Upload folder using huggingface_hub

README.md

@@ -0,0 +1,202 @@
+---
+base_model: meta-llama/Llama-2-7b-hf
+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 0.14.0

adapter_config.json

@@ -0,0 +1,37 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
+  "bias": "none",
+  "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": 64,
+  "lora_bias": false,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 32,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "v_proj",
+    "o_proj",
+    "q_proj",
+    "k_proj",
+    "up_proj",
+    "gate_proj"
+  ],
+  "task_type": "FEATURE_EXTRACTION",
+  "use_dora": false,
+  "use_rslora": false
+}

adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4658821a31b42f54caae8ae01ebb1479a99c60fef9a319cbd8769d5d23104d42
+size 159965640

global_step1655/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:172b7207682725f03087160da0f57975d9487d5d3a80de657adaf7d1d9755856
+size 479727536

global_step1655/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c45ba4dcd849e9515f24a57e2b2504b80ca3ee8ef7fe7394164521cbfd98bb8
+size 479727536

global_step1655/zero_pp_rank_0_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:67ad42b12298f4b30031f3e0a35e3fd9f9759ebebb13337071a5831259258c31
+size 6608028458

global_step1655/zero_pp_rank_1_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dcc22d6b81df6a05660bd8fc85f4456c7b6233ab68b68c3b4f40cb12495cee1c
+size 6608028458

latest

@@ -0,0 +1 @@
+global_step1655

rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d44e4cb93d62c803bd53846e7c9483dec147759c6015852861ef7a54acad8c2b
+size 14512

rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:962bdb064e165c3976f599dce744351bc065adf4c64e20b0b335dc351196822e
+size 14512

