First commit

README.md

@@ -1,3 +1,202 @@
----
-license: apache-2.0
----
+---
+library_name: peft
+base_model: mistralai/Mistral-7B-Instruct-v0.2
+---
+
+# 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 0.11.1

adapter_config.json

@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "mistralai/Mistral-7B-Instruct-v0.2",
+  "bias": "none",
+  "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": 16,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "o_proj",
+    "q_proj",
+    "up_proj",
+    "gate_proj",
+    "k_proj",
+    "v_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}

adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d9a5249e01ec6383d1e61416a3a514e18ce4d88662f0cbfd86e860caf4e69b67
+size 83946192

latest

@@ -0,0 +1 @@
+global_step1090

rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28c4884e413e8c06f75c2bfaf338b4e53759b9151b0cb22926d4aebf2fc46b69
+size 15920

rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:873eff60c8e6357f5fa85a50385c88ca9cf0e0a0535ba7fc6864e2c70977f92b
+size 15920

rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7f2816f3f249fa57d22d5ffd27f55cafabe858ed45602c4d8e0a6f5ef706d307
+size 15920

rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ab5140b8841b2341003358f6a4336ce7c82f39cfe4ea84f4482ffa5ba3378216
+size 15920

rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1b7a8024c4eb46676733602d133e854d0c2c6b995293cd169a9346e6e1f495bb
+size 15920

rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ad952d06f615db1dde85fab9338d3edf363b230d27078b2ff85cbfe64f04e253
+size 15920

rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c45c7b33daf3523bdb950675e7c0e4852a141a29b86f10ec798c55d3bd47365
+size 15920

rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:378fc15897d25f81a3489c099e10ce3bacfcd815c1f3350be98373e4bf78e626
+size 15920

scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:88f7c897a26550ebcba2dd73c884c90083e3059d5fa4c2e0f2898326dd316ccb
+size 1064

special_tokens_map.json

@@ -0,0 +1,24 @@
+{
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "</s>",
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dadfd56d766715c61d2ef780a525ab43b8e6da4de6865bda3d95fdef5e134055
+size 493443

tokenizer_config.json

@@ -0,0 +1,43 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [],
+  "bos_token": "<s>",
+  "chat_template": "{% for message in messages %}\n{% if message['role'] == 'user' %}\n{{ '<|user|>\n' + message['content'] + eos_token }}\n{% elif message['role'] == 'system' %}\n{{ '<|system|>\n' + message['content'] + eos_token }}\n{% elif message['role'] == 'assistant' %}\n{{ '<|assistant|>\n'  + message['content'] + eos_token }}\n{% endif %}\n{% if loop.last and add_generation_prompt %}\n{{ '<|assistant|>' }}\n{% endif %}\n{% endfor %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "</s>",
+  "legacy": true,
+  "model_max_length": 2048,
+  "pad_token": "</s>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}

