Upload folder using huggingface_hub

45_128_e5_3e-5/checkpoint-1175/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

45_128_e5_3e-5/checkpoint-1175/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": [
+    "q_proj",
+    "down_proj",
+    "v_proj",
+    "k_proj",
+    "up_proj",
+    "o_proj",
+    "gate_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

45_128_e5_3e-5/checkpoint-1175/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9459ccee9dd9d16c70878938573abfebc900a26639c6f4d1a07fe14807398f64
+size 791751704

45_128_e5_3e-5/checkpoint-1175/latest

@@ -0,0 +1 @@
+global_step1175

45_128_e5_3e-5/checkpoint-1175/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f030986b7a143218ea16999087fc3416d4613f8bf37360491ba88c2ff9a7661f
+size 15920

45_128_e5_3e-5/checkpoint-1175/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:753331904d4e557e57aa7bf1d2caffad351ac8401b8f070c7c75da1228368bcf
+size 15920

45_128_e5_3e-5/checkpoint-1175/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f401d09c237fb75abf09135f3a0bbef3134142a863c9df643b2d3eed154709d5
+size 15920

45_128_e5_3e-5/checkpoint-1175/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b5010990215fc14c214454b5086b3f0c7e2b7f4fcc316c2a68471cdb67531b93
+size 15920

45_128_e5_3e-5/checkpoint-1175/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5517274ab27cad5ffe32c3c51d4828156026e56147d86eba78216e02829fe53e
+size 15920

45_128_e5_3e-5/checkpoint-1175/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5b113669c89ab51ca329e9b96ab3d8afa0a0fe40cbdceb417036726aae4b018c
+size 15920

45_128_e5_3e-5/checkpoint-1175/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b0362d2f2d25adfcd2f1c1d48ae2616daa9d97d4d62791a20ba6c122ab4ba5f8
+size 15920

45_128_e5_3e-5/checkpoint-1175/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8b2561866d0e4937f269d7bd0e5ec56801d2a683bed0196a2ac89505de3822ae
+size 15920

45_128_e5_3e-5/checkpoint-1175/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:82beb172afa396f877f7d443492eeb3f29023503e8776211572280a5a9c6614e
+size 1064

45_128_e5_3e-5/checkpoint-1175/special_tokens_map.json

@@ -0,0 +1,51 @@
+{
+  "additional_special_tokens": [
+    "<|endoftext|>",
+    "<fim_prefix>",
+    "<fim_middle>",
+    "<fim_suffix>",
+    "<fim_pad>",
+    "<filename>",
+    "<gh_stars>",
+    "<issue_start>",
+    "<issue_comment>",
+    "<issue_closed>",
+    "<jupyter_start>",
+    "<jupyter_text>",
+    "<jupyter_code>",
+    "<jupyter_output>",
+    "<empty_output>",
+    "<commit_before>",
+    "<commit_msg>",
+    "<commit_after>",
+    "<reponame>"
+  ],
+  "bos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

