Upload folder using huggingface_hub

137_128_e3_3e-5/adapter_config.json

@@ -0,0 +1,39 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "ibm-granite/granite-3.3-8b-base",
+  "bias": "none",
+  "corda_config": null,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 256,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 128,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "gate_proj",
+    "k_proj",
+    "q_proj",
+    "up_proj",
+    "v_proj",
+    "down_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

137_128_e3_3e-5/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05a85849cdc6ce6e35405c24f4057c07b8b655fe4da492d0b8fc1dd268026faa
+size 791751704

137_128_e3_3e-5/latest

@@ -0,0 +1 @@
+global_step960

137_128_e3_3e-5/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:88bee1a445d800f1130050f3eb39063cc6ed70ae338fd565238ed930e14dfab1
+size 16389

137_128_e3_3e-5/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:158b289230ae2aaa5dbd872dc3bcf30f2bd3cf028d0bcb052ed8ea945d8a3bc0
+size 16389

137_128_e3_3e-5/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:605960a4a8ea68a24c65e733bc129d0dd915becf89bc3379904ed9d69ac42ff5
+size 16389

137_128_e3_3e-5/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6a5d26f1d91f759d96d4fbd2f74f8bdea8ea814176c19e133ead716e55d3b7cc
+size 16389

137_128_e3_3e-5/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a7002336484b3b69154c9ad5a7128e5ed7b0ad4c8be18d9f37791c182323ac9e
+size 16389

137_128_e3_3e-5/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e7e213785ec5afcdbfaee78bfc43132aa7d438babed7cc5936969d4ac55f4d1
+size 16389

137_128_e3_3e-5/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6f7bfc666f569b37e26488c0da594c41e17ead8d7b3c56607ebffa96149ad0aa
+size 16389

137_128_e3_3e-5/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:38d8f5bf493ec4f6804565623000388b516c210bf065ab782a7c5d54b0bdbd2b
+size 16389

137_128_e3_3e-5/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:77ab7cf68795346011ba3a2b15a76ada5548bd20e9ea277828180dbdf286ff77
+size 1401

137_128_e3_3e-5/special_tokens_map.json

@@ -0,0 +1,45 @@
+{
+  "additional_special_tokens": [
+    "<|endoftext|>",
+    "<fim_prefix>",
+    "<fim_middle>",
+    "<fim_suffix>",
+    "<fim_pad>",
+    "<filename>",
+    "<gh_stars>",
+    "<issue_start>",
+    "<issue_comment>",
+    "<issue_closed>",
+    "<jupyter_start>",
+    "<jupyter_text>",
+    "<jupyter_code>",
+    "<jupyter_output>",
+    "<empty_output>",
+    "<commit_before>",
+    "<commit_msg>",
+    "<commit_after>",
+    "<reponame>"
+  ],
+  "bos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "<reponame>",
+  "unk_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