trainer_state.json

@@ -0,0 +1,803 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 1090,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001834862385321101,
+      "grad_norm": 0.5285698626925894,
+      "learning_rate": 1.8315018315018315e-07,
+      "loss": 2.1792,
+      "step": 1
+    },
+    {
+      "epoch": 0.01834862385321101,
+      "grad_norm": 0.5570483132219919,
+      "learning_rate": 1.8315018315018316e-06,
+      "loss": 2.1303,
+      "step": 10
+    },
+    {
+      "epoch": 0.03669724770642202,
+      "grad_norm": 0.43928839139908665,
+      "learning_rate": 3.663003663003663e-06,
+      "loss": 2.143,
+      "step": 20
+    },
+    {
+      "epoch": 0.05504587155963303,
+      "grad_norm": 0.4259147884901134,
+      "learning_rate": 5.494505494505494e-06,
+      "loss": 2.0498,
+      "step": 30
+    },
+    {
+      "epoch": 0.07339449541284404,
+      "grad_norm": 0.3256334236213354,
+      "learning_rate": 7.326007326007326e-06,
+      "loss": 2.1005,
+      "step": 40
+    },
+    {
+      "epoch": 0.09174311926605505,
+      "grad_norm": 0.22687147012199685,
+      "learning_rate": 9.157509157509158e-06,
+      "loss": 2.032,
+      "step": 50
+    },
+    {
+      "epoch": 0.11009174311926606,
+      "grad_norm": 0.20407900243790855,
+      "learning_rate": 1.0989010989010989e-05,
+      "loss": 2.0265,
+      "step": 60
+    },
+    {
+      "epoch": 0.12844036697247707,
+      "grad_norm": 0.17272636847518857,
+      "learning_rate": 1.282051282051282e-05,
+      "loss": 2.0593,
+      "step": 70
+    },
+    {
+      "epoch": 0.14678899082568808,
+      "grad_norm": 0.15668903317199465,
+      "learning_rate": 1.4652014652014653e-05,
+      "loss": 2.009,
+      "step": 80
+    },
+    {
+      "epoch": 0.1651376146788991,
+      "grad_norm": 0.13299844821079024,
+      "learning_rate": 1.6483516483516486e-05,
+      "loss": 2.0319,
+      "step": 90
+    },
+    {
+      "epoch": 0.1834862385321101,
+      "grad_norm": 0.130419681456145,
+      "learning_rate": 1.8315018315018315e-05,
+      "loss": 2.0057,
+      "step": 100
+    },
+    {
+      "epoch": 0.2018348623853211,
+      "grad_norm": 0.13084640056684232,
+      "learning_rate": 2.0146520146520148e-05,
+      "loss": 2.0418,
+      "step": 110
+    },
+    {
+      "epoch": 0.22018348623853212,
+      "grad_norm": 0.13600193042807687,
+      "learning_rate": 2.1978021978021977e-05,
+      "loss": 1.9938,
+      "step": 120
+    },
+    {
+      "epoch": 0.23853211009174313,
+      "grad_norm": 0.13634836092075994,
+      "learning_rate": 2.380952380952381e-05,
+      "loss": 1.9859,
+      "step": 130
+    },
+    {
+      "epoch": 0.25688073394495414,
+      "grad_norm": 0.1422134721246097,
+      "learning_rate": 2.564102564102564e-05,
+      "loss": 2.0146,
+      "step": 140
+    },
+    {
+      "epoch": 0.27522935779816515,
+      "grad_norm": 0.15208974396882444,
+      "learning_rate": 2.7472527472527476e-05,
+      "loss": 2.0141,
+      "step": 150
+    },
+    {
+      "epoch": 0.29357798165137616,
+      "grad_norm": 0.14861311683006395,
+      "learning_rate": 2.9304029304029305e-05,
+      "loss": 1.9914,
+      "step": 160
+    },
+    {
+      "epoch": 0.3119266055045872,
+      "grad_norm": 0.154560148894899,
+      "learning_rate": 3.113553113553114e-05,
+      "loss": 1.9883,
+      "step": 170
+    },
+    {
+      "epoch": 0.3302752293577982,
+      "grad_norm": 0.14659138270717795,
+      "learning_rate": 3.296703296703297e-05,
+      "loss": 1.98,
+      "step": 180
+    },
+    {