45_128_e5_3e-5/checkpoint-1175/tokenizer_config.json

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

45_128_e5_3e-5/checkpoint-1175/trainer_state.json

@@ -0,0 +1,1679 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 1175,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.021321961620469083,
+      "grad_norm": 1.1025959253311157,
+      "learning_rate": 2.033898305084746e-06,
+      "loss": 1.3396,
+      "step": 5
+    },
+    {
+      "epoch": 0.042643923240938165,
+      "grad_norm": 0.9394682049751282,
+      "learning_rate": 4.576271186440678e-06,
+      "loss": 1.3427,
+      "step": 10
+    },
+    {
+      "epoch": 0.06396588486140725,
+      "grad_norm": 0.5853310823440552,
+      "learning_rate": 7.1186440677966106e-06,
+      "loss": 1.3113,
+      "step": 15
+    },
+    {
+      "epoch": 0.08528784648187633,
+      "grad_norm": 1.0313405990600586,
+      "learning_rate": 9.661016949152542e-06,
+      "loss": 1.2611,
+      "step": 20
+    },
+    {
+      "epoch": 0.10660980810234541,
+      "grad_norm": 0.548238217830658,
+      "learning_rate": 1.2203389830508475e-05,
+      "loss": 1.2482,
+      "step": 25
+    },
+    {
+      "epoch": 0.1279317697228145,
+      "grad_norm": 0.6522016525268555,
+      "learning_rate": 1.4745762711864408e-05,
+      "loss": 1.2438,
+      "step": 30
+    },
+    {
+      "epoch": 0.14925373134328357,
+      "grad_norm": 0.6198495626449585,
+      "learning_rate": 1.728813559322034e-05,
+      "loss": 1.1851,
+      "step": 35
+    },
+    {
+      "epoch": 0.17057569296375266,
+      "grad_norm": 0.5091772079467773,
+      "learning_rate": 1.983050847457627e-05,
+      "loss": 1.2328,
+      "step": 40
+    },
+    {
+      "epoch": 0.19189765458422176,
+      "grad_norm": 0.43554070591926575,
+      "learning_rate": 2.2372881355932205e-05,
+      "loss": 1.2056,
+      "step": 45
+    },
+    {
+      "epoch": 0.21321961620469082,
+      "grad_norm": 0.4790154695510864,
+      "learning_rate": 2.4915254237288138e-05,
+      "loss": 1.1369,
+      "step": 50
+    },
+    {
+      "epoch": 0.2345415778251599,
+      "grad_norm": 0.5797903537750244,
+      "learning_rate": 2.7457627118644068e-05,
+      "loss": 1.1525,
+      "step": 55
+    },
+    {
+      "epoch": 0.255863539445629,
+      "grad_norm": 0.6414856314659119,
+      "learning_rate": 3e-05,
+      "loss": 1.118,
+      "step": 60
+    },
+    {
+      "epoch": 0.2771855010660981,
+      "grad_norm": 0.6532033085823059,
+      "learning_rate": 2.9998514182537154e-05,
+      "loss": 1.0826,
+      "step": 65
+    },
+    {
+      "epoch": 0.29850746268656714,
+      "grad_norm": 0.5140215158462524,
+      "learning_rate": 2.9994057024502427e-05,
+      "loss": 1.0808,
+      "step": 70
+    },
+    {
+      "epoch": 0.31982942430703626,
+      "grad_norm": 0.5573478937149048,
+      "learning_rate": 2.998662940889891e-05,
+      "loss": 1.0923,
+      "step": 75
+    },
+    {
+      "epoch": 0.3411513859275053,
+      "grad_norm": 0.6022165417671204,
+      "learning_rate": 2.9976232807204073e-05,
+      "loss": 1.044,
+      "step": 80
+    },
+    {
+      "epoch": 0.3624733475479744,
+      "grad_norm": 0.6599242687225342,
+      "learning_rate": 2.9962869279078226e-05,
+      "loss": 1.0641,
+      "step": 85
+    },
+    {
+      "epoch": 0.3837953091684435,
+      "grad_norm": 0.5529143810272217,
+      "learning_rate": 2.9946541471956496e-05,
+      "loss": 1.0251,
+      "step": 90
+    },
+    {
+      "epoch": 0.4051172707889126,
+      "grad_norm": 0.5225338339805603,
+      "learning_rate": 2.9927252620524346e-05,
+      "loss": 1.0353,
+      "step": 95
+    },
+    {
+      "epoch": 0.42643923240938164,
+      "grad_norm": 0.5583758354187012,
+      "learning_rate": 2.9905006546076746e-05,
+      "loss": 1.0061,
+      "step": 100
+    },
+    {
+      "epoch": 0.44776119402985076,
+      "grad_norm": 0.5726757645606995,
+      "learning_rate": 2.9879807655761145e-05,
+      "loss": 1.0173,
+      "step": 105
+    },
+    {
+      "epoch": 0.4690831556503198,
+      "grad_norm": 0.6370916366577148,
+      "learning_rate": 2.985166094170439e-05,
+      "loss": 1.0122,
+      "step": 110
+    },
+    {
+      "epoch": 0.4904051172707889,
+      "grad_norm": 0.6529775857925415,
+      "learning_rate": 2.9820571980023704e-05,
+      "loss": 0.911,
+      "step": 115
+    },
+    {
+      "epoch": 0.511727078891258,
+      "grad_norm": 0.6832720637321472,
+      "learning_rate": 2.9786546929722055e-05,
+      "loss": 0.9745,
+      "step": 120
+    },
+    {
+      "epoch": 0.5330490405117271,
+      "grad_norm": 0.6681720614433289,
+      "learning_rate": 2.974959253146796e-05,
+      "loss": 0.9218,
+      "step": 125
+    },
+    {
+      "epoch": 0.5543710021321961,
+      "grad_norm": 0.8444278836250305,
+      "learning_rate": 2.9709716106260115e-05,
+      "loss": 0.9233,
+      "step": 130
+    },
+    {
+      "epoch": 0.5756929637526652,
+      "grad_norm": 0.8203223347663879,
+      "learning_rate": 2.966692555397705e-05,
+      "loss": 0.9167,
+      "step": 135
+    },
+    {
+      "epoch": 0.5970149253731343,
+      "grad_norm": 0.6752311587333679,
+      "learning_rate": 2.962122935181207e-05,
+      "loss": 0.9187,
+      "step": 140
+    },
+    {
+      "epoch": 0.6183368869936035,
+      "grad_norm": 0.6852523684501648,
+      "learning_rate": 2.957263655259387e-05,
+      "loss": 0.8627,
+      "step": 145
+    },
+    {
+      "epoch": 0.6396588486140725,
+      "grad_norm": 0.8514593839645386,
+      "learning_rate": 2.9521156782993066e-05,
+      "loss": 0.8204,
+      "step": 150
+    },
+    {
+      "epoch": 0.6609808102345416,
+      "grad_norm": 0.8665833473205566,
+      "learning_rate": 2.9466800241615075e-05,
+      "loss": 0.8655,
+      "step": 155
+    },
+    {
+      "epoch": 0.6823027718550106,
+      "grad_norm": 0.9081581234931946,
+      "learning_rate": 2.940957769697969e-05,
+      "loss": 0.8356,
+      "step": 160
+    },
+    {
+      "epoch": 0.7036247334754797,
+      "grad_norm": 0.7875135540962219,
+      "learning_rate": 2.9349500485387718e-05,
+      "loss": 0.8547,
+      "step": 165
+    },
+    {
+      "epoch": 0.7249466950959488,
+      "grad_norm": 0.837771475315094,
+      "learning_rate": 2.9286580508675174e-05,
+      "loss": 0.7974,
+      "step": 170
+    },
+    {
+      "epoch": 0.746268656716418,
+      "grad_norm": 0.8674811720848083,
+      "learning_rate": 2.9220830231855417e-05,
+      "loss": 0.8343,
+      "step": 175
+    },
+    {
+      "epoch": 0.767590618336887,
+      "grad_norm": 0.8112357258796692,
+      "learning_rate": 2.9152262680649704e-05,
+      "loss": 0.7642,
+      "step": 180
+    },
+    {
+      "epoch": 0.7889125799573561,
+      "grad_norm": 0.855670690536499,
+      "learning_rate": 2.90808914389067e-05,
+      "loss": 0.778,
+      "step": 185
+    },
+    {
+      "epoch": 0.8102345415778252,
+      "grad_norm": 0.9481444358825684,
+      "learning_rate": 2.900673064591139e-05,
+      "loss": 0.7688,
+      "step": 190
+    },
+    {
+      "epoch": 0.8315565031982942,
+      "grad_norm": 0.9807798266410828,
+      "learning_rate": 2.8929794993583937e-05,
+      "loss": 0.7645,
+      "step": 195
+    },
+    {
+      "epoch": 0.8528784648187633,
+      "grad_norm": 1.074795126914978,
+      "learning_rate": 2.8850099723569104e-05,
+      "loss": 0.7552,
+      "step": 200
+    },
+    {
+      "epoch": 0.8742004264392325,
+      "grad_norm": 1.0988177061080933,
+      "learning_rate": 2.876766062421675e-05,
+      "loss": 0.7782,
+      "step": 205
+    },
+    {
+      "epoch": 0.8955223880597015,
+      "grad_norm": 1.0309362411499023,
+      "learning_rate": 2.8682494027454e-05,
+      "loss": 0.7621,
+      "step": 210
+    },
+    {
+      "epoch": 0.9168443496801706,
+      "grad_norm": 0.9497408866882324,
+      "learning_rate": 2.8594616805549752e-05,
+      "loss": 0.6876,
+      "step": 215
+    },
+    {
+      "epoch": 0.9381663113006397,
+      "grad_norm": 1.0013132095336914,
+      "learning_rate": 2.8504046367772117e-05,
+      "loss": 0.6893,
+      "step": 220
+    },
+    {
+      "epoch": 0.9594882729211087,
+      "grad_norm": 1.0180565118789673,
+      "learning_rate": 2.8410800656939512e-05,
+      "loss": 0.6678,
+      "step": 225
+    },
+    {
+      "epoch": 0.9808102345415778,
+      "grad_norm": 1.1235246658325195,
+      "learning_rate": 2.8314898145865996e-05,
+      "loss": 0.6193,
+      "step": 230