137_128_e3_3e-5/tokenizer_config.json

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

137_128_e3_3e-5/trainer_state.json

@@ -0,0 +1,1378 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 960,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01564945226917058,
+      "grad_norm": 1.0992707014083862,
+      "learning_rate": 2.4999999999999998e-06,
+      "loss": 1.4312,
+      "step": 5
+    },
+    {
+      "epoch": 0.03129890453834116,
+      "grad_norm": 0.8302562832832336,
+      "learning_rate": 5.625e-06,
+      "loss": 1.416,
+      "step": 10
+    },
+    {
+      "epoch": 0.046948356807511735,
+      "grad_norm": 0.6538288593292236,
+      "learning_rate": 8.750000000000001e-06,
+      "loss": 1.396,
+      "step": 15
+    },
+    {
+      "epoch": 0.06259780907668232,
+      "grad_norm": 0.6198179721832275,
+      "learning_rate": 1.1874999999999999e-05,
+      "loss": 1.4423,
+      "step": 20
+    },
+    {
+      "epoch": 0.0782472613458529,
+      "grad_norm": 0.5694576501846313,
+      "learning_rate": 1.5e-05,
+      "loss": 1.392,
+      "step": 25
+    },
+    {
+      "epoch": 0.09389671361502347,
+      "grad_norm": 0.6703488230705261,
+      "learning_rate": 1.8125e-05,
+      "loss": 1.3493,
+      "step": 30
+    },
+    {
+      "epoch": 0.10954616588419405,
+      "grad_norm": 0.6377682685852051,
+      "learning_rate": 2.125e-05,
+      "loss": 1.2945,
+      "step": 35
+    },
+    {
+      "epoch": 0.12519561815336464,
+      "grad_norm": 0.6265494227409363,
+      "learning_rate": 2.4375e-05,
+      "loss": 1.3379,
+      "step": 40
+    },
+    {
+      "epoch": 0.14084507042253522,
+      "grad_norm": 0.6431508660316467,
+      "learning_rate": 2.75e-05,
+      "loss": 1.3005,
+      "step": 45
+    },
+    {
+      "epoch": 0.1564945226917058,
+      "grad_norm": 0.5961695909500122,
+      "learning_rate": 2.9999911003914892e-05,
+      "loss": 1.3135,
+      "step": 50
+    },
+    {
+      "epoch": 0.17214397496087636,
+      "grad_norm": 0.6352348327636719,
+      "learning_rate": 2.9996796251818968e-05,
+      "loss": 1.2121,
+      "step": 55
+    },
+    {
+      "epoch": 0.18779342723004694,
+      "grad_norm": 0.7207192182540894,
+      "learning_rate": 2.9989232751451756e-05,
+      "loss": 1.1659,
+      "step": 60
+    },
+    {
+      "epoch": 0.20344287949921752,
+      "grad_norm": 0.6068947911262512,
+      "learning_rate": 2.997722274649974e-05,
+      "loss": 1.1818,
+      "step": 65
+    },
+    {
+      "epoch": 0.2190923317683881,
+      "grad_norm": 0.6064233183860779,
+      "learning_rate": 2.9960769799689797e-05,
+      "loss": 1.1812,
+      "step": 70
+    },
+    {
+      "epoch": 0.2347417840375587,
+      "grad_norm": 0.6623309850692749,
+      "learning_rate": 2.9939878791732264e-05,
+      "loss": 1.1664,
+      "step": 75
+    },
+    {
+      "epoch": 0.25039123630672927,
+      "grad_norm": 0.6840605139732361,
+      "learning_rate": 2.991455591987314e-05,
+      "loss": 1.1066,
+      "step": 80
+    },
+    {
+      "epoch": 0.26604068857589985,
+      "grad_norm": 0.7102366089820862,
+      "learning_rate": 2.9884808696055675e-05,
+      "loss": 1.063,
+      "step": 85
+    },
+    {
+      "epoch": 0.28169014084507044,
+      "grad_norm": 0.8259777426719666,
+      "learning_rate": 2.985064594469199e-05,
+      "loss": 1.042,
+      "step": 90
+    },
+    {
+      "epoch": 0.297339593114241,
+      "grad_norm": 0.7813456058502197,
+      "learning_rate": 2.9812077800045325e-05,
+      "loss": 1.017,
+      "step": 95
+    },
+    {
+      "epoch": 0.3129890453834116,
+      "grad_norm": 0.9963333606719971,
+      "learning_rate": 2.976911570322377e-05,
+      "loss": 1.0337,
+      "step": 100
+    },
+    {
+      "epoch": 0.3286384976525822,
+      "grad_norm": 0.9410939812660217,
+      "learning_rate": 2.972177239878627e-05,
+      "loss": 0.9902,
+      "step": 105
+    },
+    {
+      "epoch": 0.3442879499217527,
+      "grad_norm": 0.8461155295372009,
+      "learning_rate": 2.967006193096203e-05,
+      "loss": 0.9514,
+      "step": 110
+    },
+    {
+      "epoch": 0.3599374021909233,
+      "grad_norm": 0.9352841377258301,
+      "learning_rate": 2.9613999639484314e-05,
+      "loss": 0.9753,
+      "step": 115
+    },
+    {
+      "epoch": 0.3755868544600939,
+      "grad_norm": 1.0986818075180054,
+      "learning_rate": 2.9553602155039975e-05,
+      "loss": 0.8761,
+      "step": 120
+    },
+    {
+      "epoch": 0.39123630672926446,
+      "grad_norm": 1.0539556741714478,
+      "learning_rate": 2.9488887394336025e-05,
+      "loss": 0.8724,
+      "step": 125
+    },
+    {
+      "epoch": 0.40688575899843504,
+      "grad_norm": 1.0686397552490234,
+      "learning_rate": 2.9419874554784695e-05,
+      "loss": 0.8787,
+      "step": 130
+    },
+    {
+      "epoch": 0.4225352112676056,
+      "grad_norm": 0.989229679107666,
+      "learning_rate": 2.934658410880859e-05,
+      "loss": 0.8718,
+      "step": 135