+      "epoch": 0.3486238532110092,
+      "grad_norm": 0.14930915757375882,
+      "learning_rate": 3.47985347985348e-05,
+      "loss": 2.0039,
+      "step": 190
+    },
+    {
+      "epoch": 0.3669724770642202,
+      "grad_norm": 0.15222292163468532,
+      "learning_rate": 3.663003663003663e-05,
+      "loss": 1.9972,
+      "step": 200
+    },
+    {
+      "epoch": 0.3853211009174312,
+      "grad_norm": 0.1794465203569635,
+      "learning_rate": 3.846153846153846e-05,
+      "loss": 1.961,
+      "step": 210
+    },
+    {
+      "epoch": 0.4036697247706422,
+      "grad_norm": 0.16111917245406526,
+      "learning_rate": 4.0293040293040296e-05,
+      "loss": 1.9689,
+      "step": 220
+    },
+    {
+      "epoch": 0.42201834862385323,
+      "grad_norm": 0.16917075207279755,
+      "learning_rate": 4.212454212454213e-05,
+      "loss": 1.955,
+      "step": 230
+    },
+    {
+      "epoch": 0.44036697247706424,
+      "grad_norm": 0.16035752232756367,
+      "learning_rate": 4.3956043956043955e-05,
+      "loss": 1.9784,
+      "step": 240
+    },
+    {
+      "epoch": 0.45871559633027525,
+      "grad_norm": 0.15704206875567997,
+      "learning_rate": 4.578754578754579e-05,
+      "loss": 1.9898,
+      "step": 250
+    },
+    {
+      "epoch": 0.47706422018348627,
+      "grad_norm": 0.17540759793553962,
+      "learning_rate": 4.761904761904762e-05,
+      "loss": 1.9835,
+      "step": 260
+    },
+    {
+      "epoch": 0.4954128440366973,
+      "grad_norm": 0.15564282446806355,
+      "learning_rate": 4.945054945054945e-05,
+      "loss": 1.9849,
+      "step": 270
+    },
+    {
+      "epoch": 0.5137614678899083,
+      "grad_norm": 0.1731016854986911,
+      "learning_rate": 4.9998994546487535e-05,
+      "loss": 1.984,
+      "step": 280
+    },
+    {
+      "epoch": 0.5321100917431193,
+      "grad_norm": 0.16384332591458595,
+      "learning_rate": 4.999407007091408e-05,
+      "loss": 1.9846,
+      "step": 290
+    },
+    {
+      "epoch": 0.5504587155963303,
+      "grad_norm": 0.16272027999288557,
+      "learning_rate": 4.998504270550914e-05,
+      "loss": 1.9773,
+      "step": 300
+    },
+    {
+      "epoch": 0.5688073394495413,
+      "grad_norm": 0.1895435253320987,
+      "learning_rate": 4.997191393215565e-05,
+      "loss": 1.9544,
+      "step": 310
+    },
+    {
+      "epoch": 0.5871559633027523,
+      "grad_norm": 0.153237870046719,
+      "learning_rate": 4.995468590600123e-05,
+      "loss": 1.9627,
+      "step": 320
+    },
+    {
+      "epoch": 0.6055045871559633,
+      "grad_norm": 0.15686668474912246,
+      "learning_rate": 4.9933361455104425e-05,
+      "loss": 1.9977,
+      "step": 330
+    },
+    {
+      "epoch": 0.6238532110091743,
+      "grad_norm": 0.16748791541980212,
+      "learning_rate": 4.990794407997044e-05,
+      "loss": 1.9657,
+      "step": 340
+    },
+    {
+      "epoch": 0.6422018348623854,
+      "grad_norm": 0.1467745903785114,
+      "learning_rate": 4.9878437952976563e-05,
+      "loss": 1.9625,
+      "step": 350
+    },
+    {
+      "epoch": 0.6605504587155964,
+      "grad_norm": 0.14836352242961773,
+      "learning_rate": 4.984484791768721e-05,
+      "loss": 1.9865,
+      "step": 360
+    },
+    {
+      "epoch": 0.6788990825688074,
+      "grad_norm": 0.15162372625275922,
+      "learning_rate": 4.980717948805884e-05,
+      "loss": 1.9534,
+      "step": 370
+    },
+    {
+      "epoch": 0.6972477064220184,
+      "grad_norm": 0.15628620060925916,
+      "learning_rate": 4.9765438847534825e-05,
+      "loss": 1.976,
