40b31aa5c1659a994bc6e3aa955da616d4557
+size 15920

36_128_e5_3e-5/checkpoint-220/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5e2727f3ab4de7cb60d3a684099d740dc438dddac02a9622dcc2ba0f5858ffc5
+size 15920

36_128_e5_3e-5/checkpoint-220/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e1e5f7d3f811668960c8694d78fd8a11fa515a0324e0850df3dc2621d89a2157
+size 15920

36_128_e5_3e-5/checkpoint-220/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eab55fece28b09bdd29314effa052e3a34b7957da3c5207172631e459a6a5b23
+size 15920

36_128_e5_3e-5/checkpoint-220/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0401df5d7c0efca5359b011d0fefbfadabe1c56f1c0949b7e388244563f2ea7b
+size 15920

36_128_e5_3e-5/checkpoint-220/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f8a0656cf8ff618f81598c0b5b67ddb46ec35c98ff68dfaa416d2ffe8fef6a75
+size 15920

36_128_e5_3e-5/checkpoint-220/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f493eafb197b99b9ec0c3a520db20edb64365b85cbb586c4ab5e3e1f90df94f1
+size 15920

36_128_e5_3e-5/checkpoint-220/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dd8ce842753bc9ebe4654acc67d9e41796c399080873c1a8f48f589264283fb9
+size 1064

36_128_e5_3e-5/checkpoint-220/special_tokens_map.json

@@ -0,0 +1,45 @@
+{
+  "additional_special_tokens": [
+    "<|endoftext|>",
+    "<fim_prefix>",
+    "<fim_middle>",
+    "<fim_suffix>",
+    "<fim_pad>",
+    "<filename>",
+    "<gh_stars>",
+    "<issue_start>",
+    "<issue_comment>",
+    "<issue_closed>",
+    "<jupyter_start>",
+    "<jupyter_text>",
+    "<jupyter_code>",
+    "<jupyter_output>",
+    "<empty_output>",
+    "<commit_before>",
+    "<commit_msg>",
+    "<commit_after>",
+    "<reponame>"
+  ],
+  "bos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "<reponame>",
+  "unk_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

36_128_e5_3e-5/checkpoint-220/tokenizer_config.json

@@ -0,0 +1,188 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<fim_prefix>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<fim_middle>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<fim_suffix>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<fim_pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "5": {
+      "content": "<filename>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "6": {
+      "content": "<gh_stars>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "7": {
+      "content": "<issue_start>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "8": {
+      "content": "<issue_comment>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "9": {
+      "content": "<issue_closed>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "10": {
+      "content": "<jupyter_start>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "11": {
+      "content": "<jupyter_text>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "12": {
+      "content": "<jupyter_code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "13": {
+      "content": "<jupyter_output>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "14": {
+      "content": "<empty_output>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "15": {
+      "content": "<commit_before>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "16": {
+      "content": "<commit_msg>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "17": {
+      "content": "<commit_after>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "18": {
+      "content": "<reponame>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<|endoftext|>",
+    "<fim_prefix>",
+    "<fim_middle>",
+    "<fim_suffix>",
+    "<fim_pad>",
+    "<filename>",
+    "<gh_stars>",
+    "<issue_start>",
+    "<issue_comment>",
+    "<issue_closed>",
+    "<jupyter_start>",
+    "<jupyter_text>",
+    "<jupyter_code>",
+    "<jupyter_output>",
+    "<empty_output>",
+    "<commit_before>",
+    "<commit_msg>",
+    "<commit_after>",
+    "<reponame>"
+  ],
+  "bos_token": "<|endoftext|>",
+  "clean_up_tokenization_spaces": true,
+  "eos_token": "<|endoftext|>",
+  "extra_special_tokens": {},
+  "model_max_length": 8192,
+  "pad_token": "<reponame>",
+  "padding_side": "left",
+  "tokenizer_class": "GPT2Tokenizer",
+  "unk_token": "<|endoftext|>",
+  "vocab_size": 49152
+}

