Upload folder using huggingface_hub

README.md

@@ -0,0 +1,9 @@
+---
+library_name: peft
+---
+## Training procedure
+
+### Framework versions
+
+
+- PEFT 0.4.0

adapter_config.json

@@ -0,0 +1,23 @@
+{
+  "auto_mapping": null,
+  "base_model_name_or_path": "mistralai/Mistral-7B-v0.1",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "lora_alpha": 16,
+  "lora_dropout": 0.05,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "v_proj",
+    "o_proj",
+    "k_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}

adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5ea5492a2629a41a3834cf2de2d413d3c30977c44b370a4bdc94bfa086eb6f04
+size 13665592

all_results.json

@@ -0,0 +1,17 @@
+{
+    "epoch": 2.0,
+    "eval_accuracy": 0.4599820966423157,
+    "eval_loss": 1.1161718368530273,
+    "eval_runtime": 166.5715,
+    "eval_samples": 25899,
+    "eval_samples_per_second": 160.844,
+    "eval_steps_per_second": 2.515,
+    "max_memory_allocated (GB)": 91.87,
+    "memory_allocated (GB)": 24.39,
+    "perplexity": 3.0531438702149534,
+    "total_memory_available (GB)": 94.62,
+    "train_loss": 1.0469276263736165,
+    "train_runtime": 6085.8982,
+    "train_samples_per_second": 162.542,
+    "train_steps_per_second": 0.317
+}

checkpoint-1000/README.md

@@ -0,0 +1,9 @@
+---
+library_name: peft
+---
+## Training procedure
+
+### Framework versions
+
+
+- PEFT 0.4.0

checkpoint-1000/adapter_config.json

@@ -0,0 +1,23 @@
+{
+  "auto_mapping": null,
+  "base_model_name_or_path": "mistralai/Mistral-7B-v0.1",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "lora_alpha": 16,
+  "lora_dropout": 0.05,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "v_proj",
+    "o_proj",
+    "k_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}

checkpoint-1000/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:81c901e3a02306c80837c992a65b028aa5720bd76546736f6b293eb7dfc7140c
+size 13665592

checkpoint-1000/gaudi_config.json

@@ -0,0 +1,10 @@
+{
+  "autocast_bf16_ops": null,
+  "autocast_fp32_ops": null,
+  "optimum_version": "1.20.0",
+  "transformers_version": "4.38.2",
+  "use_dynamic_shapes": false,
+  "use_fused_adam": true,
+  "use_fused_clip_norm": true,
+  "use_torch_autocast": false
+}

checkpoint-1000/global_step1000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fbcc1217cb0120aeb98ec1de36b0d9911840455125d2ac021d411340695b0e41
+size 10229904

checkpoint-1000/global_step1000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b97ccb06cab86f7787aa54118479ed5365ea36124fa165035de40978981d7d90
+size 10229904

checkpoint-1000/global_step1000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0e88a8703d82aca1a49db083c3389258661948ca907ccc2e9cb111b1551ca4b4
+size 10229904

checkpoint-1000/global_step1000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e427e683a4b2c0eaf3f92eeb4d17130f2db3cf91beb49cdec09db31cd03bc5d0
+size 10229904

checkpoint-1000/global_step1000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:447bcfa260f2ca4e01eab0355a25bc057c887425ee46599c03014702139424e9
+size 10229904

checkpoint-1000/global_step1000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2f82c584aafdcfd6f83b4a382b39b2a21a1c6d328e15b598c757f0bbd57d7f40
+size 10229904

checkpoint-1000/global_step1000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ed340b57ccacdb58d19cb98548194affef267ef083941da271ddb8bda6982c71
+size 10229904

checkpoint-1000/global_step1000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4e1108862ec93ac55212f8e0a1b8f3e51f7907e476768c008d36049e3882d3c6
+size 10229904

checkpoint-1000/global_step1000/mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f7eaa5aa673d8036fda2b25512f384ce627430c73f15341bdbc6c9ea4bbcccca
+size 13740018