+    },
+    {
+      "epoch": 0.4381846635367762,
+      "grad_norm": 1.033861517906189,
+      "learning_rate": 2.926903779776763e-05,
+      "loss": 0.8021,
+      "step": 140
+    },
+    {
+      "epoch": 0.4538341158059468,
+      "grad_norm": 1.1175507307052612,
+      "learning_rate": 2.9187258625509518e-05,
+      "loss": 0.8256,
+      "step": 145
+    },
+    {
+      "epoch": 0.4694835680751174,
+      "grad_norm": 1.08514404296875,
+      "learning_rate": 2.9101270851545776e-05,
+      "loss": 0.8586,
+      "step": 150
+    },
+    {
+      "epoch": 0.48513302034428796,
+      "grad_norm": 1.1544030904769897,
+      "learning_rate": 2.901109998385518e-05,
+      "loss": 0.8282,
+      "step": 155
+    },
+    {
+      "epoch": 0.5007824726134585,
+      "grad_norm": 1.4994546175003052,
+      "learning_rate": 2.8916772771316974e-05,
+      "loss": 0.7327,
+      "step": 160
+    },
+    {
+      "epoch": 0.5164319248826291,
+      "grad_norm": 1.1944866180419922,
+      "learning_rate": 2.8818317195775848e-05,
+      "loss": 0.7587,
+      "step": 165
+    },
+    {
+      "epoch": 0.5320813771517997,
+      "grad_norm": 1.2246421575546265,
+      "learning_rate": 2.871576246374126e-05,
+      "loss": 0.73,
+      "step": 170
+    },
+    {
+      "epoch": 0.5477308294209703,
+      "grad_norm": 1.1955385208129883,
+      "learning_rate": 2.86091389977234e-05,
+      "loss": 0.7319,
+      "step": 175
+    },
+    {
+      "epoch": 0.5633802816901409,
+      "grad_norm": 1.2465507984161377,
+      "learning_rate": 2.849847842720847e-05,
+      "loss": 0.7343,
+      "step": 180
+    },
+    {
+      "epoch": 0.5790297339593115,
+      "grad_norm": 1.2358489036560059,
+      "learning_rate": 2.8383813579275912e-05,
+      "loss": 0.6876,
+      "step": 185
+    },
+    {
+      "epoch": 0.594679186228482,
+      "grad_norm": 1.2837319374084473,
+      "learning_rate": 2.826517846886033e-05,
+      "loss": 0.6978,
+      "step": 190
+    },
+    {
+      "epoch": 0.6103286384976526,
+      "grad_norm": 1.3210774660110474,
+      "learning_rate": 2.8142608288661133e-05,
+      "loss": 0.6167,
+      "step": 195
+    },
+    {
+      "epoch": 0.6259780907668232,
+      "grad_norm": 1.4741657972335815,
+      "learning_rate": 2.801613939870269e-05,
+      "loss": 0.6672,
+      "step": 200
+    },
+    {
+      "epoch": 0.6416275430359938,
+      "grad_norm": 1.340415596961975,
+      "learning_rate": 2.788580931554828e-05,
+      "loss": 0.6761,
+      "step": 205
+    },
+    {
+      "epoch": 0.6572769953051644,
+      "grad_norm": 1.5497334003448486,
+      "learning_rate": 2.7751656701170923e-05,
+      "loss": 0.6435,
+      "step": 210
+    },
+    {
+      "epoch": 0.672926447574335,
+      "grad_norm": 1.2098582983016968,
+      "learning_rate": 2.7613721351484457e-05,
+      "loss": 0.5881,
+      "step": 215
+    },
+    {
+      "epoch": 0.6885758998435054,
+      "grad_norm": 1.4001708030700684,
+      "learning_rate": 2.747204418453818e-05,
+      "loss": 0.5771,
+      "step": 220
+    },
+    {
+      "epoch": 0.704225352112676,
+      "grad_norm": 1.3430523872375488,
+      "learning_rate": 2.7326667228378677e-05,
+      "loss": 0.5636,
+      "step": 225
+    },
+    {
+      "epoch": 0.7198748043818466,
+      "grad_norm": 1.7932277917861938,
+      "learning_rate": 2.7177633608582347e-05,
+      "loss": 0.5389,
+      "step": 230
+    },
+    {
+      "epoch": 0.7355242566510172,
+      "grad_norm": 1.3696084022521973,
+      "learning_rate": 2.7024987535462327e-05,
+      "loss": 0.562,
+      "step": 235
+    },
+    {
+      "epoch": 0.7511737089201878,
+      "grad_norm": 1.4317994117736816,
+      "learning_rate": 2.686877429095368e-05,
+      "loss": 0.4808,
+      "step": 240
+    },
+    {
+      "epoch": 0.7668231611893583,
+      "grad_norm": 1.342526912689209,
+      "learning_rate": 2.6709040215180633e-05,
+      "loss": 0.5642,
+      "step": 245
+    },
+    {
+      "epoch": 0.7824726134585289,
+      "grad_norm": 1.3292914628982544,
+      "learning_rate": 2.6545832692709968e-05,
+      "loss": 0.5181,
+      "step": 250
+    },
+    {
+      "epoch": 0.7981220657276995,
+      "grad_norm": 1.4693751335144043,
+      "learning_rate": 2.6379200138494542e-05,
+      "loss": 0.5534,
+      "step": 255
+    },
+    {
+      "epoch": 0.8137715179968701,
+      "grad_norm": 1.4425127506256104,
+      "learning_rate": 2.620919198351114e-05,
+      "loss": 0.4958,
+      "step": 260
+    },
+    {
+      "epoch": 0.8294209702660407,
+      "grad_norm": 1.8783156871795654,
+      "learning_rate": 2.6035858660096975e-05,
+      "loss": 0.4763,
+      "step": 265
+    },
+    {
+      "epoch": 0.8450704225352113,
+      "grad_norm": 1.361272931098938,
+      "learning_rate": 2.5859251586989087e-05,
+      "loss": 0.4796,
+      "step": 270
+    },
+    {
+      "epoch": 0.8607198748043818,
+      "grad_norm": 1.470905065536499,
+      "learning_rate": 2.5679423154071146e-05,