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.4771363735198975,
+      "learning_rate": 2.8216357833701667e-05,
+      "loss": 0.6586,
+      "step": 235
+    },
+    {
+      "epoch": 1.0213219616204692,
+      "grad_norm": 1.1948978900909424,
+      "learning_rate": 2.811519924216873e-05,
+      "loss": 0.579,
+      "step": 240
+    },
+    {
+      "epoch": 1.0426439232409381,
+      "grad_norm": 1.012829303741455,
+      "learning_rate": 2.8011442411694105e-05,
+      "loss": 0.5649,
+      "step": 245
+    },
+    {
+      "epoch": 1.0639658848614073,
+      "grad_norm": 1.0202640295028687,
+      "learning_rate": 2.79051078974392e-05,
+      "loss": 0.5759,
+      "step": 250
+    },
+    {
+      "epoch": 1.0852878464818763,
+      "grad_norm": 1.1689255237579346,
+      "learning_rate": 2.779621676522777e-05,
+      "loss": 0.5548,
+      "step": 255
+    },
+    {
+      "epoch": 1.1066098081023454,
+      "grad_norm": 1.0039416551589966,
+      "learning_rate": 2.7684790587372597e-05,
+      "loss": 0.5409,
+      "step": 260
+    },
+    {
+      "epoch": 1.1279317697228146,
+      "grad_norm": 1.0827348232269287,
+      "learning_rate": 2.757085143840179e-05,
+      "loss": 0.5911,
+      "step": 265
+    },
+    {
+      "epoch": 1.1492537313432836,
+      "grad_norm": 1.0059739351272583,
+      "learning_rate": 2.7454421890685647e-05,
+      "loss": 0.5258,
+      "step": 270
+    },
+    {
+      "epoch": 1.1705756929637527,
+      "grad_norm": 1.232321858406067,
+      "learning_rate": 2.7335525009964863e-05,
+      "loss": 0.5129,
+      "step": 275
+    },
+    {
+      "epoch": 1.1918976545842217,
+      "grad_norm": 1.052078127861023,
+      "learning_rate": 2.721418435078099e-05,
+      "loss": 0.5184,
+      "step": 280
+    },
+    {
+      "epoch": 1.2132196162046909,
+      "grad_norm": 1.2947639226913452,
+      "learning_rate": 2.709042395181008e-05,
+      "loss": 0.5388,
+      "step": 285
+    },
+    {
+      "epoch": 1.2345415778251598,
+      "grad_norm": 1.1052894592285156,
+      "learning_rate": 2.6964268331100396e-05,
+      "loss": 0.5633,
+      "step": 290
+    },
+    {
+      "epoch": 1.255863539445629,
+      "grad_norm": 1.0759986639022827,
+      "learning_rate": 2.683574248121517e-05,
+      "loss": 0.5133,
+      "step": 295
+    },
+    {
+      "epoch": 1.2771855010660982,
+      "grad_norm": 1.1331149339675903,
+      "learning_rate": 2.6704871864281377e-05,
+      "loss": 0.55,
+      "step": 300
+    },
+    {
+      "epoch": 1.2985074626865671,
+      "grad_norm": 1.071257472038269,
+      "learning_rate": 2.657168240694541e-05,
+      "loss": 0.5122,
+      "step": 305
+    },
+    {
+      "epoch": 1.3198294243070363,
+      "grad_norm": 1.0727146863937378,
+      "learning_rate": 2.6436200495236806e-05,
+      "loss": 0.5708,
+      "step": 310
+    },
+    {
+      "epoch": 1.3411513859275053,
+      "grad_norm": 1.2485096454620361,
+      "learning_rate": 2.6298452969340952e-05,
+      "loss": 0.5605,
+      "step": 315
+    },
+    {
+      "epoch": 1.3624733475479744,
+      "grad_norm": 1.1561352014541626,
+      "learning_rate": 2.6158467118281765e-05,
+      "loss": 0.4775,
+      "step": 320
+    },
+    {
+      "epoch": 1.3837953091684434,
+      "grad_norm": 1.13988196849823,
+      "learning_rate": 2.6016270674515526e-05,
+      "loss": 0.5213,
+      "step": 325
+    },
+    {
+      "epoch": 1.4051172707889126,
+      "grad_norm": 1.4094964265823364,
+      "learning_rate": 2.58718918084368e-05,
+      "loss": 0.4526,
+      "step": 330
+    },
+    {
+      "epoch": 1.4264392324093818,
+      "grad_norm": 1.132122278213501,
+      "learning_rate": 2.5725359122797657e-05,
+      "loss": 0.5135,
+      "step": 335
+    },
+    {
+      "epoch": 1.4477611940298507,
+      "grad_norm": 1.223320484161377,
+      "learning_rate": 2.5576701647041192e-05,
+      "loss": 0.4754,
+      "step": 340
+    },
+    {
+      "epoch": 1.4690831556503199,
+      "grad_norm": 1.1605335474014282,
+      "learning_rate": 2.5425948831550528e-05,
+      "loss": 0.4845,
+      "step": 345
+    },
+    {
+      "epoch": 1.4904051172707888,
+      "grad_norm": 1.0932228565216064,
+      "learning_rate": 2.5273130541814446e-05,
+      "loss": 0.4388,
+      "step": 350
+    },
+    {
+      "epoch": 1.511727078891258,
+      "grad_norm": 1.064093828201294,
+      "learning_rate": 2.511827705251075e-05,
+      "loss": 0.4424,
+      "step": 355
+    },
+    {
+      "epoch": 1.533049040511727,
+      "grad_norm": 1.204820156097412,
+      "learning_rate": 2.496141904150859e-05,
+      "loss": 0.4394,
+      "step": 360
+    },
+    {
+      "epoch": 1.5543710021321961,
+      "grad_norm": 1.1509909629821777,
+      "learning_rate": 2.4802587583790875e-05,
+      "loss": 0.4334,
+      "step": 365
+    },
+    {
+      "epoch": 1.5756929637526653,
+      "grad_norm": 1.1247512102127075,
+      "learning_rate": 2.464181414529809e-05,
+      "loss": 0.4587,
+      "step": 370
+    },
+    {
+      "epoch": 1.5970149253731343,
+      "grad_norm": 1.1641768217086792,
+      "learning_rate": 2.447913057669456e-05,
+      "loss": 0.4453,
+      "step": 375
+    },
+    {
+      "epoch": 1.6183368869936035,
+      "grad_norm": 1.1374614238739014,
+      "learning_rate": 2.4314569107058572e-05,
+      "loss": 0.4602,
+      "step": 380
+    },
+    {
+      "epoch": 1.6396588486140726,
+      "grad_norm": 1.2946219444274902,
+      "learning_rate": 2.4148162337497496e-05,
+      "loss": 0.3953,
+      "step": 385
+    },
+    {
+      "epoch": 1.6609808102345416,
+      "grad_norm": 1.1880476474761963,
+      "learning_rate": 2.3979943234689226e-05,
+      "loss": 0.4298,
+      "step": 390
+    },
+    {
+      "epoch": 1.6823027718550105,
+      "grad_norm": 1.429024338722229,
+      "learning_rate": 2.3809945124351162e-05,
+      "loss": 0.4449,
+      "step": 395
+    },
+    {
+      "epoch": 1.7036247334754797,
+      "grad_norm": 1.3434990644454956,
+      "learning_rate": 2.36382016846381e-05,
+      "loss": 0.4315,
+      "step": 400
+    },
+    {
+      "epoch": 1.724946695095949,
+      "grad_norm": 1.0165458917617798,
+      "learning_rate": 2.3464746939470288e-05,
+      "loss": 0.4226,
+      "step": 405
+    },
+    {
+      "epoch": 1.7462686567164178,
+      "grad_norm": 1.1595861911773682,
+      "learning_rate": 2.3289615251792984e-05,
+      "loss": 0.3784,
+      "step": 410
+    },
+    {
+      "epoch": 1.767590618336887,
+      "grad_norm": 1.1516300439834595,
+      "learning_rate": 2.3112841316768827e-05,
+      "loss": 0.3978,
+      "step": 415
+    },
+    {
+      "epoch": 1.7889125799573562,
+      "grad_norm": 1.2137658596038818,
+      "learning_rate": 2.2934460154904436e-05,
+      "loss": 0.3699,
+      "step": 420
+    },
+    {
+      "epoch": 1.8102345415778252,
+      "grad_norm": 1.3457313776016235,
+      "learning_rate": 2.2754507105112526e-05,
+      "loss": 0.3552,
+      "step": 425
+    },
+    {
+      "epoch": 1.831556503198294,
+      "grad_norm": 1.3003560304641724,
+      "learning_rate": 2.257301781771095e-05,
+      "loss": 0.3705,
+      "step": 430
+    },
+    {
+      "epoch": 1.8528784648187633,
+      "grad_norm": 1.2907923460006714,
+      "learning_rate": 2.2390028247360042e-05,
+      "loss": 0.3763,
+      "step": 435
+    },
+    {
+      "epoch": 1.8742004264392325,
+      "grad_norm": 1.1227895021438599,
+      "learning_rate": 2.2205574645939684e-05,
+      "loss": 0.3499,
+      "step": 440
+    },
+    {
+      "epoch": 1.8955223880597014,
+      "grad_norm": 1.2199273109436035,
+      "learning_rate": 2.201969355536749e-05,
+      "loss": 0.3956,
+      "step": 445
+    },
+    {
+      "epoch": 1.9168443496801706,
+      "grad_norm": 1.1481236219406128,
+      "learning_rate": 2.183242180035951e-05,
+      "loss": 0.3847,
+      "step": 450
+    },
+    {
+      "epoch": 1.9381663113006398,
+      "grad_norm": 1.1786848306655884,
+      "learning_rate": 2.1643796481134934e-05,
+      "loss": 0.3925,
+      "step": 455
+    },
+    {
+      "epoch": 1.9594882729211087,
+      "grad_norm": 1.4188814163208008,
+      "learning_rate": 2.145385496606619e-05,
+      "loss": 0.3548,
+      "step": 460
+    },
+    {
+      "epoch": 1.9808102345415777,
+      "grad_norm": 1.1952940225601196,
+      "learning_rate": 2.1262634884275948e-05,
+      "loss": 0.3747,
+      "step": 465
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.786958932876587,
