Training in progress, step 50, checkpoint

last-checkpoint/README.md

@@ -1,6 +1,6 @@
 ---
 library_name: peft
-base_model: bn22/Mistral-7B-v0.1-sharded
+base_model: alexsherstinsky/Mistral-7B-v0.1-sharded
 ---
 
 # Model Card for Model ID

last-checkpoint/adapter_config.json

@@ -1,7 +1,7 @@
 {
   "alpha_pattern": {},
   "auto_mapping": null,
-  "base_model_name_or_path": "bn22/Mistral-7B-v0.1-sharded",
+  "base_model_name_or_path": "alexsherstinsky/Mistral-7B-v0.1-sharded",
   "bias": "none",
   "fan_in_fan_out": false,
   "inference_mode": true,
@@ -17,12 +17,12 @@
   "revision": null,
   "target_modules": [
     "o_proj",
+    "k_proj",
+    "up_proj",
+    "v_proj",
     "q_proj",
     "down_proj",
-    "v_proj",
-    "gate_proj",
-    "k_proj",
-    "up_proj"
+    "gate_proj"
   ],
   "task_type": "CAUSAL_LM"
 }

last-checkpoint/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:53ce65206cf6ca0cf316938d51c6c033e151e1e6ac908945246c9b956f4a10d9
-size 83945296
+oid sha256:c1d26ae4d91ee5406ecdb04754fcb794c49fdbf6bbb41861d8558309f1105dad
+size 42002136

last-checkpoint/global_step50/mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4361fddc8c415ce1454201b361168cd2d48d0c91ea4864257e518f77d84aca18
+size 8182659910

last-checkpoint/global_step50/zero_pp_rank_0_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:98d7331c6fcdd957f25dbd69d15a4939c741d5692b698f8c99287f9ad6fb941b
+size 251710893

last-checkpoint/latest

@@ -0,0 +1 @@
+global_step50

last-checkpoint/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:169c733cf7764f1d968254c0dd657bf888f42fc96732223a4c949c1fbe9bd0bd
+oid sha256:95108ce3daae0de505fcc447f4742cf96275124742d94331b4e3e1cbe5b3334d
 size 14244