+      "loss": 0.4855,
+      "step": 275
+    },
+    {
+      "epoch": 0.8763693270735524,
+      "grad_norm": 1.5775848627090454,
+      "learning_rate": 2.5496426706832196e-05,
+      "loss": 0.4944,
+      "step": 280
+    },
+    {
+      "epoch": 0.892018779342723,
+      "grad_norm": 1.4940083026885986,
+      "learning_rate": 2.5310316530541883e-05,
+      "loss": 0.433,
+      "step": 285
+    },
+    {
+      "epoch": 0.9076682316118936,
+      "grad_norm": 1.570082426071167,
+      "learning_rate": 2.5121147834146907e-05,
+      "loss": 0.464,
+      "step": 290
+    },
+    {
+      "epoch": 0.9233176838810642,
+      "grad_norm": 1.4290874004364014,
+      "learning_rate": 2.49289767338935e-05,
+      "loss": 0.4387,
+      "step": 295
+    },
+    {
+      "epoch": 0.9389671361502347,
+      "grad_norm": 1.547577142715454,
+      "learning_rate": 2.473386023668071e-05,
+      "loss": 0.4396,
+      "step": 300
+    },
+    {
+      "epoch": 0.9546165884194053,
+      "grad_norm": 1.7308681011199951,
+      "learning_rate": 2.4535856223149525e-05,
+      "loss": 0.3664,
+      "step": 305
+    },
+    {
+      "epoch": 0.9702660406885759,
+      "grad_norm": 1.6065853834152222,
+      "learning_rate": 2.433502343051274e-05,
+      "loss": 0.4005,
+      "step": 310
+    },
+    {
+      "epoch": 0.9859154929577465,
+      "grad_norm": 1.2723562717437744,
+      "learning_rate": 2.413142143513081e-05,
+      "loss": 0.4245,
+      "step": 315
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.0263631343841553,
+      "learning_rate": 2.39251106348387e-05,
+      "loss": 0.37,
+      "step": 320
+    },
+    {
+      "epoch": 1.0156494522691706,
+      "grad_norm": 1.5915229320526123,
+      "learning_rate": 2.3716152231029077e-05,
+      "loss": 0.3363,
+      "step": 325
+    },
+    {
+      "epoch": 1.0312989045383412,
+      "grad_norm": 1.3393783569335938,
+      "learning_rate": 2.3504608210497124e-05,
+      "loss": 0.3029,
+      "step": 330
+    },
+    {
+      "epoch": 1.0469483568075117,
+      "grad_norm": 1.3689452409744263,
+      "learning_rate": 2.3290541327052373e-05,
+      "loss": 0.2984,
+      "step": 335
+    },
+    {
+      "epoch": 1.0625978090766823,
+      "grad_norm": 1.4672577381134033,
+      "learning_rate": 2.307401508290302e-05,
+      "loss": 0.298,
+      "step": 340
+    },
+    {
+      "epoch": 1.078247261345853,
+      "grad_norm": 1.354447603225708,
+      "learning_rate": 2.285509370981817e-05,
+      "loss": 0.313,
+      "step": 345
+    },
+    {
+      "epoch": 1.0938967136150235,
+      "grad_norm": 1.8843916654586792,
+      "learning_rate": 2.2633842150073745e-05,
+      "loss": 0.308,
+      "step": 350
+    },
+    {
+      "epoch": 1.109546165884194,
+      "grad_norm": 1.2974810600280762,
+      "learning_rate": 2.241032603718756e-05,
+      "loss": 0.3137,
+      "step": 355
+    },
+    {
+      "epoch": 1.1251956181533647,
+      "grad_norm": 1.7436259984970093,
+      "learning_rate": 2.2184611676449366e-05,
+      "loss": 0.2781,
+      "step": 360
+    },
+    {
+      "epoch": 1.1408450704225352,
+      "grad_norm": 1.3938193321228027,
+      "learning_rate": 2.1956766025251607e-05,
+      "loss": 0.3377,
+      "step": 365
+    },
+    {
+      "epoch": 1.1564945226917058,
+      "grad_norm": 1.5196208953857422,
+      "learning_rate": 2.1726856673226764e-05,
+      "loss": 0.3041,
+      "step": 370
+    },
+    {
+      "epoch": 1.1721439749608764,
+      "grad_norm": 1.400933027267456,
+      "learning_rate": 2.149495182219708e-05,
+      "loss": 0.2837,
+      "step": 375
+    },
+    {
+      "epoch": 1.187793427230047,
+      "grad_norm": 1.3604736328125,
+      "learning_rate": 2.126112026594277e-05,
+      "loss": 0.2692,
+      "step": 380
+    },
+    {
+      "epoch": 1.2034428794992176,
+      "grad_norm": 1.651373267173767,
+      "learning_rate": 2.1025431369794546e-05,
+      "loss": 0.2978,
+      "step": 385
+    },
+    {
+      "epoch": 1.2190923317683882,
+      "grad_norm": 1.2913239002227783,
+      "learning_rate": 2.0787955050056634e-05,
+      "loss": 0.2683,
+      "step": 390
+    },
+    {
+      "epoch": 1.2347417840375587,
+      "grad_norm": 1.672248363494873,
+      "learning_rate": 2.054876175326637e-05,
+      "loss": 0.2806,
+      "step": 395
+    },
+    {
+      "epoch": 1.2503912363067293,
+      "grad_norm": 1.2574130296707153,
+      "learning_rate": 2.0307922435296445e-05,
+      "loss": 0.2871,
+      "step": 400
+    },
+    {
+      "epoch": 1.2660406885759,
+      "grad_norm": 1.8362149000167847,
+      "learning_rate": 2.006550854030609e-05,
+      "loss": 0.2697,
+      "step": 405
+    },
+    {
+      "epoch": 1.2816901408450705,
+      "grad_norm": 1.4573137760162354,
+      "learning_rate": 1.9821591979547425e-05,
+      "loss": 0.2224,
+      "step": 410
+    },
+    {
+      "epoch": 1.297339593114241,
+      "grad_norm": 1.3990845680236816,
+      "learning_rate": 1.9576245110033233e-05,
