Upload 13 files

config.json

@@ -0,0 +1,32 @@
+{
+  "_name_or_path": "google/flan-t5-large",
+  "architectures": [
+    "T5ForConditionalGeneration"
+  ],
+  "d_ff": 2816,
+  "d_kv": 64,
+  "d_model": 1024,
+  "decoder_start_token_id": 0,
+  "dense_act_fn": "gelu_new",
+  "dropout_rate": 0.1,
+  "eos_token_id": 1,
+  "feed_forward_proj": "gated-gelu",
+  "initializer_factor": 1.0,
+  "is_encoder_decoder": true,
+  "is_gated_act": true,
+  "layer_norm_epsilon": 1e-06,
+  "model_type": "t5",
+  "n_positions": 512,
+  "num_decoder_layers": 24,
+  "num_heads": 16,
+  "num_layers": 24,
+  "output_past": true,
+  "pad_token_id": 0,
+  "relative_attention_max_distance": 128,
+  "relative_attention_num_buckets": 32,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float32",
+  "transformers_version": "4.23.1",
+  "use_cache": true,
+  "vocab_size": 32128
+}

latest

@@ -0,0 +1 @@
+global_step20000

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:07a7b2314da1293a6edf08d4c4280576bec867516159a6a1fa70312f03e4ee96
+size 3132680487

rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b82c5c48b3bf10a6cf3b54692d100a18ac1a7ae900872a9b6ff2bcf4be4fe6ba
+size 14503

rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1518c9f7b896bf02801902f418346d22bed074b0509c617a03a02058aaac98ee
+size 14503

rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a9bf789e737823a66966c721316ff1213f40424a19a338599c996b1732a0495d
+size 14503

rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fcc3a15749576a21f005703e260037fe24cac1fdf1f5d486a97b55f1e96d09a3
+size 14503

special_tokens_map.json

@@ -0,0 +1,107 @@
+{
+  "additional_special_tokens": [
+    "<extra_id_0>",
+    "<extra_id_1>",
+    "<extra_id_2>",
+    "<extra_id_3>",
+    "<extra_id_4>",
+    "<extra_id_5>",
+    "<extra_id_6>",
+    "<extra_id_7>",
+    "<extra_id_8>",
+    "<extra_id_9>",
+    "<extra_id_10>",
+    "<extra_id_11>",
+    "<extra_id_12>",
+    "<extra_id_13>",
+    "<extra_id_14>",
+    "<extra_id_15>",
+    "<extra_id_16>",
+    "<extra_id_17>",
+    "<extra_id_18>",
+    "<extra_id_19>",
+    "<extra_id_20>",
+    "<extra_id_21>",
+    "<extra_id_22>",
+    "<extra_id_23>",
+    "<extra_id_24>",
+    "<extra_id_25>",
+    "<extra_id_26>",
+    "<extra_id_27>",
+    "<extra_id_28>",
+    "<extra_id_29>",
+    "<extra_id_30>",
+    "<extra_id_31>",
+    "<extra_id_32>",
+    "<extra_id_33>",
+    "<extra_id_34>",
+    "<extra_id_35>",
+    "<extra_id_36>",
+    "<extra_id_37>",
+    "<extra_id_38>",
+    "<extra_id_39>",
+    "<extra_id_40>",
+    "<extra_id_41>",
+    "<extra_id_42>",
+    "<extra_id_43>",
+    "<extra_id_44>",
+    "<extra_id_45>",
+    "<extra_id_46>",
+    "<extra_id_47>",
+    "<extra_id_48>",
+    "<extra_id_49>",
+    "<extra_id_50>",
+    "<extra_id_51>",
+    "<extra_id_52>",
+    "<extra_id_53>",
+    "<extra_id_54>",
+    "<extra_id_55>",
+    "<extra_id_56>",
+    "<extra_id_57>",
+    "<extra_id_58>",
+    "<extra_id_59>",
+    "<extra_id_60>",
+    "<extra_id_61>",
+    "<extra_id_62>",
+    "<extra_id_63>",
+    "<extra_id_64>",
+    "<extra_id_65>",
+    "<extra_id_66>",
+    "<extra_id_67>",
+    "<extra_id_68>",
+    "<extra_id_69>",
+    "<extra_id_70>",
+    "<extra_id_71>",
+    "<extra_id_72>",
+    "<extra_id_73>",
+    "<extra_id_74>",
+    "<extra_id_75>",
+    "<extra_id_76>",
+    "<extra_id_77>",
+    "<extra_id_78>",
+    "<extra_id_79>",
+    "<extra_id_80>",
+    "<extra_id_81>",
+    "<extra_id_82>",
+    "<extra_id_83>",
+    "<extra_id_84>",
+    "<extra_id_85>",
+    "<extra_id_86>",
+    "<extra_id_87>",
+    "<extra_id_88>",
+    "<extra_id_89>",
+    "<extra_id_90>",
+    "<extra_id_91>",
+    "<extra_id_92>",
+    "<extra_id_93>",
+    "<extra_id_94>",
+    "<extra_id_95>",
+    "<extra_id_96>",
+    "<extra_id_97>",
+    "<extra_id_98>",
+    "<extra_id_99>"
+  ],
+  "eos_token": "</s>",
+  "pad_token": "<pad>",
+  "unk_token": "<unk>"
+}