+      "learning_rate": 2.107017411818244e-05,
+      "loss": 0.3113,
+      "step": 470
+    },
+    {
+      "epoch": 2.021321961620469,
+      "grad_norm": 1.0472403764724731,
+      "learning_rate": 2.0876510795994628e-05,
+      "loss": 0.292,
+      "step": 475
+    },
+    {
+      "epoch": 2.0426439232409384,
+      "grad_norm": 1.2057167291641235,
+      "learning_rate": 2.068168328415864e-05,
+      "loss": 0.2575,
+      "step": 480
+    },
+    {
+      "epoch": 2.0639658848614073,
+      "grad_norm": 1.2329800128936768,
+      "learning_rate": 2.048573017975706e-05,
+      "loss": 0.28,
+      "step": 485
+    },
+    {
+      "epoch": 2.0852878464818763,
+      "grad_norm": 1.4483537673950195,
+      "learning_rate": 2.028869030286248e-05,
+      "loss": 0.2503,
+      "step": 490
+    },
+    {
+      "epoch": 2.106609808102345,
+      "grad_norm": 1.2167725563049316,
+      "learning_rate": 2.0090602688846884e-05,
+      "loss": 0.2692,
+      "step": 495
+    },
+    {
+      "epoch": 2.1279317697228146,
+      "grad_norm": 1.2569903135299683,
+      "learning_rate": 1.9891506580648425e-05,
+      "loss": 0.2976,
+      "step": 500
+    },
+    {
+      "epoch": 2.1492537313432836,
+      "grad_norm": 1.172814965248108,
+      "learning_rate": 1.9691441420997016e-05,
+      "loss": 0.268,
+      "step": 505
+    },
+    {
+      "epoch": 2.1705756929637525,
+      "grad_norm": 1.300709843635559,
+      "learning_rate": 1.9490446844600375e-05,
+      "loss": 0.2736,
+      "step": 510
+    },
+    {
+      "epoch": 2.191897654584222,
+      "grad_norm": 1.0922170877456665,
+      "learning_rate": 1.9288562670292038e-05,
+      "loss": 0.2674,
+      "step": 515
+    },
+    {
+      "epoch": 2.213219616204691,
+      "grad_norm": 1.196506381034851,
+      "learning_rate": 1.90858288931429e-05,
+      "loss": 0.2727,
+      "step": 520
+    },
+    {
+      "epoch": 2.23454157782516,
+      "grad_norm": 1.2611831426620483,
+      "learning_rate": 1.888228567653781e-05,
+      "loss": 0.239,
+      "step": 525
+    },
+    {
+      "epoch": 2.2558635394456292,
+      "grad_norm": 1.1467292308807373,
+      "learning_rate": 1.8677973344218864e-05,
+      "loss": 0.2495,
+      "step": 530
+    },
+    {
+      "epoch": 2.277185501066098,
+      "grad_norm": 1.273137092590332,
+      "learning_rate": 1.8472932372296887e-05,
+      "loss": 0.2461,
+      "step": 535
+    },
+    {
+      "epoch": 2.298507462686567,
+      "grad_norm": 1.2586289644241333,
+      "learning_rate": 1.8267203381232774e-05,
+      "loss": 0.2663,
+      "step": 540
+    },
+    {
+      "epoch": 2.319829424307036,
+      "grad_norm": 1.1329915523529053,
+      "learning_rate": 1.8060827127790193e-05,
+      "loss": 0.2577,
+      "step": 545
+    },
+    {
+      "epoch": 2.3411513859275055,
+      "grad_norm": 1.5486836433410645,
+      "learning_rate": 1.7853844496961317e-05,
+      "loss": 0.2373,
+      "step": 550
+    },
+    {
+      "epoch": 2.3624733475479744,
+      "grad_norm": 1.3483792543411255,
+      "learning_rate": 1.764629649386713e-05,
+      "loss": 0.2219,
+      "step": 555
+    },
+    {
+      "epoch": 2.3837953091684434,
+      "grad_norm": 1.3648035526275635,
+      "learning_rate": 1.7438224235633945e-05,
+      "loss": 0.2506,
+      "step": 560
+    },
+    {
+      "epoch": 2.405117270788913,
+      "grad_norm": 1.1507258415222168,
+      "learning_rate": 1.7229668943247746e-05,
+      "loss": 0.2403,
+      "step": 565
+    },
+    {
+      "epoch": 2.4264392324093818,
+      "grad_norm": 1.390272855758667,
+      "learning_rate": 1.7020671933387917e-05,
+      "loss": 0.2288,
+      "step": 570
+    },
+    {
+      "epoch": 2.4477611940298507,
+      "grad_norm": 1.0744136571884155,
+      "learning_rate": 1.6811274610242043e-05,
+      "loss": 0.2239,
+      "step": 575
+    },
+    {
+      "epoch": 2.4690831556503197,
+      "grad_norm": 1.1554702520370483,
+      "learning_rate": 1.6601518457303384e-05,
+      "loss": 0.2251,
+      "step": 580
+    },
+    {
+      "epoch": 2.490405117270789,
+      "grad_norm": 1.432008147239685,
+      "learning_rate": 1.63914450291526e-05,
+      "loss": 0.271,
+      "step": 585
+    },
+    {
+      "epoch": 2.511727078891258,
+      "grad_norm": 1.2994996309280396,
+      "learning_rate": 1.6181095943225427e-05,
+      "loss": 0.2368,
+      "step": 590
+    },
+    {
+      "epoch": 2.533049040511727,
+      "grad_norm": 1.1314690113067627,
+      "learning_rate": 1.5970512871567895e-05,
+      "loss": 0.245,
+      "step": 595
+    },
+    {
+      "epoch": 2.5543710021321964,
+      "grad_norm": 1.2437481880187988,
+      "learning_rate": 1.5759737532580692e-05,
+      "loss": 0.2131,
+      "step": 600
+    },
+    {
+      "epoch": 2.5756929637526653,
+      "grad_norm": 1.0608434677124023,
+      "learning_rate": 1.5548811682754417e-05,
+      "loss": 0.2168,
+      "step": 605
+    },
+    {
+      "epoch": 2.5970149253731343,
+      "grad_norm": 1.1898307800292969,
+      "learning_rate": 1.5337777108397193e-05,
+      "loss": 0.2325,
+      "step": 610
+    },
+    {
+      "epoch": 2.6183368869936032,
+      "grad_norm": 1.2492387294769287,
+      "learning_rate": 1.5126675617356478e-05,
+      "loss": 0.2036,
+      "step": 615
+    },
+    {
+      "epoch": 2.6396588486140726,
+      "grad_norm": 1.3561724424362183,
+      "learning_rate": 1.4915549030736515e-05,
+      "loss": 0.2161,
+      "step": 620
+    },
+    {
+      "epoch": 2.6609808102345416,
+      "grad_norm": 1.2246372699737549,
+      "learning_rate": 1.4704439174613197e-05,
+      "loss": 0.2279,
+      "step": 625
+    },
+    {
+      "epoch": 2.6823027718550105,
+      "grad_norm": 1.1990751028060913,
+      "learning_rate": 1.4493387871747978e-05,
+      "loss": 0.2056,
+      "step": 630
+    },
+    {
+      "epoch": 2.70362473347548,
+      "grad_norm": 1.1299760341644287,
+      "learning_rate": 1.4282436933302364e-05,
+      "loss": 0.2034,
+      "step": 635
+    },
+    {
+      "epoch": 2.724946695095949,
+      "grad_norm": 1.505993127822876,
+      "learning_rate": 1.4071628150554774e-05,
+      "loss": 0.2242,
+      "step": 640
+    },
+    {
+      "epoch": 2.746268656716418,
+      "grad_norm": 1.0007997751235962,
+      "learning_rate": 1.3861003286621309e-05,
+      "loss": 0.2032,
+      "step": 645
+    },
+    {
+      "epoch": 2.767590618336887,
+      "grad_norm": 1.2401964664459229,
+      "learning_rate": 1.3650604068182083e-05,
+      "loss": 0.2213,
+      "step": 650
+    },
+    {
+      "epoch": 2.788912579957356,
+      "grad_norm": 1.1087535619735718,
+      "learning_rate": 1.344047217721484e-05,
+      "loss": 0.1952,
+      "step": 655
+    },
+    {
+      "epoch": 2.810234541577825,
+      "grad_norm": 1.085770845413208,
+      "learning_rate": 1.3230649242737312e-05,
+      "loss": 0.2045,
+      "step": 660
+    },
+    {
+      "epoch": 2.831556503198294,
+      "grad_norm": 1.3074979782104492,
+      "learning_rate": 1.3021176832560182e-05,
+      "loss": 0.1987,
+      "step": 665
+    },
+    {
+      "epoch": 2.8528784648187635,
+      "grad_norm": 1.2325820922851562,
+      "learning_rate": 1.2812096445052137e-05,
+      "loss": 0.2175,
+      "step": 670
+    },
+    {
+      "epoch": 2.8742004264392325,
+      "grad_norm": 1.1663273572921753,
+      "learning_rate": 1.2603449500918618e-05,
+      "loss": 0.1774,
+      "step": 675
+    },
+    {
+      "epoch": 2.8955223880597014,
+      "grad_norm": 1.1572972536087036,
+      "learning_rate": 1.2395277334996045e-05,
+      "loss": 0.1892,
+      "step": 680
+    },
+    {
+      "epoch": 2.9168443496801704,
+      "grad_norm": 1.1489613056182861,
+      "learning_rate": 1.2187621188063014e-05,
+      "loss": 0.1801,
+      "step": 685
+    },
+    {
+      "epoch": 2.9381663113006398,
+      "grad_norm": 1.2037951946258545,
+      "learning_rate": 1.1980522198670097e-05,
+      "loss": 0.1869,
+      "step": 690
+    },
+    {
+      "epoch": 2.9594882729211087,
+      "grad_norm": 1.103147268295288,
+      "learning_rate": 1.1774021394989973e-05,
+      "loss": 0.1657,
+      "step": 695
+    },
+    {
+      "epoch": 2.9808102345415777,
+      "grad_norm": 1.200391173362732,
+      "learning_rate": 1.1568159686689326e-05,
+      "loss": 0.1695,
+      "step": 700
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 1.577278971672058,
+      "learning_rate": 1.1362977856824299e-05,
+      "loss": 0.1736,
+      "step": 705