trainer_state.json

@@ -0,0 +1,1188 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9996979764421625,
+  "eval_steps": 500,
+  "global_step": 1655,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.006040471156750227,
+      "grad_norm": 6.352119345594041,
+      "learning_rate": 4.9999999999999996e-05,
+      "loss": 5.7625,
+      "step": 10
+    },
+    {
+      "epoch": 0.012080942313500454,
+      "grad_norm": 3.5107589040708578,
+      "learning_rate": 6.505149978319905e-05,
+      "loss": 3.1477,
+      "step": 20
+    },
+    {
+      "epoch": 0.018121413470250678,
+      "grad_norm": 1.7597502982401718,
+      "learning_rate": 7.385606273598311e-05,
+      "loss": 1.8234,
+      "step": 30
+    },
+    {
+      "epoch": 0.024161884627000908,
+      "grad_norm": 2.0430582664699704,
+      "learning_rate": 8.01029995663981e-05,
+      "loss": 1.482,
+      "step": 40
+    },
+    {
+      "epoch": 0.030202355783751134,
+      "grad_norm": 2.30138022647804,
+      "learning_rate": 8.494850021680092e-05,
+      "loss": 1.427,
+      "step": 50
+    },
+    {
+      "epoch": 0.036242826940501356,
+      "grad_norm": 1.961300808555845,
+      "learning_rate": 8.890756251918216e-05,
+      "loss": 1.3652,
+      "step": 60
+    },
+    {
+      "epoch": 0.042283298097251586,
+      "grad_norm": 1.5670057056368645,
+      "learning_rate": 9.225490200071284e-05,
+      "loss": 1.2859,
+      "step": 70
+    },
+    {
+      "epoch": 0.048323769254001815,
+      "grad_norm": 1.5648604780816326,
+      "learning_rate": 9.515449934959716e-05,
+      "loss": 1.2166,
+      "step": 80
+    },
+    {
+      "epoch": 0.05436424041075204,
+      "grad_norm": 1.5594489363452173,
+      "learning_rate": 9.771212547196623e-05,
+      "loss": 1.2299,
+      "step": 90
+    },
+    {
+      "epoch": 0.06040471156750227,
+      "grad_norm": 1.478498798125124,
+      "learning_rate": 9.999999999999999e-05,
+      "loss": 1.1406,
+      "step": 100
+    },
+    {
+      "epoch": 0.0664451827242525,
+      "grad_norm": 1.3456415286113537,
+      "learning_rate": 9.942122186495178e-05,
+      "loss": 1.1512,
+      "step": 110
+    },
+    {
+      "epoch": 0.07248565388100271,
+      "grad_norm": 1.4169868695799486,
+      "learning_rate": 9.877813504823152e-05,
+      "loss": 1.2473,
+      "step": 120
+    },
+    {
+      "epoch": 0.07852612503775294,
+      "grad_norm": 1.5442406663559958,
+      "learning_rate": 9.813504823151127e-05,
+      "loss": 1.1609,
+      "step": 130
+    },
+    {
+      "epoch": 0.08456659619450317,
+      "grad_norm": 1.3840740480619829,
+      "learning_rate": 9.7491961414791e-05,
+      "loss": 1.1617,
+      "step": 140
+    },
+    {
+      "epoch": 0.0906070673512534,
+      "grad_norm": 1.1099163069476086,
+      "learning_rate": 9.684887459807074e-05,
+      "loss": 1.1746,
+      "step": 150
+    },
+    {
+      "epoch": 0.09664753850800363,
+      "grad_norm": 1.2420434112131873,
+      "learning_rate": 9.620578778135049e-05,
+      "loss": 1.1504,
+      "step": 160
+    },
+    {
+      "epoch": 0.10268800966475385,
+      "grad_norm": 1.081321720135299,
+      "learning_rate": 9.556270096463023e-05,
+      "loss": 1.1605,
+      "step": 170
+    },
+    {
+      "epoch": 0.10872848082150408,
+      "grad_norm": 1.0982034537770153,
+      "learning_rate": 9.491961414790998e-05,
+      "loss": 1.0799,
+      "step": 180
+    },
+    {
+      "epoch": 0.1147689519782543,
+      "grad_norm": 1.1736811633063773,
+      "learning_rate": 9.427652733118972e-05,
+      "loss": 1.1242,
+      "step": 190
+    },
+    {
+      "epoch": 0.12080942313500453,
+      "grad_norm": 1.3595797342301437,
+      "learning_rate": 9.363344051446946e-05,
+      "loss": 1.1016,
+      "step": 200
+    },
+    {
+      "epoch": 0.12684989429175475,
+      "grad_norm": 1.2621841618934533,
+      "learning_rate": 9.29903536977492e-05,
+      "loss": 1.0695,
+      "step": 210
+    },
+    {
+      "epoch": 0.132890365448505,
+      "grad_norm": 1.2070699106064489,
+      "learning_rate": 9.234726688102894e-05,
+      "loss": 1.1049,
+      "step": 220
+    },
+    {
+      "epoch": 0.1389308366052552,
+      "grad_norm": 1.0416383062176056,
+      "learning_rate": 9.17041800643087e-05,
+      "loss": 1.1467,
+      "step": 230