+      "loss": 0.218,
+      "step": 415
+    },
+    {
+      "epoch": 1.3129890453834117,
+      "grad_norm": 1.4442147016525269,
+      "learning_rate": 1.9329540713072482e-05,
+      "loss": 0.2514,
+      "step": 420
+    },
+    {
+      "epoch": 1.3286384976525822,
+      "grad_norm": 1.3912596702575684,
+      "learning_rate": 1.9081551972680025e-05,
+      "loss": 0.2709,
+      "step": 425
+    },
+    {
+      "epoch": 1.3442879499217528,
+      "grad_norm": 1.3834530115127563,
+      "learning_rate": 1.883235245386678e-05,
+      "loss": 0.2438,
+      "step": 430
+    },
+    {
+      "epoch": 1.3599374021909232,
+      "grad_norm": 1.5764304399490356,
+      "learning_rate": 1.858201608081695e-05,
+      "loss": 0.2551,
+      "step": 435
+    },
+    {
+      "epoch": 1.375586854460094,
+      "grad_norm": 1.5206382274627686,
+      "learning_rate": 1.8330617114958648e-05,
+      "loss": 0.2586,
+      "step": 440
+    },
+    {
+      "epoch": 1.3912363067292643,
+      "grad_norm": 1.3278956413269043,
+      "learning_rate": 1.8078230132934514e-05,
+      "loss": 0.2039,
+      "step": 445
+    },
+    {
+      "epoch": 1.4068857589984352,
+      "grad_norm": 1.231637716293335,
+      "learning_rate": 1.78249300044788e-05,
+      "loss": 0.2205,
+      "step": 450
+    },
+    {
+      "epoch": 1.4225352112676055,
+      "grad_norm": 1.3460469245910645,
+      "learning_rate": 1.7570791870207518e-05,
+      "loss": 0.2047,
+      "step": 455
+    },
+    {
+      "epoch": 1.4381846635367763,
+      "grad_norm": 1.4266237020492554,
+      "learning_rate": 1.7315891119328233e-05,
+      "loss": 0.1765,
+      "step": 460
+    },
+    {
+      "epoch": 1.4538341158059467,
+      "grad_norm": 1.191825270652771,
+      "learning_rate": 1.7060303367276123e-05,
+      "loss": 0.197,
+      "step": 465
+    },
+    {
+      "epoch": 1.4694835680751175,
+      "grad_norm": 1.285810112953186,
+      "learning_rate": 1.6804104433282914e-05,
+      "loss": 0.2231,
+      "step": 470
+    },
+    {
+      "epoch": 1.4851330203442878,
+      "grad_norm": 1.6323802471160889,
+      "learning_rate": 1.6547370317885358e-05,
+      "loss": 0.2183,
+      "step": 475
+    },
+    {
+      "epoch": 1.5007824726134587,
+      "grad_norm": 1.278344750404358,
+      "learning_rate": 1.629017718037993e-05,
+      "loss": 0.1975,
+      "step": 480
+    },
+    {
+      "epoch": 1.516431924882629,
+      "grad_norm": 1.2715482711791992,
+      "learning_rate": 1.6032601316230466e-05,
+      "loss": 0.1864,
+      "step": 485
+    },
+    {
+      "epoch": 1.5320813771517998,
+      "grad_norm": 1.5322340726852417,
+      "learning_rate": 1.577471913443532e-05,
+      "loss": 0.2181,
+      "step": 490
+    },
+    {
+      "epoch": 1.5477308294209702,
+      "grad_norm": 1.3825780153274536,
+      "learning_rate": 1.5516607134860956e-05,
+      "loss": 0.1736,
+      "step": 495
+    },
+    {
+      "epoch": 1.563380281690141,
+      "grad_norm": 1.4987000226974487,
+      "learning_rate": 1.5258341885548474e-05,
+      "loss": 0.1838,
+      "step": 500
+    },
+    {
+      "epoch": 1.5790297339593113,
+      "grad_norm": 1.4554896354675293,
+      "learning_rate": 1.5e-05,
+      "loss": 0.1682,
+      "step": 505
+    },
+    {
+      "epoch": 1.5946791862284821,
+      "grad_norm": 1.4363731145858765,
+      "learning_rate": 1.4741658114451525e-05,
+      "loss": 0.1667,
+      "step": 510
+    },
+    {
+      "epoch": 1.6103286384976525,
+      "grad_norm": 1.2509009838104248,
+      "learning_rate": 1.4483392865139048e-05,
+      "loss": 0.1646,
+      "step": 515
+    },
+    {
+      "epoch": 1.6259780907668233,
+      "grad_norm": 1.5473783016204834,
+      "learning_rate": 1.422528086556468e-05,
+      "loss": 0.1807,
+      "step": 520
+    },
+    {
+      "epoch": 1.6416275430359937,
+      "grad_norm": 1.3637259006500244,
+      "learning_rate": 1.3967398683769543e-05,
+      "loss": 0.1755,
+      "step": 525
+    },
+    {
+      "epoch": 1.6572769953051645,
+      "grad_norm": 1.6053063869476318,
+      "learning_rate": 1.3709822819620072e-05,
+      "loss": 0.158,
+      "step": 530
+    },
+    {
+      "epoch": 1.6729264475743348,
+      "grad_norm": 1.5469564199447632,
+      "learning_rate": 1.3452629682114646e-05,
+      "loss": 0.1894,
+      "step": 535
+    },
+    {
+      "epoch": 1.6885758998435054,
+      "grad_norm": 1.249707818031311,
+      "learning_rate": 1.319589556671709e-05,
+      "loss": 0.17,
+      "step": 540
+    },
+    {
+      "epoch": 1.704225352112676,
+      "grad_norm": 1.175848364830017,
+      "learning_rate": 1.2939696632723877e-05,
+      "loss": 0.1608,
+      "step": 545
+    },
+    {
+      "epoch": 1.7198748043818466,
+      "grad_norm": 1.4929946660995483,
+      "learning_rate": 1.2684108880671773e-05,
+      "loss": 0.1602,
+      "step": 550
+    },
+    {
+      "epoch": 1.7355242566510172,
+      "grad_norm": 1.2490898370742798,