+    },
+    {
+      "epoch": 3.021321961620469,
+      "grad_norm": 0.9511349201202393,
+      "learning_rate": 1.1158516553761025e-05,
+      "loss": 0.1223,
+      "step": 710
+    },
+    {
+      "epoch": 3.0426439232409384,
+      "grad_norm": 1.420677661895752,
+      "learning_rate": 1.0954816283122771e-05,
+      "loss": 0.1621,
+      "step": 715
+    },
+    {
+      "epoch": 3.0639658848614073,
+      "grad_norm": 1.0316070318222046,
+      "learning_rate": 1.075191739976544e-05,
+      "loss": 0.1232,
+      "step": 720
+    },
+    {
+      "epoch": 3.0852878464818763,
+      "grad_norm": 1.1894824504852295,
+      "learning_rate": 1.0549860099782917e-05,
+      "loss": 0.1294,
+      "step": 725
+    },
+    {
+      "epoch": 3.106609808102345,
+      "grad_norm": 1.2497349977493286,
+      "learning_rate": 1.034868441254384e-05,
+      "loss": 0.159,
+      "step": 730
+    },
+    {
+      "epoch": 3.1279317697228146,
+      "grad_norm": 1.0093135833740234,
+      "learning_rate": 1.0148430192761429e-05,
+      "loss": 0.1453,
+      "step": 735
+    },
+    {
+      "epoch": 3.1492537313432836,
+      "grad_norm": 1.2263211011886597,
+      "learning_rate": 9.949137112597932e-06,
+      "loss": 0.1422,
+      "step": 740
+    },
+    {
+      "epoch": 3.1705756929637525,
+      "grad_norm": 1.1466641426086426,
+      "learning_rate": 9.750844653805166e-06,
+      "loss": 0.1507,
+      "step": 745
+    },
+    {
+      "epoch": 3.191897654584222,
+      "grad_norm": 0.9492596387863159,
+      "learning_rate": 9.553592099902882e-06,
+      "loss": 0.1326,
+      "step": 750
+    },
+    {
+      "epoch": 3.213219616204691,
+      "grad_norm": 1.2892968654632568,
+      "learning_rate": 9.357418528396295e-06,
+      "loss": 0.1346,
+      "step": 755
+    },
+    {
+      "epoch": 3.23454157782516,
+      "grad_norm": 1.1566206216812134,
+      "learning_rate": 9.162362803034512e-06,
+      "loss": 0.1534,
+      "step": 760
+    },
+    {
+      "epoch": 3.2558635394456292,
+      "grad_norm": 1.026112675666809,
+      "learning_rate": 8.968463566111263e-06,
+      "loss": 0.1322,
+      "step": 765
+    },
+    {
+      "epoch": 3.277185501066098,
+      "grad_norm": 1.202764868736267,
+      "learning_rate": 8.775759230809515e-06,
+      "loss": 0.1363,
+      "step": 770
+    },
+    {
+      "epoch": 3.298507462686567,
+      "grad_norm": 1.0371228456497192,
+      "learning_rate": 8.58428797359147e-06,
+      "loss": 0.1244,
+      "step": 775
+    },
+    {
+      "epoch": 3.319829424307036,
+      "grad_norm": 1.1627793312072754,
+      "learning_rate": 8.394087726635483e-06,
+      "loss": 0.1388,
+      "step": 780
+    },
+    {
+      "epoch": 3.3411513859275055,
+      "grad_norm": 0.9767821431159973,
+      "learning_rate": 8.205196170321336e-06,
+      "loss": 0.1279,
+      "step": 785
+    },
+    {
+      "epoch": 3.3624733475479744,
+      "grad_norm": 1.08342444896698,
+      "learning_rate": 8.017650725765425e-06,
+      "loss": 0.1298,
+      "step": 790
+    },
+    {
+      "epoch": 3.3837953091684434,
+      "grad_norm": 1.0018621683120728,
+      "learning_rate": 7.83148854740729e-06,
+      "loss": 0.112,
+      "step": 795
+    },
+    {
+      "epoch": 3.405117270788913,
+      "grad_norm": 1.4605910778045654,
+      "learning_rate": 7.646746515649e-06,
+      "loss": 0.1003,
+      "step": 800
+    },
+    {
+      "epoch": 3.4264392324093818,
+      "grad_norm": 0.9175068140029907,
+      "learning_rate": 7.4634612295488135e-06,
+      "loss": 0.1467,
+      "step": 805
+    },
+    {
+      "epoch": 3.4477611940298507,
+      "grad_norm": 1.04617178440094,
+      "learning_rate": 7.281668999570572e-06,
+      "loss": 0.1314,
+      "step": 810
+    },
+    {
+      "epoch": 3.4690831556503197,
+      "grad_norm": 0.9718805551528931,
+      "learning_rate": 7.101405840390246e-06,
+      "loss": 0.1468,
+      "step": 815
+    },
+    {
+      "epoch": 3.490405117270789,
+      "grad_norm": 0.9743406176567078,
+      "learning_rate": 6.922707463761168e-06,
+      "loss": 0.1107,
+      "step": 820
+    },
+    {
+      "epoch": 3.511727078891258,
+      "grad_norm": 1.0118330717086792,
+      "learning_rate": 6.74560927143913e-06,
+      "loss": 0.1362,
+      "step": 825
+    },
+    {
+      "epoch": 3.533049040511727,
+      "grad_norm": 1.0520509481430054,
+      "learning_rate": 6.570146348169044e-06,
+      "loss": 0.1027,
+      "step": 830
+    },
+    {
+      "epoch": 3.5543710021321964,
+      "grad_norm": 0.9696875810623169,
+      "learning_rate": 6.3963534547343126e-06,
+      "loss": 0.1206,
+      "step": 835
+    },
+    {
+      "epoch": 3.5756929637526653,
+      "grad_norm": 1.1316139698028564,
+      "learning_rate": 6.224265021070383e-06,
+      "loss": 0.1095,
+      "step": 840
+    },
+    {
+      "epoch": 3.5970149253731343,
+      "grad_norm": 0.9595882296562195,
+      "learning_rate": 6.0539151394439305e-06,
+      "loss": 0.1166,
+      "step": 845
+    },
+    {
+      "epoch": 3.6183368869936032,
+      "grad_norm": 0.9978195428848267,
+      "learning_rate": 5.885337557698791e-06,
+      "loss": 0.1144,
+      "step": 850
+    },
+    {
+      "epoch": 3.6396588486140726,
+      "grad_norm": 0.8326097726821899,
+      "learning_rate": 5.718565672570282e-06,
+      "loss": 0.1024,
+      "step": 855
+    },
+    {
+      "epoch": 3.6609808102345416,
+      "grad_norm": 1.2166277170181274,
+      "learning_rate": 5.553632523068963e-06,
+      "loss": 0.1106,
+      "step": 860
+    },
+    {
+      "epoch": 3.6823027718550105,
+      "grad_norm": 1.2250356674194336,
+      "learning_rate": 5.390570783935339e-06,
+      "loss": 0.1294,
+      "step": 865
+    },
+    {
+      "epoch": 3.70362473347548,
+      "grad_norm": 1.0315333604812622,
+      "learning_rate": 5.2294127591666695e-06,
+      "loss": 0.1129,
+      "step": 870
+    },
+    {
+      "epoch": 3.724946695095949,
+      "grad_norm": 0.9550820589065552,
+      "learning_rate": 5.07019037561729e-06,
+      "loss": 0.1027,
+      "step": 875
+    },
+    {
+      "epoch": 3.746268656716418,
+      "grad_norm": 0.977424681186676,
+      "learning_rate": 4.912935176673595e-06,
+      "loss": 0.1161,
+      "step": 880
+    },
+    {
+      "epoch": 3.767590618336887,
+      "grad_norm": 1.1264369487762451,
+      "learning_rate": 4.757678316005013e-06,
+      "loss": 0.1129,
+      "step": 885
+    },
+    {
+      "epoch": 3.788912579957356,
+      "grad_norm": 0.9593406915664673,
+      "learning_rate": 4.604450551392189e-06,
+      "loss": 0.1125,
+      "step": 890
+    },
+    {
+      "epoch": 3.810234541577825,
+      "grad_norm": 0.9950923919677734,
+      "learning_rate": 4.453282238633568e-06,
+      "loss": 0.1021,
+      "step": 895
+    },
+    {
+      "epoch": 3.831556503198294,
+      "grad_norm": 0.9612956047058105,
+      "learning_rate": 4.304203325531697e-06,
+      "loss": 0.1193,
+      "step": 900
+    },
+    {
+      "epoch": 3.8528784648187635,
+      "grad_norm": 0.9740220904350281,
+      "learning_rate": 4.157243345960214e-06,
+      "loss": 0.1049,
+      "step": 905
+    },
+    {
+      "epoch": 3.8742004264392325,
+      "grad_norm": 1.0198404788970947,
+      "learning_rate": 4.0124314140129915e-06,
+      "loss": 0.11,
+      "step": 910
+    },
+    {
+      "epoch": 3.8955223880597014,
+      "grad_norm": 0.9168893694877625,
+      "learning_rate": 3.869796218236342e-06,
+      "loss": 0.0927,
+      "step": 915
+    },
+    {
+      "epoch": 3.9168443496801704,
+      "grad_norm": 0.9452557563781738,
+      "learning_rate": 3.7293660159455495e-06,
+      "loss": 0.1122,
+      "step": 920
+    },
+    {
+      "epoch": 3.9381663113006398,
+      "grad_norm": 0.835529625415802,
+      "learning_rate": 3.5911686276268677e-06,
+      "loss": 0.1008,
+      "step": 925
+    },
+    {
+      "epoch": 3.9594882729211087,
+      "grad_norm": 0.8801108598709106,
+      "learning_rate": 3.455231431426016e-06,
+      "loss": 0.1054,
+      "step": 930
+    },
+    {
+      "epoch": 3.9808102345415777,
+      "grad_norm": 0.9656331539154053,
+      "learning_rate": 3.321581357724323e-06,
+      "loss": 0.1021,
+      "step": 935
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 1.2191102504730225,
+      "learning_rate": 3.190244883803577e-06,
+      "loss": 0.0969,
+      "step": 940
+    },
+    {
+      "epoch": 4.021321961620469,
+      "grad_norm": 0.7025108337402344,