+    },
+    {
+      "epoch": 0.14497130776200542,
+      "grad_norm": 1.1856097509180394,
+      "learning_rate": 9.106109324758843e-05,
+      "loss": 1.0746,
+      "step": 240
+    },
+    {
+      "epoch": 0.15101177891875567,
+      "grad_norm": 1.144246874115688,
+      "learning_rate": 9.041800643086817e-05,
+      "loss": 1.2018,
+      "step": 250
+    },
+    {
+      "epoch": 0.15705225007550588,
+      "grad_norm": 0.9353587652819314,
+      "learning_rate": 8.977491961414792e-05,
+      "loss": 1.0789,
+      "step": 260
+    },
+    {
+      "epoch": 0.16309272123225613,
+      "grad_norm": 1.0285527563460544,
+      "learning_rate": 8.913183279742766e-05,
+      "loss": 1.0555,
+      "step": 270
+    },
+    {
+      "epoch": 0.16913319238900634,
+      "grad_norm": 0.9852353032151006,
+      "learning_rate": 8.848874598070739e-05,
+      "loss": 1.1451,
+      "step": 280
+    },
+    {
+      "epoch": 0.17517366354575656,
+      "grad_norm": 1.175932073617665,
+      "learning_rate": 8.784565916398714e-05,
+      "loss": 1.0789,
+      "step": 290
+    },
+    {
+      "epoch": 0.1812141347025068,
+      "grad_norm": 1.213820312892746,
+      "learning_rate": 8.720257234726688e-05,
+      "loss": 1.0475,
+      "step": 300
+    },
+    {
+      "epoch": 0.18725460585925702,
+      "grad_norm": 1.1846528600649535,
+      "learning_rate": 8.655948553054663e-05,
+      "loss": 1.0521,
+      "step": 310
+    },
+    {
+      "epoch": 0.19329507701600726,
+      "grad_norm": 1.1497197430459096,
+      "learning_rate": 8.591639871382637e-05,
+      "loss": 1.082,
+      "step": 320
+    },
+    {
+      "epoch": 0.19933554817275748,
+      "grad_norm": 1.1612797319381625,
+      "learning_rate": 8.52733118971061e-05,
+      "loss": 1.0793,
+      "step": 330
+    },
+    {
+      "epoch": 0.2053760193295077,
+      "grad_norm": 1.082739707844854,
+      "learning_rate": 8.463022508038586e-05,
+      "loss": 1.1061,
+      "step": 340
+    },
+    {
+      "epoch": 0.21141649048625794,
+      "grad_norm": 1.2757899011878036,
+      "learning_rate": 8.398713826366559e-05,
+      "loss": 1.073,
+      "step": 350
+    },
+    {
+      "epoch": 0.21745696164300815,
+      "grad_norm": 0.9540956853203417,
+      "learning_rate": 8.334405144694534e-05,
+      "loss": 1.0135,
+      "step": 360
+    },
+    {
+      "epoch": 0.2234974327997584,
+      "grad_norm": 1.0200709241255603,
+      "learning_rate": 8.270096463022508e-05,
+      "loss": 1.1389,
+      "step": 370
+    },
+    {
+      "epoch": 0.2295379039565086,
+      "grad_norm": 0.8455317026975807,
+      "learning_rate": 8.205787781350482e-05,
+      "loss": 1.0598,
+      "step": 380
+    },
+    {
+      "epoch": 0.23557837511325883,
+      "grad_norm": 1.1381349305391164,
+      "learning_rate": 8.141479099678457e-05,
+      "loss": 1.0777,
+      "step": 390
+    },
+    {
+      "epoch": 0.24161884627000907,
+      "grad_norm": 1.103747383663609,
+      "learning_rate": 8.07717041800643e-05,
+      "loss": 1.0617,
+      "step": 400
+    },
+    {
+      "epoch": 0.24765931742675928,
+      "grad_norm": 1.138904626814912,
+      "learning_rate": 8.012861736334406e-05,
+      "loss": 1.0324,
+      "step": 410
+    },
+    {
+      "epoch": 0.2536997885835095,
+      "grad_norm": 1.0781428262426909,
+      "learning_rate": 7.94855305466238e-05,
+      "loss": 1.0859,
+      "step": 420
+    },
+    {
+      "epoch": 0.2597402597402597,
+      "grad_norm": 1.0846995831727089,
+      "learning_rate": 7.884244372990353e-05,
+      "loss": 1.0553,
+      "step": 430
+    },
+    {
+      "epoch": 0.26578073089701,
+      "grad_norm": 0.8641072003497396,
+      "learning_rate": 7.819935691318328e-05,
+      "loss": 1.0068,
+      "step": 440
+    },
+    {
+      "epoch": 0.2718212020537602,
+      "grad_norm": 1.110226877216393,
+      "learning_rate": 7.755627009646302e-05,
+      "loss": 1.0115,
+      "step": 450
+    },
+    {
+      "epoch": 0.2778616732105104,
+      "grad_norm": 1.1418636977442045,
+      "learning_rate": 7.691318327974277e-05,
+      "loss": 1.0373,
+      "step": 460
+    },
+    {
+      "epoch": 0.28390214436726063,
+      "grad_norm": 0.9476740742119087,
+      "learning_rate": 7.62700964630225e-05,
+      "loss": 1.0023,