+      "learning_rate": 1.2429208129792488e-05,
+      "loss": 0.1701,
+      "step": 555
+    },
+    {
+      "epoch": 1.7511737089201878,
+      "grad_norm": 1.461625099182129,
+      "learning_rate": 1.21750699955212e-05,
+      "loss": 0.1469,
+      "step": 560
+    },
+    {
+      "epoch": 1.7668231611893583,
+      "grad_norm": 1.2867664098739624,
+      "learning_rate": 1.1921769867065487e-05,
+      "loss": 0.1495,
+      "step": 565
+    },
+    {
+      "epoch": 1.782472613458529,
+      "grad_norm": 1.3397834300994873,
+      "learning_rate": 1.1669382885041353e-05,
+      "loss": 0.1479,
+      "step": 570
+    },
+    {
+      "epoch": 1.7981220657276995,
+      "grad_norm": 1.3621459007263184,
+      "learning_rate": 1.1417983919183056e-05,
+      "loss": 0.1576,
+      "step": 575
+    },
+    {
+      "epoch": 1.81377151799687,
+      "grad_norm": 1.152160406112671,
+      "learning_rate": 1.1167647546133222e-05,
+      "loss": 0.1278,
+      "step": 580
+    },
+    {
+      "epoch": 1.8294209702660407,
+      "grad_norm": 1.2958347797393799,
+      "learning_rate": 1.0918448027319972e-05,
+      "loss": 0.1235,
+      "step": 585
+    },
+    {
+      "epoch": 1.8450704225352113,
+      "grad_norm": 1.6260055303573608,
+      "learning_rate": 1.0670459286927517e-05,
+      "loss": 0.1558,
+      "step": 590
+    },
+    {
+      "epoch": 1.8607198748043818,
+      "grad_norm": 1.293577790260315,
+      "learning_rate": 1.042375488996677e-05,
+      "loss": 0.1558,
+      "step": 595
+    },
+    {
+      "epoch": 1.8763693270735524,
+      "grad_norm": 1.1326173543930054,
+      "learning_rate": 1.017840802045258e-05,
+      "loss": 0.1441,
+      "step": 600
+    },
+    {
+      "epoch": 1.892018779342723,
+      "grad_norm": 1.125647783279419,
+      "learning_rate": 9.934491459693916e-06,
+      "loss": 0.1255,
+      "step": 605
+    },
+    {
+      "epoch": 1.9076682316118936,
+      "grad_norm": 1.032149076461792,
+      "learning_rate": 9.692077564703556e-06,
+      "loss": 0.1295,
+      "step": 610
+    },
+    {
+      "epoch": 1.9233176838810642,
+      "grad_norm": 1.2166255712509155,
+      "learning_rate": 9.451238246733632e-06,
+      "loss": 0.1495,
+      "step": 615
+    },
+    {
+      "epoch": 1.9389671361502347,
+      "grad_norm": 1.255165696144104,
+      "learning_rate": 9.212044949943364e-06,
+      "loss": 0.126,
+      "step": 620
+    },
+    {
+      "epoch": 1.9546165884194053,
+      "grad_norm": 1.1294087171554565,
+      "learning_rate": 8.974568630205462e-06,
+      "loss": 0.1301,
+      "step": 625
+    },
+    {
+      "epoch": 1.970266040688576,
+      "grad_norm": 1.3508412837982178,
+      "learning_rate": 8.73887973405723e-06,
+      "loss": 0.1187,
+      "step": 630
+    },
+    {
+      "epoch": 1.9859154929577465,
+      "grad_norm": 1.0194400548934937,
+      "learning_rate": 8.50504817780292e-06,
+      "loss": 0.1026,
+      "step": 635
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.3527121543884277,
+      "learning_rate": 8.273143326773242e-06,
+      "loss": 0.1172,
+      "step": 640
+    },
+    {
+      "epoch": 2.0156494522691704,
+      "grad_norm": 0.9387474060058594,
+      "learning_rate": 8.043233974748392e-06,
+      "loss": 0.0957,
+      "step": 645
+    },
+    {
+      "epoch": 2.031298904538341,
+      "grad_norm": 0.9070900678634644,
+      "learning_rate": 7.815388323550638e-06,
+      "loss": 0.0971,
+      "step": 650
+    },
+    {
+      "epoch": 2.0469483568075115,
+      "grad_norm": 1.0321788787841797,
+      "learning_rate": 7.589673962812442e-06,
+      "loss": 0.0944,
+      "step": 655
+    },
+    {
+      "epoch": 2.0625978090766823,
+      "grad_norm": 0.9838140606880188,
+      "learning_rate": 7.36615784992626e-06,
+      "loss": 0.093,
+      "step": 660
+    },
+    {
+      "epoch": 2.0782472613458527,
+      "grad_norm": 1.0341883897781372,
+      "learning_rate": 7.144906290181832e-06,
+      "loss": 0.0927,
+      "step": 665
+    },
+    {
+      "epoch": 2.0938967136150235,
+      "grad_norm": 1.152293086051941,
+      "learning_rate": 6.925984917096985e-06,
+      "loss": 0.0994,
+      "step": 670
+    },
+    {
+      "epoch": 2.109546165884194,
+      "grad_norm": 0.9845132231712341,
+      "learning_rate": 6.709458672947631e-06,
+      "loss": 0.0937,
+      "step": 675
+    },
+    {
+      "epoch": 2.1251956181533647,
+      "grad_norm": 1.330306887626648,
+      "learning_rate": 6.4953917895028795e-06,
+      "loss": 0.0915,
+      "step": 680
+    },
+    {
+      "epoch": 2.140845070422535,
+      "grad_norm": 0.9394160509109497,
+      "learning_rate": 6.283847768970927e-06,
+      "loss": 0.1023,
+      "step": 685
+    },
+    {
+      "epoch": 2.156494522691706,
+      "grad_norm": 0.9315147399902344,
+      "learning_rate": 6.074889365161303e-06,
+      "loss": 0.0863,
+      "step": 690
+    },
+    {
+      "epoch": 2.172143974960876,
+      "grad_norm": 0.9891120791435242,
+      "learning_rate": 5.86857856486919e-06,