+      "learning_rate": 3.061248028600644e-06,
+      "loss": 0.0873,
+      "step": 945
+    },
+    {
+      "epoch": 4.042643923240938,
+      "grad_norm": 0.858396053314209,
+      "learning_rate": 2.934616347552864e-06,
+      "loss": 0.0844,
+      "step": 950
+    },
+    {
+      "epoch": 4.063965884861407,
+      "grad_norm": 0.8153722882270813,
+      "learning_rate": 2.8103749275353134e-06,
+      "loss": 0.0868,
+      "step": 955
+    },
+    {
+      "epoch": 4.085287846481877,
+      "grad_norm": 0.8988513350486755,
+      "learning_rate": 2.688548381890859e-06,
+      "loss": 0.0838,
+      "step": 960
+    },
+    {
+      "epoch": 4.106609808102346,
+      "grad_norm": 0.8394201993942261,
+      "learning_rate": 2.5691608455540176e-06,
+      "loss": 0.083,
+      "step": 965
+    },
+    {
+      "epoch": 4.127931769722815,
+      "grad_norm": 0.7381258010864258,
+      "learning_rate": 2.452235970269649e-06,
+      "loss": 0.0895,
+      "step": 970
+    },
+    {
+      "epoch": 4.149253731343284,
+      "grad_norm": 0.875696063041687,
+      "learning_rate": 2.3377969199072767e-06,
+      "loss": 0.0686,
+      "step": 975
+    },
+    {
+      "epoch": 4.1705756929637525,
+      "grad_norm": 0.8519154191017151,
+      "learning_rate": 2.225866365872178e-06,
+      "loss": 0.0906,
+      "step": 980
+    },
+    {
+      "epoch": 4.1918976545842215,
+      "grad_norm": 0.8513870239257812,
+      "learning_rate": 2.1164664826139135e-06,
+      "loss": 0.0687,
+      "step": 985
+    },
+    {
+      "epoch": 4.21321961620469,
+      "grad_norm": 0.7496639490127563,
+      "learning_rate": 2.0096189432334194e-06,
+      "loss": 0.0968,
+      "step": 990
+    },
+    {
+      "epoch": 4.23454157782516,
+      "grad_norm": 0.7566796541213989,
+      "learning_rate": 1.9053449151893497e-06,
+      "loss": 0.087,
+      "step": 995
+    },
+    {
+      "epoch": 4.255863539445629,
+      "grad_norm": 0.7599784731864929,
+      "learning_rate": 1.8036650561046004e-06,
+      "loss": 0.0708,
+      "step": 1000
+    },
+    {
+      "epoch": 4.277185501066098,
+      "grad_norm": 0.7668546438217163,
+      "learning_rate": 1.7045995096738782e-06,
+      "loss": 0.0912,
+      "step": 1005
+    },
+    {
+      "epoch": 4.298507462686567,
+      "grad_norm": 0.7944720387458801,
+      "learning_rate": 1.6081679016730262e-06,
+      "loss": 0.0839,
+      "step": 1010
+    },
+    {
+      "epoch": 4.319829424307036,
+      "grad_norm": 0.7812253832817078,
+      "learning_rate": 1.5143893360709976e-06,
+      "loss": 0.0845,
+      "step": 1015
+    },
+    {
+      "epoch": 4.341151385927505,
+      "grad_norm": 1.0616964101791382,
+      "learning_rate": 1.4232823912451808e-06,
+      "loss": 0.0749,
+      "step": 1020
+    },
+    {
+      "epoch": 4.362473347547974,
+      "grad_norm": 0.8399378061294556,
+      "learning_rate": 1.3348651163008612e-06,
+      "loss": 0.0816,
+      "step": 1025
+    },
+    {
+      "epoch": 4.383795309168444,
+      "grad_norm": 0.7324888706207275,
+      "learning_rate": 1.249155027495515e-06,
+      "loss": 0.0862,
+      "step": 1030
+    },
+    {
+      "epoch": 4.405117270788913,
+      "grad_norm": 0.7757644057273865,
+      "learning_rate": 1.1661691047687045e-06,
+      "loss": 0.0775,
+      "step": 1035
+    },
+    {
+      "epoch": 4.426439232409382,
+      "grad_norm": 0.7491752505302429,
+      "learning_rate": 1.085923788378187e-06,
+      "loss": 0.0901,
+      "step": 1040
+    },
+    {
+      "epoch": 4.447761194029851,
+      "grad_norm": 0.8399827480316162,
+      "learning_rate": 1.00843497564294e-06,
+      "loss": 0.0808,
+      "step": 1045
+    },
+    {
+      "epoch": 4.46908315565032,
+      "grad_norm": 0.8545715808868408,
+      "learning_rate": 9.337180177937954e-07,
+      "loss": 0.0971,
+      "step": 1050
+    },
+    {
+      "epoch": 4.490405117270789,
+      "grad_norm": 0.8249493837356567,
+      "learning_rate": 8.617877169321742e-07,
+      "loss": 0.0812,
+      "step": 1055
+    },
+    {
+      "epoch": 4.5117270788912585,
+      "grad_norm": 0.7233726382255554,
+      "learning_rate": 7.926583230976997e-07,
+      "loss": 0.0916,
+      "step": 1060
+    },
+    {
+      "epoch": 4.533049040511727,
+      "grad_norm": 0.7911732196807861,
+      "learning_rate": 7.263435314451089e-07,
+      "loss": 0.0837,
+      "step": 1065
+    },
+    {
+      "epoch": 4.554371002132196,
+      "grad_norm": 0.8477759957313538,
+      "learning_rate": 6.62856479531136e-07,
+      "loss": 0.0968,
+      "step": 1070
+    },
+    {
+      "epoch": 4.575692963752665,
+      "grad_norm": 0.725795328617096,
+      "learning_rate": 6.022097447118341e-07,
+      "loss": 0.0692,
+      "step": 1075
+    },
+    {
+      "epoch": 4.597014925373134,
+      "grad_norm": 0.6900057196617126,
+      "learning_rate": 5.444153416508873e-07,
+      "loss": 0.0804,
+      "step": 1080
+    },
+    {
+      "epoch": 4.618336886993603,
+      "grad_norm": 0.7194017171859741,
+      "learning_rate": 4.894847199394093e-07,
+      "loss": 0.0809,
+      "step": 1085
+    },
+    {
+      "epoch": 4.639658848614072,
+      "grad_norm": 0.7520719170570374,
+      "learning_rate": 4.3742876182766233e-07,
+      "loss": 0.0743,
+      "step": 1090
+    },
+    {
+      "epoch": 4.660980810234541,
+      "grad_norm": 0.7377017736434937,
+      "learning_rate": 3.882577800691961e-07,
+      "loss": 0.0963,
+      "step": 1095
+    },
+    {
+      "epoch": 4.682302771855011,
+      "grad_norm": 0.9448336362838745,
+      "learning_rate": 3.419815158777878e-07,
+      "loss": 0.0929,
+      "step": 1100
+    },
+    {
+      "epoch": 4.70362473347548,
+      "grad_norm": 0.6936815977096558,
+      "learning_rate": 2.9860913699763624e-07,
+      "loss": 0.0732,
+      "step": 1105
+    },
+    {
+      "epoch": 4.724946695095949,
+      "grad_norm": 0.8193686604499817,
+      "learning_rate": 2.581492358871312e-07,
+      "loss": 0.0728,
+      "step": 1110
+    },
+    {
+      "epoch": 4.746268656716418,
+      "grad_norm": 0.8287943601608276,
+      "learning_rate": 2.2060982801662188e-07,
+      "loss": 0.0835,
+      "step": 1115
+    },
+    {
+      "epoch": 4.767590618336887,
+      "grad_norm": 0.7442787885665894,
+      "learning_rate": 1.8599835028047518e-07,
+      "loss": 0.0802,
+      "step": 1120
+    },
+    {
+      "epoch": 4.788912579957356,
+      "grad_norm": 0.8334233164787292,
+      "learning_rate": 1.5432165952376455e-07,
+      "loss": 0.081,
+      "step": 1125
+    },
+    {
+      "epoch": 4.810234541577826,
+      "grad_norm": 0.7535020112991333,
+      "learning_rate": 1.2558603118386137e-07,
+      "loss": 0.0846,
+      "step": 1130
+    },
+    {
+      "epoch": 4.8315565031982945,
+      "grad_norm": 0.7498242259025574,
+      "learning_rate": 9.97971580472179e-08,
+      "loss": 0.0695,
+      "step": 1135
+    },
+    {
+      "epoch": 4.8528784648187635,
+      "grad_norm": 0.8455618619918823,
+      "learning_rate": 7.696014912157267e-08,
+      "loss": 0.0803,
+      "step": 1140
+    },
+    {
+      "epoch": 4.8742004264392325,
+      "grad_norm": 0.703239917755127,
+      "learning_rate": 5.7079528623816824e-08,
+      "loss": 0.071,
+      "step": 1145
+    },
+    {
+      "epoch": 4.895522388059701,
+      "grad_norm": 0.8822847604751587,
+      "learning_rate": 4.0159235083697676e-08,
+      "loss": 0.0812,
+      "step": 1150
+    },
+    {
+      "epoch": 4.91684434968017,
+      "grad_norm": 0.7031238079071045,
+      "learning_rate": 2.6202620563561263e-08,
+      "loss": 0.0847,
+      "step": 1155
+    },
+    {
+      "epoch": 4.938166311300639,
+      "grad_norm": 0.707079291343689,
+      "learning_rate": 1.5212449994288502e-08,
+      "loss": 0.0691,
+      "step": 1160
+    },
+    {
+      "epoch": 4.959488272921108,
+      "grad_norm": 0.7360175848007202,
+      "learning_rate": 7.190900627525032e-09,
+      "loss": 0.0673,
+      "step": 1165
+    },
+    {
+      "epoch": 4.980810234541578,
+      "grad_norm": 0.76508629322052,
+      "learning_rate": 2.139561604352913e-09,
+      "loss": 0.0875,
+      "step": 1170
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.9940358996391296,
+      "learning_rate": 5.943364047078514e-11,
+      "loss": 0.0771,
+      "step": 1175
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1175,
+  "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": 1.703012282457391e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