tokenizer_config.json

@@ -0,0 +1,113 @@
+{
+  "additional_special_tokens": [
+    "<extra_id_0>",
+    "<extra_id_1>",
+    "<extra_id_2>",
+    "<extra_id_3>",
+    "<extra_id_4>",
+    "<extra_id_5>",
+    "<extra_id_6>",
+    "<extra_id_7>",
+    "<extra_id_8>",
+    "<extra_id_9>",
+    "<extra_id_10>",
+    "<extra_id_11>",
+    "<extra_id_12>",
+    "<extra_id_13>",
+    "<extra_id_14>",
+    "<extra_id_15>",
+    "<extra_id_16>",
+    "<extra_id_17>",
+    "<extra_id_18>",
+    "<extra_id_19>",
+    "<extra_id_20>",
+    "<extra_id_21>",
+    "<extra_id_22>",
+    "<extra_id_23>",
+    "<extra_id_24>",
+    "<extra_id_25>",
+    "<extra_id_26>",
+    "<extra_id_27>",
+    "<extra_id_28>",
+    "<extra_id_29>",
+    "<extra_id_30>",
+    "<extra_id_31>",
+    "<extra_id_32>",
+    "<extra_id_33>",
+    "<extra_id_34>",
+    "<extra_id_35>",
+    "<extra_id_36>",
+    "<extra_id_37>",
+    "<extra_id_38>",
+    "<extra_id_39>",
+    "<extra_id_40>",
+    "<extra_id_41>",
+    "<extra_id_42>",
+    "<extra_id_43>",
+    "<extra_id_44>",
+    "<extra_id_45>",
+    "<extra_id_46>",
+    "<extra_id_47>",
+    "<extra_id_48>",
+    "<extra_id_49>",
+    "<extra_id_50>",
+    "<extra_id_51>",
+    "<extra_id_52>",
+    "<extra_id_53>",
+    "<extra_id_54>",
+    "<extra_id_55>",
+    "<extra_id_56>",
+    "<extra_id_57>",
+    "<extra_id_58>",
+    "<extra_id_59>",
+    "<extra_id_60>",
+    "<extra_id_61>",
+    "<extra_id_62>",
+    "<extra_id_63>",
+    "<extra_id_64>",
+    "<extra_id_65>",
+    "<extra_id_66>",
+    "<extra_id_67>",
+    "<extra_id_68>",
+    "<extra_id_69>",
+    "<extra_id_70>",
+    "<extra_id_71>",
+    "<extra_id_72>",
+    "<extra_id_73>",
+    "<extra_id_74>",
+    "<extra_id_75>",
+    "<extra_id_76>",
+    "<extra_id_77>",
+    "<extra_id_78>",
+    "<extra_id_79>",
+    "<extra_id_80>",
+    "<extra_id_81>",
+    "<extra_id_82>",
+    "<extra_id_83>",
+    "<extra_id_84>",
+    "<extra_id_85>",
+    "<extra_id_86>",
+    "<extra_id_87>",
+    "<extra_id_88>",
+    "<extra_id_89>",
+    "<extra_id_90>",
+    "<extra_id_91>",
+    "<extra_id_92>",
+    "<extra_id_93>",
+    "<extra_id_94>",
+    "<extra_id_95>",
+    "<extra_id_96>",
+    "<extra_id_97>",
+    "<extra_id_98>",
+    "<extra_id_99>"
+  ],
+  "eos_token": "</s>",
+  "extra_ids": 100,
+  "model_max_length": 512,
+  "name_or_path": "google/flan-t5-large",
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "special_tokens_map_file": "/home/younes_huggingface_co/.cache/huggingface/hub/models--google--t5-v1_1-large/snapshots/314bc112b191ec17b625ba81438dc73d6c23659d/special_tokens_map.json",
+  "tokenizer_class": "T5Tokenizer",
+  "unk_token": "<unk>"
+}