checkpoint-1000/latest

@@ -0,0 +1 @@
+global_step1000

checkpoint-1000/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:da3853db45e30a134bde6283db65f4d2e12b4f06b534a780433dd598db8504cb
+size 18032

checkpoint-1000/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d0664a5f5d0f31c8412dfa2b9c211689c20cc952d4ce4660ee1761e2ec10776d
+size 18032

checkpoint-1000/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8418b1622ae3a1586b0d267b3c8c1d9d5d403a56cc73434016d717b501b2c0fd
+size 18032

checkpoint-1000/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:51d300bdb34d5407f13acbfb278589f1a2907891fba4e020420c6053746cf8e2
+size 18032

checkpoint-1000/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9d597fad31002fa89ec1fd8aefe8f76311a32275c74a84439354f630b9b001a4
+size 18032

checkpoint-1000/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1fbf5d648175322bc9d62f67af45fa2e710bdc93b49adb768d8df629d2b0347f
+size 18032

checkpoint-1000/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fcfd1907de0ce290b15c0e3cda02ca34b46efde1a255be0c9b8e383cbf93e363
+size 18032

checkpoint-1000/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6a4d45767ae6a4db5ed86b012add567287d0976c731678dabe7dedeb253ea63b
+size 18032

checkpoint-1000/special_tokens_map.json

@@ -0,0 +1,12 @@
+{
+  "bos_token": "<s>",
+  "eos_token": "</s>",
+  "pad_token": "</s>",
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

checkpoint-1000/tokenizer.model

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

checkpoint-1000/tokenizer_config.json

@@ -0,0 +1,42 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [],
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "</s>",
+  "legacy": true,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "</s>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}