+      "step": 470
+    },
+    {
+      "epoch": 0.28994261552401085,
+      "grad_norm": 1.1736350313366837,
+      "learning_rate": 7.562700964630224e-05,
+      "loss": 1.0338,
+      "step": 480
+    },
+    {
+      "epoch": 0.2959830866807611,
+      "grad_norm": 1.113923619981689,
+      "learning_rate": 7.4983922829582e-05,
+      "loss": 0.9855,
+      "step": 490
+    },
+    {
+      "epoch": 0.30202355783751134,
+      "grad_norm": 1.0389339180007815,
+      "learning_rate": 7.434083601286173e-05,
+      "loss": 1.0525,
+      "step": 500
+    },
+    {
+      "epoch": 0.30806402899426155,
+      "grad_norm": 1.161270624803982,
+      "learning_rate": 7.369774919614148e-05,
+      "loss": 1.0697,
+      "step": 510
+    },
+    {
+      "epoch": 0.31410450015101177,
+      "grad_norm": 0.8892068789215148,
+      "learning_rate": 7.305466237942122e-05,
+      "loss": 0.9889,
+      "step": 520
+    },
+    {
+      "epoch": 0.320144971307762,
+      "grad_norm": 1.0455859065719944,
+      "learning_rate": 7.241157556270097e-05,
+      "loss": 0.9898,
+      "step": 530
+    },
+    {
+      "epoch": 0.32618544246451225,
+      "grad_norm": 1.0186894188406512,
+      "learning_rate": 7.17684887459807e-05,
+      "loss": 0.9904,
+      "step": 540
+    },
+    {
+      "epoch": 0.33222591362126247,
+      "grad_norm": 1.0072092839183986,
+      "learning_rate": 7.112540192926044e-05,
+      "loss": 0.9873,
+      "step": 550
+    },
+    {
+      "epoch": 0.3382663847780127,
+      "grad_norm": 0.9663462582485958,
+      "learning_rate": 7.04823151125402e-05,
+      "loss": 1.0375,
+      "step": 560
+    },
+    {
+      "epoch": 0.3443068559347629,
+      "grad_norm": 0.9103712505082897,
+      "learning_rate": 6.983922829581993e-05,
+      "loss": 1.0383,
+      "step": 570
+    },
+    {
+      "epoch": 0.3503473270915131,
+      "grad_norm": 1.0485963233672557,
+      "learning_rate": 6.919614147909968e-05,
+      "loss": 0.9758,
+      "step": 580
+    },
+    {
+      "epoch": 0.3563877982482634,
+      "grad_norm": 0.9912979171159205,
+      "learning_rate": 6.855305466237942e-05,
+      "loss": 0.993,
+      "step": 590
+    },
+    {
+      "epoch": 0.3624282694050136,
+      "grad_norm": 1.1550179596081722,
+      "learning_rate": 6.790996784565917e-05,
+      "loss": 1.032,
+      "step": 600
+    },
+    {
+      "epoch": 0.3684687405617638,
+      "grad_norm": 1.0176105219446732,
+      "learning_rate": 6.726688102893891e-05,
+      "loss": 1.0094,
+      "step": 610
+    },
+    {
+      "epoch": 0.37450921171851403,
+      "grad_norm": 1.1210756115920897,
+      "learning_rate": 6.662379421221864e-05,
+      "loss": 0.968,
+      "step": 620
+    },
+    {
+      "epoch": 0.38054968287526425,
+      "grad_norm": 0.9310657150450071,
+      "learning_rate": 6.59807073954984e-05,
+      "loss": 1.0102,
+      "step": 630
+    },
+    {
+      "epoch": 0.3865901540320145,
+      "grad_norm": 1.1633016127800648,
+      "learning_rate": 6.533762057877813e-05,
+      "loss": 1.0256,
+      "step": 640
+    },
+    {
+      "epoch": 0.39263062518876474,
+      "grad_norm": 0.8195008101746646,
+      "learning_rate": 6.469453376205788e-05,
+      "loss": 0.9906,
+      "step": 650
+    },
+    {
+      "epoch": 0.39867109634551495,
+      "grad_norm": 0.9310761172857367,
+      "learning_rate": 6.405144694533762e-05,
+      "loss": 0.9754,
+      "step": 660
+    },
+    {
+      "epoch": 0.40471156750226517,
+      "grad_norm": 1.0516630367824455,
+      "learning_rate": 6.340836012861736e-05,
+      "loss": 1.0275,
+      "step": 670
+    },
+    {
+      "epoch": 0.4107520386590154,
+      "grad_norm": 1.0629036969985795,
+      "learning_rate": 6.276527331189711e-05,
+      "loss": 1.0088,
+      "step": 680
+    },
+    {
+      "epoch": 0.41679250981576565,
+      "grad_norm": 1.0110994904370134,
+      "learning_rate": 6.212218649517684e-05,
+      "loss": 1.0416,
+      "step": 690
+    },
+    {
+      "epoch": 0.42283298097251587,
+      "grad_norm": 0.8895584335650736,
+      "learning_rate": 6.14790996784566e-05,
+      "loss": 1.0762,
+      "step": 700
+    },
+    {
+      "epoch": 0.4288734521292661,
+      "grad_norm": 1.0448634604827216,
+      "learning_rate": 6.083601286173633e-05,
+      "loss": 0.9771,