trainer_state.json

@@ -0,0 +1,376 @@
+{
+  "best_metric": 0.436233788728714,
+  "best_model_checkpoint": "flan_large_ft_adam_filtd/checkpoint-20000",
+  "epoch": 8.32639467110741,
+  "global_step": 20000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.8e-05,
+      "loss": 1.6461,
+      "step": 1000
+    },
+    {
+      "epoch": 0.42,
+      "eval_f1": 0.7336931843203959,
+      "eval_loss": 0.543613076210022,
+      "eval_precision": 0.7462491253306569,
+      "eval_recall": 0.7237358282616747,
+      "eval_runtime": 339.9295,
+      "eval_sacrebleu": 5.478519586104642,
+      "eval_samples_per_second": 28.262,
+      "eval_steps_per_second": 0.885,
+      "step": 1000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6e-05,
+      "loss": 0.6266,
+      "step": 2000
+    },
+    {
+      "epoch": 0.83,
+      "eval_f1": 0.7411911502965993,
+      "eval_loss": 0.5126737356185913,
+      "eval_precision": 0.7532233481006616,
+      "eval_recall": 0.7315065660194513,
+      "eval_runtime": 330.4901,
+      "eval_sacrebleu": 5.581437847090348,
+      "eval_samples_per_second": 29.069,
+      "eval_steps_per_second": 0.911,
+      "step": 2000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.4e-05,
+      "loss": 0.591,
+      "step": 3000
+    },
+    {
+      "epoch": 1.25,
+      "eval_f1": 0.744462796992661,
+      "eval_loss": 0.49647408723831177,
+      "eval_precision": 0.7574931879507412,
+      "eval_recall": 0.7338014463765621,
+      "eval_runtime": 328.9086,
+      "eval_sacrebleu": 5.712014723912081,
+      "eval_samples_per_second": 29.209,
+      "eval_steps_per_second": 0.915,
+      "step": 3000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.5723,
+      "step": 4000
+    },
+    {
+      "epoch": 1.67,
+      "eval_f1": 0.7482227764808884,
+      "eval_loss": 0.4846822917461395,
+      "eval_precision": 0.7609444108747153,
+      "eval_recall": 0.7378004755756169,
+      "eval_runtime": 328.5657,
+      "eval_sacrebleu": 5.759016321774072,
+      "eval_samples_per_second": 29.239,
+      "eval_steps_per_second": 0.916,
+      "step": 4000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.5576,
+      "step": 5000
+    },
+    {
+      "epoch": 2.08,
+      "eval_f1": 0.7495765637889841,
+      "eval_loss": 0.4766782820224762,
+      "eval_precision": 0.7626191373513915,
+      "eval_recall": 0.7388611246473127,
+      "eval_runtime": 326.106,
+      "eval_sacrebleu": 5.811597759442428,
+      "eval_samples_per_second": 29.46,
+      "eval_steps_per_second": 0.923,
+      "step": 5000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.8e-05,
+      "loss": 0.5393,
+      "step": 6000
+    },
+    {
+      "epoch": 2.5,
+      "eval_f1": 0.7505085720869012,
+      "eval_loss": 0.47017449140548706,
+      "eval_precision": 0.7641689959470273,
+      "eval_recall": 0.7391936523188186,
+      "eval_runtime": 328.1093,
+      "eval_sacrebleu": 5.885476099157212,
+      "eval_samples_per_second": 29.28,
+      "eval_steps_per_second": 0.917,
+      "step": 6000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 0.5318,
+      "step": 7000
+    },
+    {
+      "epoch": 2.91,
+      "eval_f1": 0.7533312253634614,
+      "eval_loss": 0.46316930651664734,
+      "eval_precision": 0.7659040383844901,
+      "eval_recall": 0.7430121202963726,
+      "eval_runtime": 330.001,
+      "eval_sacrebleu": 5.928167844047436,
+      "eval_samples_per_second": 29.112,
+      "eval_steps_per_second": 0.912,
+      "step": 7000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.4e-05,
+      "loss": 0.514,
+      "step": 8000
+    },
+    {
+      "epoch": 3.33,
+      "eval_f1": 0.7534677800234317,
+      "eval_loss": 0.45805472135543823,
+      "eval_precision": 0.7670932310055877,
+      "eval_recall": 0.7421523856331236,
+      "eval_runtime": 326.6901,
+      "eval_sacrebleu": 5.977964366842188,
+      "eval_samples_per_second": 29.407,
+      "eval_steps_per_second": 0.921,
+      "step": 8000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 0.5084,
+      "step": 9000
+    },
+    {
+      "epoch": 3.75,
+      "eval_f1": 0.7547050486360937,
+      "eval_loss": 0.4548051953315735,
+      "eval_precision": 0.7675773970595853,
+      "eval_recall": 0.7440757734278951,
+      "eval_runtime": 329.4886,
+      "eval_sacrebleu": 5.986154752583561,
+      "eval_samples_per_second": 29.157,
+      "eval_steps_per_second": 0.914,
+      "step": 9000
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2e-05,
+      "loss": 0.4987,
+      "step": 10000
+    },
+    {
+      "epoch": 4.16,
+      "eval_f1": 0.7559209833584902,
+      "eval_loss": 0.45204678177833557,
+      "eval_precision": 0.7687782156203135,
+      "eval_recall": 0.7452970269140299,
+      "eval_runtime": 328.8363,
+      "eval_sacrebleu": 6.112626970123219,
+      "eval_samples_per_second": 29.215,
+      "eval_steps_per_second": 0.915,
+      "step": 10000