last-checkpoint/trainer_state.json

@@ -1,619 +1,319 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.0024875621890547263,
+  "epoch": 0.0012437810945273632,
   "eval_steps": 1000,
-  "global_step": 100,
+  "global_step": 50,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 0.0,
-      "loss": 1.9492,
+      "learning_rate": 0,
+      "loss": 1.5524,
       "step": 1
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 1.9778,
+      "loss": 2.1402,
       "step": 2
     },
     {
       "epoch": 0.0,
-      "learning_rate": 4e-05,
-      "loss": 2.0908,
+      "learning_rate": 0.0,
+      "loss": 1.664,
       "step": 3
     },
     {
       "epoch": 0.0,
-      "learning_rate": 4e-05,
-      "loss": 2.0456,
+      "learning_rate": 8.613531161467861e-05,
+      "loss": 1.7923,
       "step": 4
     },
     {
       "epoch": 0.0,
-      "learning_rate": 8e-05,
-      "loss": 2.2288,
+      "learning_rate": 0.00013652123889719707,
+      "loss": 1.8119,
       "step": 5
     },
     {
       "epoch": 0.0,
-      "learning_rate": 0.00012,
-      "loss": 1.7521,
+      "learning_rate": 0.00017227062322935723,
+      "loss": 1.6156,
       "step": 6
     },
     {
       "epoch": 0.0,
-      "learning_rate": 0.00016,
-      "loss": 1.746,
+      "learning_rate": 0.0002,
+      "loss": 1.4321,
       "step": 7
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.0002,
-      "loss": 1.6457,
+      "loss": 1.4738,
       "step": 8
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00019789473684210526,
-      "loss": 1.6226,
+      "loss": 1.7251,
       "step": 9
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00019578947368421054,
-      "loss": 1.643,
+      "loss": 1.6262,
       "step": 10
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.0001936842105263158,
-      "loss": 1.7807,
+      "loss": 1.4947,
       "step": 11
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00019157894736842104,
-      "loss": 1.7257,
+      "loss": 1.8006,
       "step": 12
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00018947368421052632,
-      "loss": 1.6412,
+      "loss": 1.6487,
       "step": 13
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.0001873684210526316,
-      "loss": 1.6792,
+      "loss": 1.7926,
       "step": 14
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00018526315789473685,
-      "loss": 1.4892,
+      "loss": 1.5979,
       "step": 15
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.0001831578947368421,
-      "loss": 1.4418,
+      "loss": 1.6008,
       "step": 16
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00018105263157894739,
-      "loss": 1.5064,
+      "loss": 1.7569,
       "step": 17
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00017894736842105264,
-      "loss": 1.5943,
+      "loss": 1.7647,
       "step": 18
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.0001768421052631579,
-      "loss": 1.419,
+      "loss": 1.7386,
       "step": 19
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00017473684210526317,
-      "loss": 1.5235,
+      "loss": 1.5272,
       "step": 20
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00017263157894736842,
-      "loss": 1.6562,
+      "loss": 1.8591,
       "step": 21
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.0001705263157894737,
-      "loss": 1.4182,
+      "loss": 1.5613,
       "step": 22
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00016842105263157895,
-      "loss": 1.7132,
+      "loss": 1.3797,
       "step": 23
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00016631578947368423,
-      "loss": 1.7307,
+      "loss": 1.3594,
       "step": 24
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00016421052631578948,
-      "loss": 1.7079,
+      "loss": 1.6027,
       "step": 25
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00016210526315789473,
-      "loss": 1.4506,
+      "loss": 1.5134,
       "step": 26
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00016,
-      "loss": 1.673,
+      "loss": 1.7149,
       "step": 27
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00015789473684210527,
-      "loss": 1.5159,
+      "loss": 1.8614,
       "step": 28
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00015578947368421052,
-      "loss": 1.8044,
+      "loss": 1.4819,
       "step": 29
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.0001536842105263158,
-      "loss": 1.311,
+      "loss": 1.623,
       "step": 30
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00015157894736842108,
-      "loss": 1.5876,
+      "loss": 1.7829,
       "step": 31
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00014947368421052633,
-      "loss": 1.5843,
+      "loss": 1.4878,
       "step": 32
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00014736842105263158,
-      "loss": 1.7256,
+      "loss": 1.7675,
       "step": 33
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00014526315789473686,
-      "loss": 1.5446,
+      "loss": 1.6652,
       "step": 34
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.0001431578947368421,
-      "loss": 1.9188,
+      "loss": 1.4829,
       "step": 35
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00014105263157894736,
-      "loss": 1.6284,
+      "loss": 1.6084,
       "step": 36
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00013894736842105264,
-      "loss": 1.6446,
+      "loss": 1.5299,
       "step": 37
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.0001368421052631579,
-      "loss": 1.5255,
+      "loss": 1.5337,
       "step": 38
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00013473684210526317,
-      "loss": 1.4796,
+      "loss": 1.4584,
       "step": 39
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00013263157894736842,
-      "loss": 1.8069,
+      "loss": 1.5648,
       "step": 40
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.0001305263157894737,
-      "loss": 1.5269,
+      "loss": 1.6003,
       "step": 41
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00012842105263157895,
-      "loss": 1.3387,
+      "loss": 1.5679,
       "step": 42
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.0001263157894736842,
-      "loss": 1.3193,
+      "loss": 1.425,
       "step": 43
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00012421052631578949,
-      "loss": 1.6728,
+      "loss": 1.2781,
       "step": 44
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00012210526315789474,
-      "loss": 1.5991,
+      "loss": 1.7008,
       "step": 45
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00012,
-      "loss": 1.6619,
+      "loss": 1.4367,
       "step": 46
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00011789473684210525,
-      "loss": 1.4312,
+      "loss": 1.4628,
       "step": 47