+      "step": 710
+    },
+    {
+      "epoch": 0.4349139232860163,
+      "grad_norm": 0.7905625978703504,
+      "learning_rate": 6.019292604501608e-05,
+      "loss": 1.0104,
+      "step": 720
+    },
+    {
+      "epoch": 0.4409543944427665,
+      "grad_norm": 1.0522004404580236,
+      "learning_rate": 5.954983922829582e-05,
+      "loss": 0.9977,
+      "step": 730
+    },
+    {
+      "epoch": 0.4469948655995168,
+      "grad_norm": 0.989475716298655,
+      "learning_rate": 5.8906752411575564e-05,
+      "loss": 0.9752,
+      "step": 740
+    },
+    {
+      "epoch": 0.453035336756267,
+      "grad_norm": 1.1463565668787514,
+      "learning_rate": 5.826366559485531e-05,
+      "loss": 0.9922,
+      "step": 750
+    },
+    {
+      "epoch": 0.4590758079130172,
+      "grad_norm": 1.189865689753586,
+      "learning_rate": 5.7620578778135045e-05,
+      "loss": 1.0211,
+      "step": 760
+    },
+    {
+      "epoch": 0.46511627906976744,
+      "grad_norm": 1.0920686333427474,
+      "learning_rate": 5.6977491961414796e-05,
+      "loss": 0.9664,
+      "step": 770
+    },
+    {
+      "epoch": 0.47115675022651765,
+      "grad_norm": 1.0174913677757231,
+      "learning_rate": 5.633440514469453e-05,
+      "loss": 0.9992,
+      "step": 780
+    },
+    {
+      "epoch": 0.4771972213832679,
+      "grad_norm": 1.1913080680366728,
+      "learning_rate": 5.5691318327974284e-05,
+      "loss": 1.033,
+      "step": 790
+    },
+    {
+      "epoch": 0.48323769254001814,
+      "grad_norm": 1.0204323369307062,
+      "learning_rate": 5.504823151125402e-05,
+      "loss": 0.9807,
+      "step": 800
+    },
+    {
+      "epoch": 0.48927816369676835,
+      "grad_norm": 1.118864062431584,
+      "learning_rate": 5.440514469453376e-05,
+      "loss": 0.974,
+      "step": 810
+    },
+    {
+      "epoch": 0.49531863485351857,
+      "grad_norm": 1.0350012597705058,
+      "learning_rate": 5.376205787781351e-05,
+      "loss": 0.9705,
+      "step": 820
+    },
+    {
+      "epoch": 0.5013591060102688,
+      "grad_norm": 1.0845595520731828,
+      "learning_rate": 5.3118971061093246e-05,
+      "loss": 0.9582,
+      "step": 830
+    },
+    {
+      "epoch": 0.507399577167019,
+      "grad_norm": 0.9453749574005488,
+      "learning_rate": 5.2475884244372996e-05,
+      "loss": 1.0293,
+      "step": 840
+    },
+    {
+      "epoch": 0.5134400483237692,
+      "grad_norm": 1.042699724920899,
+      "learning_rate": 5.183279742765273e-05,
+      "loss": 0.952,
+      "step": 850
+    },
+    {
+      "epoch": 0.5194805194805194,
+      "grad_norm": 1.024020216858902,
+      "learning_rate": 5.118971061093247e-05,
+      "loss": 1.018,
+      "step": 860
+    },
+    {
+      "epoch": 0.5255209906372698,
+      "grad_norm": 1.0360224854752502,
+      "learning_rate": 5.054662379421222e-05,
+      "loss": 0.9295,
+      "step": 870
+    },
+    {
+      "epoch": 0.53156146179402,
+      "grad_norm": 1.1683508369643563,
+      "learning_rate": 4.9903536977491965e-05,
+      "loss": 1.0133,
+      "step": 880
+    },
+    {
+      "epoch": 0.5376019329507702,
+      "grad_norm": 0.8614966755997071,
+      "learning_rate": 4.92604501607717e-05,
+      "loss": 0.993,
+      "step": 890
+    },
+    {
+      "epoch": 0.5436424041075204,
+      "grad_norm": 0.9857361206695465,
+      "learning_rate": 4.8617363344051446e-05,
+      "loss": 0.951,
+      "step": 900
+    },
+    {
+      "epoch": 0.5496828752642706,
+      "grad_norm": 0.8956078547131043,
+      "learning_rate": 4.797427652733119e-05,
+      "loss": 0.9268,
+      "step": 910
+    },
+    {
+      "epoch": 0.5557233464210208,
+      "grad_norm": 0.8885268581836658,
+      "learning_rate": 4.7331189710610934e-05,
+      "loss": 0.9688,
+      "step": 920
+    },
+    {
+      "epoch": 0.561763817577771,
+      "grad_norm": 1.136712299735102,
+      "learning_rate": 4.668810289389068e-05,
+      "loss": 0.9545,
+      "step": 930
+    },
+    {
+      "epoch": 0.5678042887345213,
+      "grad_norm": 1.114040839440962,
+      "learning_rate": 4.604501607717042e-05,
+      "loss": 1.0,
+      "step": 940
+    },
+    {
+      "epoch": 0.5738447598912715,
+      "grad_norm": 1.2120423884348053,
+      "learning_rate": 4.5401929260450165e-05,
+      "loss": 0.9525,