+      "loss": 0.0859,
+      "step": 695
+    },
+    {
+      "epoch": 2.187793427230047,
+      "grad_norm": 1.1360231637954712,
+      "learning_rate": 5.664976569487264e-06,
+      "loss": 0.0935,
+      "step": 700
+    },
+    {
+      "epoch": 2.2034428794992174,
+      "grad_norm": 0.9550563097000122,
+      "learning_rate": 5.464143776850483e-06,
+      "loss": 0.0819,
+      "step": 705
+    },
+    {
+      "epoch": 2.219092331768388,
+      "grad_norm": 0.8992927074432373,
+      "learning_rate": 5.266139763319294e-06,
+      "loss": 0.0929,
+      "step": 710
+    },
+    {
+      "epoch": 2.2347417840375585,
+      "grad_norm": 0.9905206561088562,
+      "learning_rate": 5.071023266106502e-06,
+      "loss": 0.0771,
+      "step": 715
+    },
+    {
+      "epoch": 2.2503912363067293,
+      "grad_norm": 0.8893988728523254,
+      "learning_rate": 4.878852165853097e-06,
+      "loss": 0.0849,
+      "step": 720
+    },
+    {
+      "epoch": 2.2660406885758997,
+      "grad_norm": 1.1274504661560059,
+      "learning_rate": 4.689683469458124e-06,
+      "loss": 0.0768,
+      "step": 725
+    },
+    {
+      "epoch": 2.2816901408450705,
+      "grad_norm": 0.9970494508743286,
+      "learning_rate": 4.503573293167806e-06,
+      "loss": 0.0667,
+      "step": 730
+    },
+    {
+      "epoch": 2.297339593114241,
+      "grad_norm": 0.950003981590271,
+      "learning_rate": 4.320576845928859e-06,
+      "loss": 0.0847,
+      "step": 735
+    },
+    {
+      "epoch": 2.3129890453834117,
+      "grad_norm": 1.0764044523239136,
+      "learning_rate": 4.14074841301092e-06,
+      "loss": 0.0852,
+      "step": 740
+    },
+    {
+      "epoch": 2.328638497652582,
+      "grad_norm": 0.9176291227340698,
+      "learning_rate": 3.964141339903026e-06,
+      "loss": 0.0868,
+      "step": 745
+    },
+    {
+      "epoch": 2.344287949921753,
+      "grad_norm": 0.933357298374176,
+      "learning_rate": 3.790808016488862e-06,
+      "loss": 0.0792,
+      "step": 750
+    },
+    {
+      "epoch": 2.359937402190923,
+      "grad_norm": 0.954981803894043,
+      "learning_rate": 3.620799861505462e-06,
+      "loss": 0.0814,
+      "step": 755
+    },
+    {
+      "epoch": 2.375586854460094,
+      "grad_norm": 1.0217819213867188,
+      "learning_rate": 3.4541673072900364e-06,
+      "loss": 0.0914,
+      "step": 760
+    },
+    {
+      "epoch": 2.3912363067292643,
+      "grad_norm": 0.814157247543335,
+      "learning_rate": 3.2909597848193703e-06,
+      "loss": 0.0756,
+      "step": 765
+    },
+    {
+      "epoch": 2.406885758998435,
+      "grad_norm": 1.1114461421966553,
+      "learning_rate": 3.1312257090463214e-06,
+      "loss": 0.0691,
+      "step": 770
+    },
+    {
+      "epoch": 2.4225352112676055,
+      "grad_norm": 0.7841159701347351,
+      "learning_rate": 2.975012464537676e-06,
+      "loss": 0.0645,
+      "step": 775
+    },
+    {
+      "epoch": 2.4381846635367763,
+      "grad_norm": 0.8511492013931274,
+      "learning_rate": 2.8223663914176546e-06,
+      "loss": 0.0741,
+      "step": 780
+    },
+    {
+      "epoch": 2.4538341158059467,
+      "grad_norm": 0.8271607160568237,
+      "learning_rate": 2.673332771621324e-06,
+      "loss": 0.0729,
+      "step": 785
+    },
+    {
+      "epoch": 2.4694835680751175,
+      "grad_norm": 0.769767701625824,
+      "learning_rate": 2.52795581546182e-06,
+      "loss": 0.0778,
+      "step": 790
+    },
+    {
+      "epoch": 2.485133020344288,
+      "grad_norm": 0.7408167123794556,
+      "learning_rate": 2.3862786485155436e-06,
+      "loss": 0.076,
+      "step": 795
+    },
+    {
+      "epoch": 2.5007824726134587,
+      "grad_norm": 0.7294985055923462,
+      "learning_rate": 2.2483432988290766e-06,
+      "loss": 0.0832,
+      "step": 800
+    },
+    {
+      "epoch": 2.516431924882629,
+      "grad_norm": 0.7242084741592407,
+      "learning_rate": 2.1141906844517207e-06,
+      "loss": 0.0638,
+      "step": 805
+    },
+    {
+      "epoch": 2.5320813771518,
+      "grad_norm": 0.7555140852928162,
+      "learning_rate": 1.9838606012973152e-06,
+      "loss": 0.0776,
+      "step": 810
+    },
+    {
+      "epoch": 2.54773082942097,
+      "grad_norm": 0.9583760499954224,
+      "learning_rate": 1.857391711338869e-06,
+      "loss": 0.0648,
+      "step": 815
+    },
+    {
+      "epoch": 2.563380281690141,
+      "grad_norm": 0.7715667486190796,
+      "learning_rate": 1.734821531139667e-06,
+      "loss": 0.0722,
+      "step": 820
+    },
+    {
+      "epoch": 2.5790297339593113,
+      "grad_norm": 0.8475284576416016,
+      "learning_rate": 1.6161864207240895e-06,
+      "loss": 0.0693,
+      "step": 825
+    },
+    {
+      "epoch": 2.594679186228482,
+      "grad_norm": 0.7296764850616455,
+      "learning_rate": 1.5015215727915278e-06,
+      "loss": 0.0792,
+      "step": 830
+    },
+    {
+      "epoch": 2.6103286384976525,
+      "grad_norm": 0.660068154335022,