+      "step": 380
+    },
+    {
+      "epoch": 0.7155963302752294,
+      "grad_norm": 0.15072809988708077,
+      "learning_rate": 4.9719632848030405e-05,
+      "loss": 1.9598,
+      "step": 390
+    },
+    {
+      "epoch": 0.7339449541284404,
+      "grad_norm": 0.14670892778532468,
+      "learning_rate": 4.966976900880791e-05,
+      "loss": 2.0001,
+      "step": 400
+    },
+    {
+      "epoch": 0.7522935779816514,
+      "grad_norm": 0.15020025168070633,
+      "learning_rate": 4.9615855515242434e-05,
+      "loss": 1.9876,
+      "step": 410
+    },
+    {
+      "epoch": 0.7706422018348624,
+      "grad_norm": 0.14345074549432202,
+      "learning_rate": 4.955790121747821e-05,
+      "loss": 1.9839,
+      "step": 420
+    },
+    {
+      "epoch": 0.7889908256880734,
+      "grad_norm": 0.15176626779360905,
+      "learning_rate": 4.949591562897574e-05,
+      "loss": 1.9738,
+      "step": 430
+    },
+    {
+      "epoch": 0.8073394495412844,
+      "grad_norm": 0.15234818977369388,
+      "learning_rate": 4.942990892495021e-05,
+      "loss": 2.0086,
+      "step": 440
+    },
+    {
+      "epoch": 0.8256880733944955,
+      "grad_norm": 0.14268394349668895,
+      "learning_rate": 4.9359891940701086e-05,
+      "loss": 1.9797,
+      "step": 450
+    },
+    {
+      "epoch": 0.8440366972477065,
+      "grad_norm": 0.15852698385870923,
+      "learning_rate": 4.9285876169833544e-05,
+      "loss": 1.9532,
+      "step": 460
+    },
+    {
+      "epoch": 0.8623853211009175,
+      "grad_norm": 0.15294331275157483,
+      "learning_rate": 4.920787376237168e-05,
+      "loss": 1.992,
+      "step": 470
+    },
+    {
+      "epoch": 0.8807339449541285,
+      "grad_norm": 0.15177058058835824,
+      "learning_rate": 4.9125897522764044e-05,
+      "loss": 1.9902,
+      "step": 480
+    },
+    {
+      "epoch": 0.8990825688073395,
+      "grad_norm": 0.17927892676829146,
+      "learning_rate": 4.9039960907781746e-05,
+      "loss": 1.9694,
+      "step": 490
+    },
+    {
+      "epoch": 0.9174311926605505,
+      "grad_norm": 0.14983509077226922,
+      "learning_rate": 4.895007802430944e-05,
+      "loss": 1.9506,
+      "step": 500
+    },
+    {
+      "epoch": 0.9357798165137615,
+      "grad_norm": 0.16471555436795565,
+      "learning_rate": 4.885626362702966e-05,
+      "loss": 1.9841,
+      "step": 510
+    },
+    {
+      "epoch": 0.9541284403669725,
+      "grad_norm": 0.1472502715125968,
+      "learning_rate": 4.8758533116000696e-05,
+      "loss": 1.9832,
+      "step": 520
+    },
+    {
+      "epoch": 0.9724770642201835,
+      "grad_norm": 0.15070629202588426,
+      "learning_rate": 4.86569025341287e-05,
+      "loss": 1.9774,
+      "step": 530
+    },
+    {
+      "epoch": 0.9908256880733946,
+      "grad_norm": 0.14316992610781615,
+      "learning_rate": 4.855138856453408e-05,
+      "loss": 1.9692,
+      "step": 540
+    },
+    {
+      "epoch": 1.0091743119266054,
+      "grad_norm": 0.14585698001384728,
+      "learning_rate": 4.844200852781295e-05,
+      "loss": 1.9598,
+      "step": 550
+    },
+    {
+      "epoch": 1.0275229357798166,
+      "grad_norm": 0.15101510050666817,
+      "learning_rate": 4.8328780379193885e-05,
+      "loss": 1.9773,
+      "step": 560
+    },
+    {
+      "epoch": 1.0458715596330275,
+      "grad_norm": 0.15415482414608545,
+      "learning_rate": 4.821172270559039e-05,
+      "loss": 1.9699,
+      "step": 570
+    },
+    {
+      "epoch": 1.0642201834862386,
+      "grad_norm": 0.15483797940425081,