45_128_e5_3e-5/checkpoint-1175/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3a879a5f8730abcf4ec5a5e22828c8fa360372fd1fe5d8c1bcbd49ca2b8d3d20
+size 7736

45_128_e5_3e-5/checkpoint-1175/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)

45_128_e5_3e-5/checkpoint-235/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

45_128_e5_3e-5/checkpoint-235/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": [
+    "q_proj",
+    "down_proj",
+    "v_proj",
+    "k_proj",
+    "up_proj",
+    "o_proj",
+    "gate_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

45_128_e5_3e-5/checkpoint-235/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c2c7c80f371a0592a0fc559bbc8c6525b89d1e8c3e1634ae558946881e8ffdb0
+size 791751704

45_128_e5_3e-5/checkpoint-235/latest

@@ -0,0 +1 @@
+global_step235

45_128_e5_3e-5/checkpoint-235/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:78899daf21815027461349772ff80b3c612fec193b24e4549e03ccb4ed7b1e4a
+size 15920

45_128_e5_3e-5/checkpoint-235/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3863f067b68130a509c5cd4171fd33ec7a0fbc8f2cfb848e7737d419f7cf99ab
+size 15920

45_128_e5_3e-5/checkpoint-235/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1bf597b4b510ba18263a21364489d6700ad17b7a9fe168e20f73ac1dd31cdb4d
+size 15920

45_128_e5_3e-5/checkpoint-235/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dde43f20f24405b03e905cb041878c12ad76f8731b40ca9fa1289cfb3b298238
+size 15920

45_128_e5_3e-5/checkpoint-235/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:757047e2022b403eea1ee5468e3d9fc326bd1573e987c425d67cd8834122a238
+size 15920

45_128_e5_3e-5/checkpoint-235/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7d51dc46cf50d46d350b6671e8037eba1d919fd02288478558f9d23610b55a17
+size 15920

45_128_e5_3e-5/checkpoint-235/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d1ec67b3b82bf97dcbb55a386cf7efc8188ab14fafe1c3e7f1be34e7a782ceac
+size 15920

45_128_e5_3e-5/checkpoint-235/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05c8b91be0cc5e8e1a44325058b77fedb7c7fc230cc74a79a8e68d02e95aeb5e
+size 15920

45_128_e5_3e-5/checkpoint-235/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b82123c3aa39d06d3decbdc3dfe308d337a33b24e52e59de3d40dc2705483380
+size 1064

45_128_e5_3e-5/checkpoint-235/special_tokens_map.json

@@ -0,0 +1,51 @@
+{
+  "additional_special_tokens": [
+    "<|endoftext|>",
+    "<fim_prefix>",
+    "<fim_middle>",
+    "<fim_suffix>",
+    "<fim_pad>",
+    "<filename>",
+    "<gh_stars>",
+    "<issue_start>",
+    "<issue_comment>",
+    "<issue_closed>",
+    "<jupyter_start>",
+    "<jupyter_text>",
+    "<jupyter_code>",
+    "<jupyter_output>",
+    "<empty_output>",
+    "<commit_before>",
+    "<commit_msg>",
+    "<commit_after>",
+    "<reponame>"
+  ],
+  "bos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