+      "learning_rate": 1.390861002276602e-06,
+      "loss": 0.0623,
+      "step": 835
+    },
+    {
+      "epoch": 2.6259780907668233,
+      "grad_norm": 0.6773574948310852,
+      "learning_rate": 1.2842375362587412e-06,
+      "loss": 0.0715,
+      "step": 840
+    },
+    {
+      "epoch": 2.6416275430359937,
+      "grad_norm": 1.030470609664917,
+      "learning_rate": 1.1816828042241528e-06,
+      "loss": 0.0684,
+      "step": 845
+    },
+    {
+      "epoch": 2.6572769953051645,
+      "grad_norm": 0.8967737555503845,
+      "learning_rate": 1.0832272286830287e-06,
+      "loss": 0.0749,
+      "step": 850
+    },
+    {
+      "epoch": 2.672926447574335,
+      "grad_norm": 0.8080452680587769,
+      "learning_rate": 9.889000161448175e-07,
+      "loss": 0.0724,
+      "step": 855
+    },
+    {
+      "epoch": 2.6885758998435056,
+      "grad_norm": 0.8337356448173523,
+      "learning_rate": 8.987291484542259e-07,
+      "loss": 0.0715,
+      "step": 860
+    },
+    {
+      "epoch": 2.704225352112676,
+      "grad_norm": 0.7395936846733093,
+      "learning_rate": 8.127413744904805e-07,
+      "loss": 0.0736,
+      "step": 865
+    },
+    {
+      "epoch": 2.7198748043818464,
+      "grad_norm": 0.9157183170318604,
+      "learning_rate": 7.30962202232372e-07,
+      "loss": 0.0737,
+      "step": 870
+    },
+    {
+      "epoch": 2.735524256651017,
+      "grad_norm": 0.8858577609062195,
+      "learning_rate": 6.534158911914101e-07,
+      "loss": 0.0711,
+      "step": 875
+    },
+    {
+      "epoch": 2.751173708920188,
+      "grad_norm": 0.836662769317627,
+      "learning_rate": 5.801254452153082e-07,
+      "loss": 0.0775,
+      "step": 880
+    },
+    {
+      "epoch": 2.7668231611893583,
+      "grad_norm": 0.6758183836936951,
+      "learning_rate": 5.11112605663977e-07,
+      "loss": 0.065,
+      "step": 885
+    },
+    {
+      "epoch": 2.7824726134585287,
+      "grad_norm": 0.8250333070755005,
+      "learning_rate": 4.4639784496002533e-07,
+      "loss": 0.0757,
+      "step": 890
+    },
+    {
+      "epoch": 2.7981220657276995,
+      "grad_norm": 0.6689587235450745,
+      "learning_rate": 3.860003605156881e-07,
+      "loss": 0.0602,
+      "step": 895
+    },
+    {
+      "epoch": 2.8137715179968703,
+      "grad_norm": 0.7403008937835693,
+      "learning_rate": 3.299380690379716e-07,
+      "loss": 0.06,
+      "step": 900
+    },
+    {
+      "epoch": 2.8294209702660407,
+      "grad_norm": 0.747850775718689,
+      "learning_rate": 2.7822760121373193e-07,
+      "loss": 0.0844,
+      "step": 905
+    },
+    {
+      "epoch": 2.845070422535211,
+      "grad_norm": 0.6903601884841919,
+      "learning_rate": 2.3088429677623423e-07,
+      "loss": 0.073,
+      "step": 910
+    },
+    {
+      "epoch": 2.860719874804382,
+      "grad_norm": 0.7659348845481873,
+      "learning_rate": 1.8792219995467653e-07,
+      "loss": 0.0732,
+      "step": 915
+    },
+    {
+      "epoch": 2.8763693270735526,
+      "grad_norm": 0.776434063911438,
+      "learning_rate": 1.4935405530801193e-07,
+      "loss": 0.0842,
+      "step": 920
+    },
+    {
+      "epoch": 2.892018779342723,
+      "grad_norm": 0.7690606117248535,
+      "learning_rate": 1.1519130394432476e-07,
+      "loss": 0.0719,
+      "step": 925
+    },
+    {
+      "epoch": 2.9076682316118934,
+      "grad_norm": 0.7178393006324768,
+      "learning_rate": 8.544408012686022e-08,
+      "loss": 0.0677,
+      "step": 930
+    },
+    {
+      "epoch": 2.923317683881064,
+      "grad_norm": 0.6252371072769165,
+      "learning_rate": 6.012120826773604e-08,
+      "loss": 0.0606,
+      "step": 935
+    },
+    {
+      "epoch": 2.938967136150235,
+      "grad_norm": 0.8659175038337708,
+      "learning_rate": 3.9230200310202966e-08,
+      "loss": 0.0638,
+      "step": 940
+    },
+    {
+      "epoch": 2.9546165884194053,
+      "grad_norm": 0.7876169681549072,
+      "learning_rate": 2.2777253500257388e-08,
+      "loss": 0.0689,
+      "step": 945
+    },
+    {
+      "epoch": 2.9702660406885757,
+      "grad_norm": 0.6571369767189026,
+      "learning_rate": 1.076724854824851e-08,
+      "loss": 0.0646,
+      "step": 950
+    },
+    {
+      "epoch": 2.9859154929577465,
+      "grad_norm": 0.7026804089546204,
+      "learning_rate": 3.20374818103486e-09,
+      "loss": 0.069,
+      "step": 955
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.8314445614814758,
+      "learning_rate": 8.899608510803958e-11,
+      "loss": 0.0635,
+      "step": 960
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 960,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 2000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.1734029151512822e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

137_128_e3_3e-5/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7b5eca953172854069b24eccf54f0ceaf07bd04ac672df9c8a4cf47bb44c1a45
+size 8209

137_128_e3_3e-5/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)