36_128_e5_3e-5/checkpoint-220/trainer_state.json

@@ -0,0 +1,342 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 220,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.022779043280182234,
+      "grad_norm": 1.0761960744857788,
+      "learning_rate": 2.181818181818182e-06,
+      "loss": 1.4131,
+      "step": 5
+    },
+    {
+      "epoch": 0.04555808656036447,
+      "grad_norm": 0.7671697735786438,
+      "learning_rate": 4.90909090909091e-06,
+      "loss": 1.3993,
+      "step": 10
+    },
+    {
+      "epoch": 0.0683371298405467,
+      "grad_norm": 0.5325023531913757,
+      "learning_rate": 7.636363636363636e-06,
+      "loss": 1.3575,
+      "step": 15
+    },
+    {
+      "epoch": 0.09111617312072894,
+      "grad_norm": 0.5349851846694946,
+      "learning_rate": 1.0363636363636364e-05,
+      "loss": 1.3727,
+      "step": 20
+    },
+    {
+      "epoch": 0.11389521640091116,
+      "grad_norm": 0.5694962739944458,
+      "learning_rate": 1.309090909090909e-05,
+      "loss": 1.3057,
+      "step": 25
+    },
+    {
+      "epoch": 0.1366742596810934,
+      "grad_norm": 0.541211724281311,
+      "learning_rate": 1.5818181818181818e-05,
+      "loss": 1.2344,
+      "step": 30
+    },
+    {
+      "epoch": 0.15945330296127563,
+      "grad_norm": 0.5115331411361694,
+      "learning_rate": 1.8545454545454545e-05,
+      "loss": 1.3365,
+      "step": 35
+    },
+    {
+      "epoch": 0.18223234624145787,
+      "grad_norm": 0.46041619777679443,
+      "learning_rate": 2.1272727272727273e-05,
+      "loss": 1.2517,
+      "step": 40
+    },
+    {
+      "epoch": 0.20501138952164008,
+      "grad_norm": 0.4588950276374817,
+      "learning_rate": 2.4e-05,
+      "loss": 1.233,
+      "step": 45
+    },
+    {
+      "epoch": 0.22779043280182232,
+      "grad_norm": 0.571067750453949,
+      "learning_rate": 2.6727272727272728e-05,
+      "loss": 1.1614,
+      "step": 50
+    },
+    {
+      "epoch": 0.2505694760820046,
+      "grad_norm": 0.47105151414871216,
+      "learning_rate": 2.9454545454545456e-05,
+      "loss": 1.1789,
+      "step": 55
+    },
+    {
+      "epoch": 0.2733485193621868,
+      "grad_norm": 0.5167170763015747,
+      "learning_rate": 2.9998915465986464e-05,
+      "loss": 1.1674,
+      "step": 60
+    },
+    {
+      "epoch": 0.296127562642369,
+      "grad_norm": 0.5363845229148865,
+      "learning_rate": 2.999450981533838e-05,
+      "loss": 1.1646,
+      "step": 65
+    },
+    {
+      "epoch": 0.31890660592255127,
+      "grad_norm": 0.5026624798774719,
+      "learning_rate": 2.9986716259340288e-05,
+      "loss": 1.1268,
+      "step": 70
+    },
+    {
+      "epoch": 0.3416856492027335,
+      "grad_norm": 0.46400442719459534,
+      "learning_rate": 2.9975536558892034e-05,
+      "loss": 1.0501,
+      "step": 75
+    },
+    {
+      "epoch": 0.36446469248291574,
+      "grad_norm": 0.6948593258857727,
+      "learning_rate": 2.9960973239969295e-05,
+      "loss": 1.0567,
+      "step": 80
+    },
+    {
+      "epoch": 0.38724373576309795,
+      "grad_norm": 0.6658697724342346,
+      "learning_rate": 2.9943029593052822e-05,
+      "loss": 1.1199,
+      "step": 85
+    },
+    {
+      "epoch": 0.41002277904328016,
+      "grad_norm": 0.5796182155609131,
+      "learning_rate": 2.992170967238502e-05,
+      "loss": 1.0407,
+      "step": 90
+    },
+    {
+      "epoch": 0.4328018223234624,
+      "grad_norm": 0.5862748622894287,
+      "learning_rate": 2.9897018295053883e-05,
+      "loss": 1.0524,
+      "step": 95
+    },
+    {
+      "epoch": 0.45558086560364464,
+      "grad_norm": 0.714149534702301,
+      "learning_rate": 2.9868961039904628e-05,
+      "loss": 1.0367,
+      "step": 100
+    },
+    {
+      "epoch": 0.4783599088838269,
+      "grad_norm": 0.7176491618156433,
+      "learning_rate": 2.983754424627919e-05,
+      "loss": 1.0013,
+      "step": 105
+    },
+    {
+      "epoch": 0.5011389521640092,
+      "grad_norm": 0.6981073617935181,
+      "learning_rate": 2.9802775012583884e-05,
+      "loss": 1.0079,
+      "step": 110
+    },
+    {
+      "epoch": 0.5239179954441914,
+      "grad_norm": 0.5984258651733398,
+      "learning_rate": 2.9764661194685583e-05,
+      "loss": 0.975,
+      "step": 115
+    },
+    {
+      "epoch": 0.5466970387243736,
+      "grad_norm": 0.7986994981765747,
+      "learning_rate": 2.972321140413672e-05,
+      "loss": 0.9417,
+      "step": 120
+    },
+    {
+      "epoch": 0.5694760820045558,
+      "grad_norm": 0.6552079916000366,
+      "learning_rate": 2.9678435006229585e-05,
+      "loss": 0.9208,
+      "step": 125
+    },
+    {
+      "epoch": 0.592255125284738,
+      "grad_norm": 0.7615071535110474,
+      "learning_rate": 2.9630342117880293e-05,
+      "loss": 0.9208,
+      "step": 130
+    },
+    {
+      "epoch": 0.6150341685649203,
+      "grad_norm": 0.73598313331604,
+      "learning_rate": 2.957894360534295e-05,
+      "loss": 0.9459,
+      "step": 135
+    },
+    {
+      "epoch": 0.6378132118451025,
+      "grad_norm": 0.9029515981674194,
+      "learning_rate": 2.9524251081754475e-05,
+      "loss": 0.9338,
+      "step": 140
+    },
+    {
+      "epoch": 0.6605922551252847,
+      "grad_norm": 0.8563511967658997,
+      "learning_rate": 2.9466276904510713e-05,
+      "loss": 0.8437,
+      "step": 145
+    },
+    {
+      "epoch": 0.683371298405467,
+      "grad_norm": 0.8966438174247742,
+      "learning_rate": 2.9405034172474363e-05,
+      "loss": 0.8995,
+      "step": 150
+    },
+    {
+      "epoch": 0.7061503416856492,
+      "grad_norm": 0.9085351228713989,
+      "learning_rate": 2.9340536723015367e-05,
+      "loss": 0.8988,
+      "step": 155
+    },
+    {
+      "epoch": 0.7289293849658315,
+      "grad_norm": 0.920749843120575,
+      "learning_rate": 2.927279912888447e-05,
+      "loss": 0.8457,
+      "step": 160
+    },
+    {
+      "epoch": 0.7517084282460137,
+      "grad_norm": 0.844048023223877,
+      "learning_rate": 2.920183669492061e-05,
+      "loss": 0.8699,
+      "step": 165
+    },
+    {
+      "epoch": 0.7744874715261959,
+      "grad_norm": 1.1362037658691406,
+      "learning_rate": 2.9127665454592872e-05,
+      "loss": 0.8561,
+      "step": 170
+    },
+    {
+      "epoch": 0.7972665148063781,
+      "grad_norm": 0.9490585923194885,
+      "learning_rate": 2.9050302166377858e-05,
+      "loss": 0.8016,
+      "step": 175
+    },
+    {
+      "epoch": 0.8200455580865603,
+      "grad_norm": 1.047890543937683,
+      "learning_rate": 2.896976430997323e-05,
+      "loss": 0.8047,
+      "step": 180
+    },
+    {
+      "epoch": 0.8428246013667426,
+      "grad_norm": 1.252584457397461,
+      "learning_rate": 2.8886070082348268e-05,
+      "loss": 0.8087,
+      "step": 185
+    },
+    {
+      "epoch": 0.8656036446469249,
+      "grad_norm": 1.0744786262512207,
+      "learning_rate": 2.879923839363242e-05,
+      "loss": 0.7823,
+      "step": 190
+    },
+    {
+      "epoch": 0.8883826879271071,
+      "grad_norm": 0.9743952751159668,
+      "learning_rate": 2.870928886284267e-05,
+      "loss": 0.727,
+      "step": 195
+    },
+    {
+      "epoch": 0.9111617312072893,
+      "grad_norm": 1.029543399810791,
+      "learning_rate": 2.8616241813450755e-05,
+      "loss": 0.7648,
+      "step": 200
+    },
+    {
+      "epoch": 0.9339407744874715,
+      "grad_norm": 1.031134009361267,
+      "learning_rate": 2.852011826879125e-05,
+      "loss": 0.7444,
+      "step": 205
+    },
+    {
+      "epoch": 0.9567198177676538,
+      "grad_norm": 1.1572555303573608,
+      "learning_rate": 2.8420939947311454e-05,
+      "loss": 0.7408,
+      "step": 210
+    },
+    {
+      "epoch": 0.979498861047836,
+      "grad_norm": 1.0397342443466187,
+      "learning_rate": 2.8318729257664265e-05,
+      "loss": 0.7015,
+      "step": 215
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.4852145910263062,
+      "learning_rate": 2.821350929364512e-05,
+      "loss": 0.6917,
+      "step": 220
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1100,
+  "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": 3.180798646723543e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