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00011578947368421053,
-      "loss": 1.4846,
+      "loss": 1.6098,
       "step": 48
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.0001136842105263158,
-      "loss": 1.3929,
+      "loss": 1.4618,
       "step": 49
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.00011157894736842105,
-      "loss": 1.647,
+      "loss": 1.3719,
       "step": 50
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 0.00010947368421052633,
-      "loss": 1.5989,
-      "step": 51
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 0.00010736842105263158,
-      "loss": 1.5857,
-      "step": 52
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 0.00010526315789473685,
-      "loss": 1.0283,
-      "step": 53
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 0.00010315789473684211,
-      "loss": 1.6175,
-      "step": 54
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 0.00010105263157894738,
-      "loss": 1.6465,
-      "step": 55
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 9.894736842105263e-05,
-      "loss": 1.6387,
-      "step": 56
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 9.68421052631579e-05,
-      "loss": 1.6037,
-      "step": 57
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 9.473684210526316e-05,
-      "loss": 1.7261,
-      "step": 58
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 9.263157894736843e-05,
-      "loss": 1.6225,
-      "step": 59
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 9.052631578947369e-05,
-      "loss": 1.273,
-      "step": 60
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 8.842105263157894e-05,
-      "loss": 1.5125,
-      "step": 61
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 8.631578947368421e-05,
-      "loss": 1.4634,
-      "step": 62
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 8.421052631578948e-05,
-      "loss": 1.5137,
-      "step": 63
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 8.210526315789474e-05,
-      "loss": 1.6608,
-      "step": 64
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 8e-05,
-      "loss": 1.7165,
-      "step": 65
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 7.789473684210526e-05,
-      "loss": 1.4347,
-      "step": 66
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 7.578947368421054e-05,
-      "loss": 1.9672,
-      "step": 67
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 7.368421052631579e-05,
-      "loss": 1.2761,
-      "step": 68
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 7.157894736842105e-05,
-      "loss": 1.6604,
-      "step": 69
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 6.947368421052632e-05,
-      "loss": 1.3673,
-      "step": 70
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 6.736842105263159e-05,
-      "loss": 1.6446,
-      "step": 71
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 6.526315789473685e-05,
-      "loss": 1.6273,
-      "step": 72
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 6.31578947368421e-05,
-      "loss": 1.657,
-      "step": 73
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 6.105263157894737e-05,
-      "loss": 1.689,
-      "step": 74
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.894736842105263e-05,
-      "loss": 1.5599,
-      "step": 75
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.68421052631579e-05,
-      "loss": 1.3117,
-      "step": 76
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.4736842105263165e-05,
-      "loss": 1.5121,
-      "step": 77
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.2631578947368424e-05,
-      "loss": 1.6346,
-      "step": 78
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.052631578947369e-05,
-      "loss": 1.5391,
-      "step": 79
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 4.842105263157895e-05,
-      "loss": 1.6792,
-      "step": 80
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 4.6315789473684214e-05,
-      "loss": 1.6389,
-      "step": 81
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 4.421052631578947e-05,
-      "loss": 1.6191,
-      "step": 82
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 4.210526315789474e-05,
-      "loss": 1.6297,
-      "step": 83
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 4e-05,
-      "loss": 1.5525,
-      "step": 84
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 3.789473684210527e-05,
-      "loss": 1.4816,
-      "step": 85
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 3.578947368421053e-05,
-      "loss": 1.5453,
-      "step": 86
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 3.368421052631579e-05,
-      "loss": 1.7132,
-      "step": 87
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 3.157894736842105e-05,
-      "loss": 1.7889,
-      "step": 88
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 2.9473684210526314e-05,
-      "loss": 1.3749,
-      "step": 89
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 2.7368421052631583e-05,
-      "loss": 1.337,
-      "step": 90
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 2.5263157894736845e-05,
-      "loss": 1.4863,
-      "step": 91
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 2.3157894736842107e-05,
-      "loss": 1.4914,
-      "step": 92
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 2.105263157894737e-05,
-      "loss": 1.3957,
-      "step": 93
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 1.8947368421052634e-05,
-      "loss": 1.4742,
-      "step": 94
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 1.6842105263157896e-05,
-      "loss": 1.6525,
-      "step": 95
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 1.4736842105263157e-05,
-      "loss": 1.6347,
-      "step": 96
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 1.2631578947368422e-05,
-      "loss": 1.5477,
-      "step": 97
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 1.0526315789473684e-05,
-      "loss": 1.318,
-      "step": 98
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 8.421052631578948e-06,
-      "loss": 1.8458,
-      "step": 99
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 6.315789473684211e-06,
-      "loss": 2.0586,
-      "step": 100
     }
   ],
   "logging_steps": 1,
   "max_steps": 100,
   "num_train_epochs": 1,
-  "save_steps": 25,
-  "total_flos": 5587576318083072.0,
+  "save_steps": 50,
+  "total_flos": 3065802843488256.0,
   "trial_name": null,
   "trial_params": null
 }

last-checkpoint/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:59ea6bf03014c5571a57a6c0be1bd7d4982e07b90a7aa5f37b84f0659c1586a3
-size 4600
+oid sha256:e3b4ec3761ef4757a95a527eb558426cc30befdddfb2e65ccd9b15755214e1aa
+size 6328

last-checkpoint/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)