+      "learning_rate": 4.8090854722549914e-05,
+      "loss": 1.993,
+      "step": 580
+    },
+    {
+      "epoch": 1.0825688073394495,
+      "grad_norm": 0.15815400043247343,
+      "learning_rate": 4.796619627109944e-05,
+      "loss": 1.939,
+      "step": 590
+    },
+    {
+      "epoch": 1.1009174311926606,
+      "grad_norm": 0.1645249210360427,
+      "learning_rate": 4.7837767814488486e-05,
+      "loss": 1.9623,
+      "step": 600
+    },
+    {
+      "epoch": 1.1192660550458715,
+      "grad_norm": 0.1647619918199487,
+      "learning_rate": 4.770559043483003e-05,
+      "loss": 1.9816,
+      "step": 610
+    },
+    {
+      "epoch": 1.1376146788990826,
+      "grad_norm": 0.1679679060709561,
+      "learning_rate": 4.7569685829639734e-05,
+      "loss": 1.9611,
+      "step": 620
+    },
+    {
+      "epoch": 1.1559633027522935,
+      "grad_norm": 0.1676170375866798,
+      "learning_rate": 4.743007630827423e-05,
+      "loss": 1.959,
+      "step": 630
+    },
+    {
+      "epoch": 1.1743119266055047,
+      "grad_norm": 0.14905089408466188,
+      "learning_rate": 4.7286784788268904e-05,
+      "loss": 1.9269,
+      "step": 640
+    },
+    {
+      "epoch": 1.1926605504587156,
+      "grad_norm": 0.162817312686442,
+      "learning_rate": 4.713983479157592e-05,
+      "loss": 1.9638,
+      "step": 650
+    },
+    {
+      "epoch": 1.2110091743119267,
+      "grad_norm": 0.18370789078991157,
+      "learning_rate": 4.698925044070296e-05,
+      "loss": 1.9494,
+      "step": 660
+    },
+    {
+      "epoch": 1.2293577981651376,
+      "grad_norm": 0.15795771627671365,
+      "learning_rate": 4.683505645475339e-05,
+      "loss": 1.96,
+      "step": 670
+    },
+    {
+      "epoch": 1.2477064220183487,
+      "grad_norm": 0.15531127222157262,
+      "learning_rate": 4.6677278145368554e-05,
+      "loss": 1.969,
+      "step": 680
+    },
+    {
+      "epoch": 1.2660550458715596,
+      "grad_norm": 0.15599004624517146,
+      "learning_rate": 4.65159414125727e-05,
+      "loss": 1.9435,
+      "step": 690
+    },
+    {
+      "epoch": 1.2844036697247707,
+      "grad_norm": 0.16702581452844592,
+      "learning_rate": 4.6351072740521415e-05,
+      "loss": 1.9323,
+      "step": 700
+    },
+    {
+      "epoch": 1.3027522935779816,
+      "grad_norm": 0.15898151168116278,
+      "learning_rate": 4.6182699193154125e-05,
+      "loss": 1.9442,
+      "step": 710
+    },
+    {
+      "epoch": 1.3211009174311927,
+      "grad_norm": 0.15969818021992918,
+      "learning_rate": 4.601084840975139e-05,
+      "loss": 1.973,
+      "step": 720
+    },
+    {
+      "epoch": 1.3394495412844036,
+      "grad_norm": 0.15378573113733301,
+      "learning_rate": 4.583554860039784e-05,
+      "loss": 1.9366,
+      "step": 730
+    },
+    {
+      "epoch": 1.3577981651376148,
+      "grad_norm": 0.15121087319924134,
+      "learning_rate": 4.565682854135132e-05,
+      "loss": 1.9698,
+      "step": 740
+    },
+    {
+      "epoch": 1.3761467889908257,
+      "grad_norm": 0.15898725981112244,
+      "learning_rate": 4.547471757031919e-05,
+      "loss": 1.9604,
+      "step": 750
+    },
+    {
+      "epoch": 1.3944954128440368,
+      "grad_norm": 0.15306064791439133,
+      "learning_rate": 4.528924558164233e-05,
+      "loss": 1.962,
+      "step": 760
+    },
+    {
+      "epoch": 1.4128440366972477,
+      "grad_norm": 0.16371512899536222,
+      "learning_rate": 4.510044302138793e-05,
+      "loss": 1.9793,
+      "step": 770
+    },
+    {
+      "epoch": 1.4311926605504588,