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.8e-05,
+      "loss": 0.4916,
+      "step": 11000
+    },
+    {
+      "epoch": 4.58,
+      "eval_f1": 0.7562574939163441,
+      "eval_loss": 0.44847676157951355,
+      "eval_precision": 0.7693036325555866,
+      "eval_recall": 0.7454933088714478,
+      "eval_runtime": 328.9876,
+      "eval_sacrebleu": 6.110669366899824,
+      "eval_samples_per_second": 29.202,
+      "eval_steps_per_second": 0.915,
+      "step": 11000
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.4855,
+      "step": 12000
+    },
+    {
+      "epoch": 5.0,
+      "eval_f1": 0.7573605920082946,
+      "eval_loss": 0.44527965784072876,
+      "eval_precision": 0.7699207649359262,
+      "eval_recall": 0.7470148278161243,
+      "eval_runtime": 326.1238,
+      "eval_sacrebleu": 6.199619598523688,
+      "eval_samples_per_second": 29.458,
+      "eval_steps_per_second": 0.923,
+      "step": 12000
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.4e-05,
+      "loss": 0.4735,
+      "step": 13000
+    },
+    {
+      "epoch": 5.41,
+      "eval_f1": 0.757606573654931,
+      "eval_loss": 0.44323351979255676,
+      "eval_precision": 0.7707262306620629,
+      "eval_recall": 0.7467577414308141,
+      "eval_runtime": 325.2009,
+      "eval_sacrebleu": 6.170302966384815,
+      "eval_samples_per_second": 29.542,
+      "eval_steps_per_second": 0.926,
+      "step": 13000
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.2e-05,
+      "loss": 0.4714,
+      "step": 14000
+    },
+    {
+      "epoch": 5.83,
+      "eval_f1": 0.7582219671704239,
+      "eval_loss": 0.44084230065345764,
+      "eval_precision": 0.7707039175730187,
+      "eval_recall": 0.7479663680666155,
+      "eval_runtime": 328.6225,
+      "eval_sacrebleu": 6.217397957778798,
+      "eval_samples_per_second": 29.234,
+      "eval_steps_per_second": 0.916,
+      "step": 14000
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1e-05,
+      "loss": 0.4619,
+      "step": 15000
+    },
+    {
+      "epoch": 6.24,
+      "eval_f1": 0.7582127043374982,
+      "eval_loss": 0.44005897641181946,
+      "eval_precision": 0.7708874883637588,
+      "eval_recall": 0.7477565708537125,
+      "eval_runtime": 329.0722,
+      "eval_sacrebleu": 6.251480154755987,
+      "eval_samples_per_second": 29.194,
+      "eval_steps_per_second": 0.915,
+      "step": 15000
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.4594,
+      "step": 16000
+    },
+    {
+      "epoch": 6.66,
+      "eval_f1": 0.7590658383431736,
+      "eval_loss": 0.4385643005371094,
+      "eval_precision": 0.7722058984135644,
+      "eval_recall": 0.7481865650026411,
+      "eval_runtime": 329.0887,
+      "eval_sacrebleu": 6.274907412338864,
+      "eval_samples_per_second": 29.193,
+      "eval_steps_per_second": 0.915,
+      "step": 16000
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 6e-06,
+      "loss": 0.4548,
+      "step": 17000
+    },
+    {
+      "epoch": 7.08,
+      "eval_f1": 0.7591303865470561,
+      "eval_loss": 0.4375361502170563,
+      "eval_precision": 0.7716383277265539,
+      "eval_recall": 0.7488526040952809,
+      "eval_runtime": 326.0628,
+      "eval_sacrebleu": 6.2691323707280056,
+      "eval_samples_per_second": 29.464,
+      "eval_steps_per_second": 0.923,
+      "step": 17000
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.4496,
+      "step": 18000
+    },
+    {
+      "epoch": 7.49,
+      "eval_f1": 0.7595453337276964,
+      "eval_loss": 0.4368315041065216,
+      "eval_precision": 0.7720739444758674,
+      "eval_recall": 0.7492328676017723,
+      "eval_runtime": 326.1933,
+      "eval_sacrebleu": 6.332761567803722,
+      "eval_samples_per_second": 29.452,
+      "eval_steps_per_second": 0.923,
+      "step": 18000
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.4484,
+      "step": 19000
+    },
+    {
+      "epoch": 7.91,
+      "eval_f1": 0.7595180715794393,
+      "eval_loss": 0.4362909495830536,
+      "eval_precision": 0.7722789199881863,
+      "eval_recall": 0.7489893756232178,
+      "eval_runtime": 326.3639,
+      "eval_sacrebleu": 6.311690383163712,
+      "eval_samples_per_second": 29.436,
+      "eval_steps_per_second": 0.922,
+      "step": 19000
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 0.0,
+      "loss": 0.4446,
+      "step": 20000
+    },
+    {
+      "epoch": 8.33,
+      "eval_f1": 0.7592586470783173,
+      "eval_loss": 0.436233788728714,
+      "eval_precision": 0.7721431963591666,
+      "eval_recall": 0.7486093006745032,
+      "eval_runtime": 326.7518,
+      "eval_sacrebleu": 6.312650510263652,
+      "eval_samples_per_second": 29.402,
+      "eval_steps_per_second": 0.921,
+      "step": 20000
+    }
+  ],
+  "max_steps": 20000,
+  "num_train_epochs": 9,
+  "total_flos": 2.033175326799954e+17,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4fb7b4bbb84d900928d8580e142e5dfdadda232b8b027f8de41a7f202adc7585
+size 4143