36_128_e5_3e-5/checkpoint-220/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eef332d20abd54b13c49f72e87471df99ec0c4e4488c78179cdbc413b0c8caa1
+size 7736

36_128_e5_3e-5/checkpoint-220/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)

36_128_e5_3e-5/checkpoint-440/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

36_128_e5_3e-5/checkpoint-440/adapter_config.json

@@ -0,0 +1,39 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "ibm-granite/granite-3.3-8b-base",
+  "bias": "none",
+  "corda_config": null,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 256,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 128,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "up_proj",
+    "gate_proj",
+    "q_proj",
+    "v_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

36_128_e5_3e-5/checkpoint-440/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b95236b849aff24c1febbb26b865daf528b4f2f57a33c9caa4c0e1ec4f8341e2
+size 791751704

36_128_e5_3e-5/checkpoint-440/latest

@@ -0,0 +1 @@
+global_step440

36_128_e5_3e-5/checkpoint-440/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ef1feddc4db058cbdf3d5ebfd20d6be6f035915e11f78b5faf32d5041f1b786f
+size 15920

36_128_e5_3e-5/checkpoint-440/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:53d312b23be06ab051e9bbc9bf82de8396551fc12b7cc5e055dc260be4dcbeac
+size 15920

36_128_e5_3e-5/checkpoint-440/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b44fbf5c0c44be7acfa9455c80c2fd9e373c959f3fd67cff7fc807d94494d8e
+size 15920