+      "grad_norm": 0.15788911392614066,
+      "learning_rate": 4.490834088235157e-05,
+      "loss": 1.9801,
+      "step": 780
+    },
+    {
+      "epoch": 1.4495412844036697,
+      "grad_norm": 0.16398987312372718,
+      "learning_rate": 4.4712970698969645e-05,
+      "loss": 1.9236,
+      "step": 790
+    },
+    {
+      "epoch": 1.4678899082568808,
+      "grad_norm": 0.15020427466606878,
+      "learning_rate": 4.451436454214285e-05,
+      "loss": 1.9438,
+      "step": 800
+    },
+    {
+      "epoch": 1.4862385321100917,
+      "grad_norm": 0.1528061160370563,
+      "learning_rate": 4.4312555013971534e-05,
+      "loss": 1.9364,
+      "step": 810
+    },
+    {
+      "epoch": 1.5045871559633026,
+      "grad_norm": 0.15222177715019794,
+      "learning_rate": 4.4107575242404013e-05,
+      "loss": 1.9399,
+      "step": 820
+    },
+    {
+      "epoch": 1.5229357798165137,
+      "grad_norm": 0.1540985658911008,
+      "learning_rate": 4.38994588757984e-05,
+      "loss": 1.9483,
+      "step": 830
+    },
+    {
+      "epoch": 1.5412844036697249,
+      "grad_norm": 0.1599074284657503,
+      "learning_rate": 4.3688240077399074e-05,
+      "loss": 1.9748,
+      "step": 840
+    },
+    {
+      "epoch": 1.5596330275229358,
+      "grad_norm": 0.1516707606485465,
+      "learning_rate": 4.3473953519728685e-05,
+      "loss": 1.9213,
+      "step": 850
+    },
+    {
+      "epoch": 1.5779816513761467,
+      "grad_norm": 0.15122814764762293,
+      "learning_rate": 4.325663437889643e-05,
+      "loss": 1.9893,
+      "step": 860
+    },
+    {
+      "epoch": 1.5963302752293578,
+      "grad_norm": 0.15959987778428422,
+      "learning_rate": 4.30363183288238e-05,
+      "loss": 1.9602,
+      "step": 870
+    },
+    {
+      "epoch": 1.614678899082569,
+      "grad_norm": 0.15266741829090746,
+      "learning_rate": 4.2813041535388496e-05,
+      "loss": 1.9529,
+      "step": 880
+    },
+    {
+      "epoch": 1.6330275229357798,
+      "grad_norm": 0.16386079884752924,
+      "learning_rate": 4.258684065048766e-05,
+      "loss": 1.9606,
+      "step": 890
+    },
+    {
+      "epoch": 1.6513761467889907,
+      "grad_norm": 0.15583648586262694,
+      "learning_rate": 4.23577528060213e-05,
+      "loss": 1.9581,
+      "step": 900
+    },
+    {
+      "epoch": 1.6697247706422018,
+      "grad_norm": 0.1585736145888436,
+      "learning_rate": 4.212581560779689e-05,
+      "loss": 1.9552,
+      "step": 910
+    },
+    {
+      "epoch": 1.688073394495413,
+      "grad_norm": 0.16965341576331627,
+      "learning_rate": 4.1891067129356276e-05,
+      "loss": 1.9296,
+      "step": 920
+    },
+    {
+      "epoch": 1.7064220183486238,
+      "grad_norm": 0.15927292960064882,
+      "learning_rate": 4.165354590572564e-05,
+      "loss": 1.9762,
+      "step": 930
+    },
+    {
+      "epoch": 1.7247706422018347,
+      "grad_norm": 0.15450025224674474,
+      "learning_rate": 4.14132909270899e-05,
+      "loss": 1.9429,
+      "step": 940
+    },
+    {
+      "epoch": 1.7431192660550459,
+      "grad_norm": 0.15355485102004446,
+      "learning_rate": 4.117034163239219e-05,
+      "loss": 1.9233,
+      "step": 950
+    },
+    {
+      "epoch": 1.761467889908257,
+      "grad_norm": 0.15539982061507696,
+      "learning_rate": 4.092473790285986e-05,
+      "loss": 1.944,
+      "step": 960
+    },
+    {
+      "epoch": 1.7798165137614679,
+      "grad_norm": 0.1551191369865497,
+      "learning_rate": 4.0676520055457765e-05,
+      "loss": 1.945,
+      "step": 970
+    },
+    {