45_128_e5_3e-5/checkpoint-235/tokenizer_config.json

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

45_128_e5_3e-5/checkpoint-235/trainer_state.json

@@ -0,0 +1,363 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 235,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.021321961620469083,
+      "grad_norm": 1.1025959253311157,
+      "learning_rate": 2.033898305084746e-06,
+      "loss": 1.3396,
+      "step": 5
+    },
+    {
+      "epoch": 0.042643923240938165,
+      "grad_norm": 0.9394682049751282,
+      "learning_rate": 4.576271186440678e-06,
+      "loss": 1.3427,
+      "step": 10
+    },
+    {
+      "epoch": 0.06396588486140725,
+      "grad_norm": 0.5853310823440552,
+      "learning_rate": 7.1186440677966106e-06,
+      "loss": 1.3113,
+      "step": 15
+    },
+    {
+      "epoch": 0.08528784648187633,
+      "grad_norm": 1.0313405990600586,
+      "learning_rate": 9.661016949152542e-06,
+      "loss": 1.2611,
+      "step": 20
+    },
+    {
+      "epoch": 0.10660980810234541,
+      "grad_norm": 0.548238217830658,
+      "learning_rate": 1.2203389830508475e-05,
+      "loss": 1.2482,
+      "step": 25
+    },
+    {
+      "epoch": 0.1279317697228145,
+      "grad_norm": 0.6522016525268555,
+      "learning_rate": 1.4745762711864408e-05,
+      "loss": 1.2438,
+      "step": 30
+    },
+    {
+      "epoch": 0.14925373134328357,
+      "grad_norm": 0.6198495626449585,
+      "learning_rate": 1.728813559322034e-05,
+      "loss": 1.1851,
+      "step": 35
+    },
+    {
+      "epoch": 0.17057569296375266,
+      "grad_norm": 0.5091772079467773,
+      "learning_rate": 1.983050847457627e-05,
+      "loss": 1.2328,
+      "step": 40
+    },
+    {
+      "epoch": 0.19189765458422176,
+      "grad_norm": 0.43554070591926575,
+      "learning_rate": 2.2372881355932205e-05,
+      "loss": 1.2056,
+      "step": 45
+    },
+    {
+      "epoch": 0.21321961620469082,
+      "grad_norm": 0.4790154695510864,
+      "learning_rate": 2.4915254237288138e-05,
+      "loss": 1.1369,
+      "step": 50
+    },
+    {
+      "epoch": 0.2345415778251599,
+      "grad_norm": 0.5797903537750244,
+      "learning_rate": 2.7457627118644068e-05,
+      "loss": 1.1525,
+      "step": 55
+    },
+    {
+      "epoch": 0.255863539445629,
+      "grad_norm": 0.6414856314659119,
+      "learning_rate": 3e-05,
+      "loss": 1.118,
+      "step": 60
+    },
+    {
+      "epoch": 0.2771855010660981,
+      "grad_norm": 0.6532033085823059,
+      "learning_rate": 2.9998514182537154e-05,
+      "loss": 1.0826,
+      "step": 65
+    },
+    {
+      "epoch": 0.29850746268656714,
+      "grad_norm": 0.5140215158462524,
+      "learning_rate": 2.9994057024502427e-05,
+      "loss": 1.0808,
+      "step": 70
+    },
+    {
+      "epoch": 0.31982942430703626,
+      "grad_norm": 0.5573478937149048,
+      "learning_rate": 2.998662940889891e-05,
+      "loss": 1.0923,
+      "step": 75
+    },
+    {
+      "epoch": 0.3411513859275053,
+      "grad_norm": 0.6022165417671204,
+      "learning_rate": 2.9976232807204073e-05,
+      "loss": 1.044,
+      "step": 80
+    },
+    {
+      "epoch": 0.3624733475479744,
+      "grad_norm": 0.6599242687225342,
+      "learning_rate": 2.9962869279078226e-05,
+      "loss": 1.0641,
+      "step": 85
+    },
+    {
+      "epoch": 0.3837953091684435,
+      "grad_norm": 0.5529143810272217,
+      "learning_rate": 2.9946541471956496e-05,
+      "loss": 1.0251,
+      "step": 90
+    },
+    {
+      "epoch": 0.4051172707889126,
+      "grad_norm": 0.5225338339805603,
+      "learning_rate": 2.9927252620524346e-05,
+      "loss": 1.0353,
+      "step": 95
+    },
+    {
+      "epoch": 0.42643923240938164,
+      "grad_norm": 0.5583758354187012,
+      "learning_rate": 2.9905006546076746e-05,
+      "loss": 1.0061,
+      "step": 100
+    },
+    {
+      "epoch": 0.44776119402985076,
+      "grad_norm": 0.5726757645606995,
+      "learning_rate": 2.9879807655761145e-05,
+      "loss": 1.0173,
+      "step": 105
+    },
+    {
+      "epoch": 0.4690831556503198,
+      "grad_norm": 0.6370916366577148,
+      "learning_rate": 2.985166094170439e-05,
+      "loss": 1.0122,
+      "step": 110
+    },
+    {
+      "epoch": 0.4904051172707889,
+      "grad_norm": 0.6529775857925415,
+      "learning_rate": 2.9820571980023704e-05,
+      "loss": 0.911,
+      "step": 115
+    },
+    {
+      "epoch": 0.511727078891258,
+      "grad_norm": 0.6832720637321472,
+      "learning_rate": 2.9786546929722055e-05,
+      "loss": 0.9745,
+      "step": 120
+    },
+    {
+      "epoch": 0.5330490405117271,
+      "grad_norm": 0.6681720614433289,
+      "learning_rate": 2.974959253146796e-05,
+      "loss": 0.9218,
+      "step": 125
+    },
+    {
+      "epoch": 0.5543710021321961,
+      "grad_norm": 0.8444278836250305,
+      "learning_rate": 2.9709716106260115e-05,
+      "loss": 0.9233,
+      "step": 130
+    },
+    {
+      "epoch": 0.5756929637526652,
+      "grad_norm": 0.8203223347663879,
+      "learning_rate": 2.966692555397705e-05,
+      "loss": 0.9167,
+      "step": 135
+    },
+    {
+      "epoch": 0.5970149253731343,
+      "grad_norm": 0.6752311587333679,
+      "learning_rate": 2.962122935181207e-05,
+      "loss": 0.9187,
+      "step": 140
+    },
+    {
+      "epoch": 0.6183368869936035,
+      "grad_norm": 0.6852523684501648,
+      "learning_rate": 2.957263655259387e-05,
+      "loss": 0.8627,
+      "step": 145
+    },
+    {
+      "epoch": 0.6396588486140725,
+      "grad_norm": 0.8514593839645386,
+      "learning_rate": 2.9521156782993066e-05,
+      "loss": 0.8204,
+      "step": 150
+    },
+    {
+      "epoch": 0.6609808102345416,
+      "grad_norm": 0.8665833473205566,
+      "learning_rate": 2.9466800241615075e-05,
+      "loss": 0.8655,
+      "step": 155
+    },
+    {
+      "epoch": 0.6823027718550106,
+      "grad_norm": 0.9081581234931946,
+      "learning_rate": 2.940957769697969e-05,
+      "loss": 0.8356,
+      "step": 160
+    },
+    {
+      "epoch": 0.7036247334754797,
+      "grad_norm": 0.7875135540962219,
+      "learning_rate": 2.9349500485387718e-05,
+      "loss": 0.8547,
+      "step": 165
+    },
+    {
+      "epoch": 0.7249466950959488,
+      "grad_norm": 0.837771475315094,
+      "learning_rate": 2.9286580508675174e-05,
+      "loss": 0.7974,
+      "step": 170
+    },
+    {
+      "epoch": 0.746268656716418,
+      "grad_norm": 0.8674811720848083,
+      "learning_rate": 2.9220830231855417e-05,
+      "loss": 0.8343,
+      "step": 175
+    },
+    {
+      "epoch": 0.767590618336887,
+      "grad_norm": 0.8112357258796692,
+      "learning_rate": 2.9152262680649704e-05,
+      "loss": 0.7642,
+      "step": 180
+    },
+    {
+      "epoch": 0.7889125799573561,
+      "grad_norm": 0.855670690536499,
+      "learning_rate": 2.90808914389067e-05,
+      "loss": 0.778,
+      "step": 185
+    },
+    {
+      "epoch": 0.8102345415778252,
+      "grad_norm": 0.9481444358825684,
+      "learning_rate": 2.900673064591139e-05,
+      "loss": 0.7688,
+      "step": 190
+    },
+    {
+      "epoch": 0.8315565031982942,
+      "grad_norm": 0.9807798266410828,
+      "learning_rate": 2.8929794993583937e-05,
+      "loss": 0.7645,
+      "step": 195
+    },
+    {
+      "epoch": 0.8528784648187633,
+      "grad_norm": 1.074795126914978,
+      "learning_rate": 2.8850099723569104e-05,
+      "loss": 0.7552,
+      "step": 200
+    },
+    {
+      "epoch": 0.8742004264392325,
+      "grad_norm": 1.0988177061080933,
+      "learning_rate": 2.876766062421675e-05,
+      "loss": 0.7782,
+      "step": 205
+    },
+    {
+      "epoch": 0.8955223880597015,
+      "grad_norm": 1.0309362411499023,
+      "learning_rate": 2.8682494027454e-05,
+      "loss": 0.7621,
+      "step": 210
+    },
+    {
+      "epoch": 0.9168443496801706,
+      "grad_norm": 0.9497408866882324,
+      "learning_rate": 2.8594616805549752e-05,
+      "loss": 0.6876,
+      "step": 215
+    },
+    {
+      "epoch": 0.9381663113006397,
+      "grad_norm": 1.0013132095336914,
+      "learning_rate": 2.8504046367772117e-05,
+      "loss": 0.6893,
+      "step": 220
+    },
+    {
+      "epoch": 0.9594882729211087,
+      "grad_norm": 1.0180565118789673,
+      "learning_rate": 2.8410800656939512e-05,
+      "loss": 0.6678,
+      "step": 225
+    },
+    {
+      "epoch": 0.9808102345415778,
+      "grad_norm": 1.1235246658325195,
+      "learning_rate": 2.8314898145865996e-05,
+      "loss": 0.6193,
+      "step": 230
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.4771363735198975,
+      "learning_rate": 2.8216357833701667e-05,
+      "loss": 0.6586,
+      "step": 235
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1175,
+  "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": 3.398096257149829e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

45_128_e5_3e-5/checkpoint-235/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3a879a5f8730abcf4ec5a5e22828c8fa360372fd1fe5d8c1bcbd49ca2b8d3d20
+size 7736

45_128_e5_3e-5/checkpoint-235/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)

45_128_e5_3e-5/checkpoint-470/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

45_128_e5_3e-5/checkpoint-470/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": [
+    "q_proj",
+    "down_proj",
+    "v_proj",
+    "k_proj",
+    "up_proj",
+    "o_proj",
+    "gate_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

45_128_e5_3e-5/checkpoint-470/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b9e05086869e795ccf017cb57993999d14241e1cb04b37d2941126b3fd86ab97
+size 791751704

45_128_e5_3e-5/checkpoint-470/latest

@@ -0,0 +1 @@
+global_step470

45_128_e5_3e-5/checkpoint-470/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9ac918dcbd0cbe4e0b57efd3f848a5ebf3fbd5a37f4c64c70ed8f7f4db3656e1
+size 15920

45_128_e5_3e-5/checkpoint-470/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ff70a83e8ed1463a4cb32a09b8f04abe2d1a8735605a4d6953f14e17830b1682
+size 15920

45_128_e5_3e-5/checkpoint-470/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7d52bc7b1da0d8ea1091bcd8056dcac0dad99a86316b0819718582d7ed2fb225
+size 15920