zero_to_fp32.py

@@ -0,0 +1,482 @@
+#!/usr/bin/env python
+
+# This script extracts fp32 consolidated weights from a zero 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
+
+# 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)
+
+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_optim_files(checkpoint_dir):
+    # XXX: need to test that this simple glob rule works for multi-node setup too
+    optim_files = sorted(glob.glob(os.path.join(checkpoint_dir,
+                                                "*_optim_states.pt")),
+                         key=natural_keys)
+
+    if len(optim_files) == 0:
+        raise FileNotFoundError(
+            f"can't find '*_optim_states.pt' files in directory '{checkpoint_dir}'")
+
+    return optim_files
+
+
+def parse_model_state(file):
+    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]
+
+    ds_version = state_dict.get(DS_VERSION, None)
+
+    return buffers, param_shapes, ds_version
+
+
+def parse_optim_states(files, ds_checkpoint_dir):
+
+    total_files = len(files)
+    state_dicts = []
+    for f in files:
+        state_dicts.append(torch.load(f, map_location=device))
+
+    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_file = get_model_state_file(ds_checkpoint_dir, zero_stage)
+    buffers, param_shapes, ds_version = parse_model_state(model_file)
+    print(f'Parsing checkpoint created by deepspeed=={ds_version}')
+
+    if zero_stage == 2:
+        return _get_fp32_state_dict_from_zero2_checkpoint(world_size,
+                                                          param_shapes,
+                                                          fp32_flat_groups,
+                                                          buffers)
+    elif zero_stage == 3:
+        return _get_fp32_state_dict_from_zero3_checkpoint(world_size,
+                                                          param_shapes,
+                                                          fp32_flat_groups,
+                                                          buffers)
+
+
+def _get_fp32_state_dict_from_zero2_checkpoint(world_size,
+                                               param_shapes,
+                                               fp32_flat_groups,
+                                               buffers):
+
+    # 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.")
+
+    state_dict = OrderedDict()
+
+    # buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    # 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"
+    )
+
+    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 _get_fp32_state_dict_from_zero3_checkpoint(world_size,
+                                               param_shapes,
+                                               fp32_flat_groups,
+                                               buffers):
+
+    # 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
+
+    avail_numel = fp32_flat_groups[0].numel() * world_size
+    # 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
+        print(f"Have {avail_numel} numels to process.")
+        print(f"Need {wanted_numel} numels in {wanted_params} params.")
+
+    state_dict = OrderedDict()
+
+    # buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    # 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"{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 fp32 state dict with {total_params} params {total_numel} elements"
+    )
+
+    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("-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)