+      "epoch": 1.7981651376146788,
+      "grad_norm": 0.15695341020759185,
+      "learning_rate": 4.0425728836270037e-05,
+      "loss": 1.9656,
+      "step": 980
+    },
+    {
+      "epoch": 1.81651376146789,
+      "grad_norm": 0.16047022345268508,
+      "learning_rate": 4.017240541381146e-05,
+      "loss": 1.9546,
+      "step": 990
+    },
+    {
+      "epoch": 1.834862385321101,
+      "grad_norm": 0.1561566895170307,
+      "learning_rate": 3.9916591372269434e-05,
+      "loss": 1.9363,
+      "step": 1000
+    },
+    {
+      "epoch": 1.853211009174312,
+      "grad_norm": 0.15117586495122826,
+      "learning_rate": 3.9658328704677794e-05,
+      "loss": 1.978,
+      "step": 1010
+    },
+    {
+      "epoch": 1.8715596330275228,
+      "grad_norm": 0.1605982135175342,
+      "learning_rate": 3.939765980602342e-05,
+      "loss": 1.9713,
+      "step": 1020
+    },
+    {
+      "epoch": 1.889908256880734,
+      "grad_norm": 0.1602836931731585,
+      "learning_rate": 3.913462746628691e-05,
+      "loss": 2.0041,
+      "step": 1030
+    },
+    {
+      "epoch": 1.908256880733945,
+      "grad_norm": 0.16286472480475192,
+      "learning_rate": 3.886927486341844e-05,
+      "loss": 1.9352,
+      "step": 1040
+    },
+    {
+      "epoch": 1.926605504587156,
+      "grad_norm": 0.15449199584717296,
+      "learning_rate": 3.860164555624988e-05,
+      "loss": 1.97,
+      "step": 1050
+    },
+    {
+      "epoch": 1.9449541284403669,
+      "grad_norm": 0.14684566302878604,
+      "learning_rate": 3.833178347734443e-05,
+      "loss": 1.9433,
+      "step": 1060
+    },
+    {
+      "epoch": 1.963302752293578,
+      "grad_norm": 0.16167305499507623,
+      "learning_rate": 3.80597329257849e-05,
+      "loss": 1.9782,
+      "step": 1070
+    },
+    {
+      "epoch": 1.981651376146789,
+      "grad_norm": 0.1544185075202575,
+      "learning_rate": 3.778553855990176e-05,
+      "loss": 1.9253,
+      "step": 1080
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.16072854970224754,
+      "learning_rate": 3.750924538994235e-05,
+      "loss": 1.9578,
+      "step": 1090
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 2725,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 5,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.5869823436193792e+16,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c90183ac8f9afe09b314a736fcd6fbbaac4738bfb6c5e97864e724f23d3f4083
+size 6392

zero_to_fp32.py

@@ -0,0 +1,587 @@
+#!/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):
+    """
+    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)
+    elif zero_stage == 3:
+        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states)
+
+
+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 _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()
+            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):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    _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):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    _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):
+    """
+    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``
+
+    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)
+
+
+def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None):
+    """
+    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``
+    """
+
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+    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("-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)