+      "step": 950
+    },
+    {
+      "epoch": 0.5798852310480217,
+      "grad_norm": 0.9841344976466476,
+      "learning_rate": 4.47588424437299e-05,
+      "loss": 1.0389,
+      "step": 960
+    },
+    {
+      "epoch": 0.585925702204772,
+      "grad_norm": 1.0459005071658793,
+      "learning_rate": 4.4115755627009646e-05,
+      "loss": 1.0041,
+      "step": 970
+    },
+    {
+      "epoch": 0.5919661733615222,
+      "grad_norm": 1.2096290639870764,
+      "learning_rate": 4.347266881028939e-05,
+      "loss": 0.9939,
+      "step": 980
+    },
+    {
+      "epoch": 0.5980066445182725,
+      "grad_norm": 1.0171142436328486,
+      "learning_rate": 4.2829581993569134e-05,
+      "loss": 0.9432,
+      "step": 990
+    },
+    {
+      "epoch": 0.6040471156750227,
+      "grad_norm": 1.0557726010616861,
+      "learning_rate": 4.218649517684888e-05,
+      "loss": 1.0041,
+      "step": 1000
+    },
+    {
+      "epoch": 0.6100875868317729,
+      "grad_norm": 0.9854936325370102,
+      "learning_rate": 4.154340836012862e-05,
+      "loss": 0.9598,
+      "step": 1010
+    },
+    {
+      "epoch": 0.6161280579885231,
+      "grad_norm": 1.0002881708013533,
+      "learning_rate": 4.090032154340836e-05,
+      "loss": 0.9537,
+      "step": 1020
+    },
+    {
+      "epoch": 0.6221685291452733,
+      "grad_norm": 1.0163254545569222,
+      "learning_rate": 4.02572347266881e-05,
+      "loss": 0.9781,
+      "step": 1030
+    },
+    {
+      "epoch": 0.6282090003020235,
+      "grad_norm": 1.0700856083800017,
+      "learning_rate": 3.961414790996785e-05,
+      "loss": 0.9418,
+      "step": 1040
+    },
+    {
+      "epoch": 0.6342494714587738,
+      "grad_norm": 0.925802326406659,
+      "learning_rate": 3.897106109324759e-05,
+      "loss": 0.9355,
+      "step": 1050
+    },
+    {
+      "epoch": 0.640289942615524,
+      "grad_norm": 1.0173964783122673,
+      "learning_rate": 3.8327974276527335e-05,
+      "loss": 0.9311,
+      "step": 1060
+    },
+    {
+      "epoch": 0.6463304137722743,
+      "grad_norm": 1.0886760638416628,
+      "learning_rate": 3.768488745980708e-05,
+      "loss": 1.0104,
+      "step": 1070
+    },
+    {
+      "epoch": 0.6523708849290245,
+      "grad_norm": 0.8784480294988184,
+      "learning_rate": 3.7041800643086816e-05,
+      "loss": 0.9631,
+      "step": 1080
+    },
+    {
+      "epoch": 0.6584113560857747,
+      "grad_norm": 1.0061477546516637,
+      "learning_rate": 3.639871382636656e-05,
+      "loss": 0.9209,
+      "step": 1090
+    },
+    {
+      "epoch": 0.6644518272425249,
+      "grad_norm": 1.0443361002098694,
+      "learning_rate": 3.5755627009646303e-05,
+      "loss": 0.9494,
+      "step": 1100
+    },
+    {
+      "epoch": 0.6704922983992752,
+      "grad_norm": 1.064188874483296,
+      "learning_rate": 3.511254019292605e-05,
+      "loss": 0.9871,
+      "step": 1110
+    },
+    {
+      "epoch": 0.6765327695560254,
+      "grad_norm": 0.9804777210337282,
+      "learning_rate": 3.446945337620579e-05,
+      "loss": 0.9357,
+      "step": 1120
+    },
+    {
+      "epoch": 0.6825732407127756,
+      "grad_norm": 1.0053083581127769,
+      "learning_rate": 3.3826366559485535e-05,
+      "loss": 0.9637,
+      "step": 1130
+    },
+    {
+      "epoch": 0.6886137118695258,
+      "grad_norm": 1.0062328413826747,
+      "learning_rate": 3.318327974276528e-05,
+      "loss": 0.9914,
+      "step": 1140
+    },
+    {
+      "epoch": 0.694654183026276,
+      "grad_norm": 1.0659942748249296,
+      "learning_rate": 3.2540192926045016e-05,
+      "loss": 0.9332,
+      "step": 1150
+    },
+    {
+      "epoch": 0.7006946541830262,
+      "grad_norm": 1.2425227959394274,
+      "learning_rate": 3.189710610932476e-05,
+      "loss": 0.9951,
+      "step": 1160
+    },
+    {
+      "epoch": 0.7067351253397764,
+      "grad_norm": 1.0269910163244402,
+      "learning_rate": 3.1254019292604504e-05,
+      "loss": 0.9805,
+      "step": 1170
+    },
+    {
+      "epoch": 0.7127755964965268,
+      "grad_norm": 1.067485855484705,
+      "learning_rate": 3.061093247588425e-05,
+      "loss": 0.975,
+      "step": 1180
+    },
+    {
+      "epoch": 0.718816067653277,
+      "grad_norm": 0.9819234749980894,
+      "learning_rate": 2.996784565916399e-05,
+      "loss": 0.9805,
+      "step": 1190
+    },
+    {
+      "epoch": 0.7248565388100272,
+      "grad_norm": 0.9915476687600292,
+      "learning_rate": 2.9324758842443732e-05,
+      "loss": 0.9365,
+      "step": 1200
+    },
+    {
+      "epoch": 0.7308970099667774,
+      "grad_norm": 1.048821106029052,
+      "learning_rate": 2.8681672025723473e-05,
+      "loss": 0.9861,
+      "step": 1210
+    },
+    {
+      "epoch": 0.7369374811235276,
+      "grad_norm": 0.9121022219179783,
+      "learning_rate": 2.8038585209003217e-05,
+      "loss": 0.9854,
+      "step": 1220
+    },
+    {
+      "epoch": 0.7429779522802779,
+      "grad_norm": 0.9525288844616361,
+      "learning_rate": 2.739549839228296e-05,
+      "loss": 0.948,
+      "step": 1230
+    },
+    {
+      "epoch": 0.7490184234370281,
+      "grad_norm": 1.1391100357014714,
+      "learning_rate": 2.67524115755627e-05,
+      "loss": 0.8887,
+      "step": 1240
+    },
+    {
+      "epoch": 0.7550588945937783,
+      "grad_norm": 1.1078585866746686,
+      "learning_rate": 2.6109324758842445e-05,
+      "loss": 0.9715,
+      "step": 1250
+    },
+    {
+      "epoch": 0.7610993657505285,
+      "grad_norm": 0.9323091046055703,
+      "learning_rate": 2.546623794212219e-05,
+      "loss": 0.9359,
+      "step": 1260
+    },
+    {
+      "epoch": 0.7671398369072787,
+      "grad_norm": 1.146046378807531,
+      "learning_rate": 2.482315112540193e-05,
+      "loss": 0.8967,
+      "step": 1270
+    },
+    {
+      "epoch": 0.773180308064029,
+      "grad_norm": 1.14163561365635,
+      "learning_rate": 2.4180064308681673e-05,
+      "loss": 0.9707,
+      "step": 1280
+    },
+    {
+      "epoch": 0.7792207792207793,
+      "grad_norm": 1.0034307361335086,
+      "learning_rate": 2.3536977491961414e-05,
+      "loss": 0.9668,
+      "step": 1290
+    },
+    {
+      "epoch": 0.7852612503775295,
+      "grad_norm": 1.1199578096409222,
+      "learning_rate": 2.2893890675241158e-05,
+      "loss": 0.968,
+      "step": 1300
+    },
+    {
+      "epoch": 0.7913017215342797,
+      "grad_norm": 1.0809213063777794,
+      "learning_rate": 2.22508038585209e-05,
+      "loss": 0.9264,
+      "step": 1310
+    },
+    {
+      "epoch": 0.7973421926910299,
+      "grad_norm": 1.1266044230874084,
+      "learning_rate": 2.1607717041800642e-05,
+      "loss": 1.0203,
+      "step": 1320
+    },
+    {
+      "epoch": 0.8033826638477801,
+      "grad_norm": 0.8540728431635001,
+      "learning_rate": 2.0964630225080386e-05,
+      "loss": 0.924,
+      "step": 1330
+    },
+    {
+      "epoch": 0.8094231350045303,
+      "grad_norm": 1.0063045365537524,
+      "learning_rate": 2.032154340836013e-05,
+      "loss": 0.9656,
+      "step": 1340
+    },
+    {
+      "epoch": 0.8154636061612806,
+      "grad_norm": 0.9432512605760064,
+      "learning_rate": 1.967845659163987e-05,
+      "loss": 1.0254,
+      "step": 1350
+    },
+    {
+      "epoch": 0.8215040773180308,
+      "grad_norm": 1.162199676010242,
+      "learning_rate": 1.9035369774919614e-05,
+      "loss": 0.9549,
+      "step": 1360
+    },
+    {
+      "epoch": 0.827544548474781,
+      "grad_norm": 1.0184734737311432,
+      "learning_rate": 1.8392282958199358e-05,
+      "loss": 0.8937,
+      "step": 1370
+    },
+    {
+      "epoch": 0.8335850196315313,
+      "grad_norm": 1.0321283898574098,
+      "learning_rate": 1.77491961414791e-05,
+      "loss": 0.9641,
+      "step": 1380
+    },
+    {
+      "epoch": 0.8396254907882815,
+      "grad_norm": 0.948342731761788,
+      "learning_rate": 1.7106109324758842e-05,
+      "loss": 0.907,
+      "step": 1390
+    },
+    {
+      "epoch": 0.8456659619450317,
+      "grad_norm": 1.016072519476065,
+      "learning_rate": 1.6463022508038586e-05,
+      "loss": 0.999,
+      "step": 1400
+    },
+    {
+      "epoch": 0.851706433101782,
+      "grad_norm": 0.8891894724654406,
+      "learning_rate": 1.581993569131833e-05,
+      "loss": 0.9238,
+      "step": 1410
+    },
+    {
+      "epoch": 0.8577469042585322,
+      "grad_norm": 1.0116016221718374,
+      "learning_rate": 1.517684887459807e-05,
+      "loss": 0.9488,
+      "step": 1420
+    },
+    {
+      "epoch": 0.8637873754152824,
+      "grad_norm": 1.092739541532347,
+      "learning_rate": 1.4533762057877815e-05,
+      "loss": 0.9971,
+      "step": 1430
+    },
+    {
+      "epoch": 0.8698278465720326,
+      "grad_norm": 0.8563957174375877,
+      "learning_rate": 1.3890675241157558e-05,
+      "loss": 1.0133,
+      "step": 1440
+    },
+    {
+      "epoch": 0.8758683177287828,
+      "grad_norm": 1.089288756096709,
+      "learning_rate": 1.3247588424437299e-05,
+      "loss": 0.9127,
+      "step": 1450
+    },
+    {
+      "epoch": 0.881908788885533,
+      "grad_norm": 0.9458299067214011,
+      "learning_rate": 1.2604501607717043e-05,
+      "loss": 0.9668,
+      "step": 1460
+    },
+    {
+      "epoch": 0.8879492600422833,
+      "grad_norm": 1.2229020868039389,
+      "learning_rate": 1.1961414790996785e-05,
+      "loss": 0.9506,
+      "step": 1470
+    },
+    {
+      "epoch": 0.8939897311990336,
+      "grad_norm": 1.0615217830735095,
+      "learning_rate": 1.1318327974276529e-05,
+      "loss": 0.9219,
+      "step": 1480
+    },
+    {
+      "epoch": 0.9000302023557838,
+      "grad_norm": 1.1622944573297567,
+      "learning_rate": 1.0675241157556271e-05,
+      "loss": 0.9391,
+      "step": 1490
+    },
+    {
+      "epoch": 0.906070673512534,
+      "grad_norm": 0.9443566657008828,
+      "learning_rate": 1.0032154340836013e-05,
+      "loss": 0.8955,
+      "step": 1500
+    },
+    {
+      "epoch": 0.9121111446692842,
+      "grad_norm": 1.1367332066610845,
+      "learning_rate": 9.389067524115757e-06,
+      "loss": 0.9385,
+      "step": 1510
+    },
+    {
+      "epoch": 0.9181516158260344,
+      "grad_norm": 0.9414922696266849,
+      "learning_rate": 8.7459807073955e-06,
+      "loss": 0.9318,
+      "step": 1520
+    },
+    {
+      "epoch": 0.9241920869827847,
+      "grad_norm": 1.0844935096035118,
+      "learning_rate": 8.102893890675242e-06,
+      "loss": 0.9311,
+      "step": 1530
+    },
+    {
+      "epoch": 0.9302325581395349,
+      "grad_norm": 1.120509038145209,
+      "learning_rate": 7.459807073954985e-06,
+      "loss": 0.9422,
+      "step": 1540
+    },
+    {
+      "epoch": 0.9362730292962851,
+      "grad_norm": 0.9486397409946379,
+      "learning_rate": 6.816720257234727e-06,
+      "loss": 0.9162,
+      "step": 1550
+    },
+    {
+      "epoch": 0.9423135004530353,
+      "grad_norm": 1.1412479158621043,
+      "learning_rate": 6.17363344051447e-06,
+      "loss": 0.9676,
+      "step": 1560
+    },
+    {
+      "epoch": 0.9483539716097855,
+      "grad_norm": 1.050318112867144,
+      "learning_rate": 5.530546623794212e-06,
+      "loss": 0.9754,
+      "step": 1570
+    },
+    {
+      "epoch": 0.9543944427665358,
+      "grad_norm": 1.0466438135565226,
+      "learning_rate": 4.887459807073955e-06,
+      "loss": 0.926,
+      "step": 1580
+    },
+    {
+      "epoch": 0.9604349139232861,
+      "grad_norm": 0.9720115456232871,
+      "learning_rate": 4.244372990353698e-06,
+      "loss": 0.9447,
+      "step": 1590
+    },
+    {
+      "epoch": 0.9664753850800363,
+      "grad_norm": 0.8786923129454015,
+      "learning_rate": 3.6012861736334403e-06,
+      "loss": 0.9443,
+      "step": 1600
+    },
+    {
+      "epoch": 0.9725158562367865,
+      "grad_norm": 1.0733513985052048,
+      "learning_rate": 2.9581993569131834e-06,
+      "loss": 0.9053,
+      "step": 1610
+    },
+    {
+      "epoch": 0.9785563273935367,
+      "grad_norm": 0.8731368575153022,
+      "learning_rate": 2.315112540192926e-06,
+      "loss": 0.9803,
+      "step": 1620
+    },
+    {
+      "epoch": 0.9845967985502869,
+      "grad_norm": 1.0060959243542735,
+      "learning_rate": 1.6720257234726688e-06,
+      "loss": 0.9563,
+      "step": 1630
+    },
+    {
+      "epoch": 0.9906372697070371,
+      "grad_norm": 0.9113226806468382,
+      "learning_rate": 1.0289389067524116e-06,
+      "loss": 0.9191,
+      "step": 1640
+    },
+    {
+      "epoch": 0.9966777408637874,
+      "grad_norm": 0.9160377841463557,
+      "learning_rate": 3.8585209003215437e-07,
+      "loss": 0.9117,
+      "step": 1650
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 1655,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:889b418411635b080223cf5b494471bfd7a53dadaa45890f705449cbb042f27f
+size 7480

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)