checkpoint-1000/trainer_state.json

@@ -0,0 +1,1021 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0404473923787227,
+  "eval_steps": 500,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8564500212669373,
+      "learning_rate": 5.670773083167062e-05,
+      "loss": 1.7378,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 10,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9088606238365173,
+      "learning_rate": 7.377845879804262e-05,
+      "loss": 1.5183,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 20,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.396087646484375,
+      "learning_rate": 8.376419451838216e-05,
+      "loss": 1.3578,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 30,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.2860943078994751,
+      "learning_rate": 9.084918676441463e-05,
+      "loss": 1.2832,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 40,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.2048167884349823,
+      "learning_rate": 9.634473369696918e-05,
+      "loss": 1.2717,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 50,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.2355910986661911,
+      "learning_rate": 9.994635193133047e-05,
+      "loss": 1.2359,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 60,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.25649788975715637,
+      "learning_rate": 9.94098712446352e-05,
+      "loss": 1.2202,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 70,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3248240649700165,
+      "learning_rate": 9.887339055793991e-05,
+      "loss": 1.1928,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 80,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.34330031275749207,
+      "learning_rate": 9.833690987124465e-05,
+      "loss": 1.1839,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 90,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.481341689825058,
+      "learning_rate": 9.780042918454936e-05,
+      "loss": 1.1682,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 100,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.33835262060165405,
+      "learning_rate": 9.726394849785409e-05,
+      "loss": 1.1528,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 110,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5484705567359924,
+      "learning_rate": 9.67274678111588e-05,
+      "loss": 1.1511,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 120,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.3785193860530853,
+      "learning_rate": 9.619098712446352e-05,
+      "loss": 1.152,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 130,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.36997660994529724,
+      "learning_rate": 9.565450643776824e-05,
+      "loss": 1.1489,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 140,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.5300387144088745,
+      "learning_rate": 9.511802575107297e-05,
+      "loss": 1.1326,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 150,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5168531537055969,
+      "learning_rate": 9.458154506437769e-05,
+      "loss": 1.1253,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 160,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.48498621582984924,
+      "learning_rate": 9.404506437768241e-05,
+      "loss": 1.1079,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 170,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4298243522644043,
+      "learning_rate": 9.350858369098713e-05,
+      "loss": 1.0987,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 180,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.4342297315597534,
+      "learning_rate": 9.297210300429185e-05,
+      "loss": 1.0975,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 190,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5405712127685547,
+      "learning_rate": 9.243562231759658e-05,
+      "loss": 1.1084,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 200,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4317038655281067,
+      "learning_rate": 9.189914163090128e-05,
+      "loss": 1.0969,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 210,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.41190406680107117,
+      "learning_rate": 9.136266094420602e-05,
+      "loss": 1.0992,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 220,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.5093066692352295,
+      "learning_rate": 9.082618025751073e-05,
+      "loss": 1.0875,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 230,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.40610459446907043,
+      "learning_rate": 9.028969957081545e-05,
+      "loss": 1.0886,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 240,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.38791623711586,
+      "learning_rate": 8.975321888412017e-05,
+      "loss": 1.0783,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 250,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3828742206096649,
+      "learning_rate": 8.92167381974249e-05,
+      "loss": 1.0816,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 260,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5216621160507202,
+      "learning_rate": 8.868025751072962e-05,
+      "loss": 1.0731,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 270,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.43531420826911926,
+      "learning_rate": 8.814377682403434e-05,
+      "loss": 1.0804,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 280,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.41790470480918884,
+      "learning_rate": 8.760729613733906e-05,
+      "loss": 1.0785,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 290,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.456264466047287,
+      "learning_rate": 8.707081545064378e-05,
+      "loss": 1.0708,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 300,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3793538212776184,
+      "learning_rate": 8.65343347639485e-05,
+      "loss": 1.0818,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 310,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.37741824984550476,
+      "learning_rate": 8.599785407725323e-05,
+      "loss": 1.0669,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 320,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3576098084449768,
+      "learning_rate": 8.546137339055795e-05,
+      "loss": 1.0678,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 330,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4295920133590698,
+      "learning_rate": 8.492489270386267e-05,
+      "loss": 1.0571,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 340,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.38484010100364685,
+      "learning_rate": 8.438841201716738e-05,
+      "loss": 1.0697,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 350,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3806072175502777,
+      "learning_rate": 8.385193133047211e-05,
+      "loss": 1.0652,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 360,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3507857918739319,
+      "learning_rate": 8.331545064377682e-05,
+      "loss": 1.061,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 370,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3869009017944336,
+      "learning_rate": 8.277896995708156e-05,
+      "loss": 1.0622,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 380,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3941822648048401,
+      "learning_rate": 8.224248927038627e-05,
+      "loss": 1.0629,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 390,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3660201132297516,
+      "learning_rate": 8.1706008583691e-05,
+      "loss": 1.0589,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 400,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3641981780529022,
+      "learning_rate": 8.116952789699571e-05,
+      "loss": 1.0611,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 410,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.38291656970977783,
+      "learning_rate": 8.063304721030043e-05,
+      "loss": 1.0554,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 420,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3162360191345215,
+      "learning_rate": 8.009656652360515e-05,
+      "loss": 1.0504,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 430,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3741254508495331,
+      "learning_rate": 7.956008583690988e-05,
+      "loss": 1.0525,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 440,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3503776788711548,
+      "learning_rate": 7.90236051502146e-05,
+      "loss": 1.0554,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 450,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3659127652645111,
+      "learning_rate": 7.848712446351931e-05,
+      "loss": 1.0464,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 460,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3401431143283844,
+      "learning_rate": 7.795064377682404e-05,
+      "loss": 1.0596,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 470,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.33970314264297485,
+      "learning_rate": 7.741416309012875e-05,
+      "loss": 1.0659,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 480,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.40222179889678955,
+      "learning_rate": 7.687768240343349e-05,
+      "loss": 1.0485,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 490,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3137299716472626,
+      "learning_rate": 7.63412017167382e-05,
+      "loss": 1.0506,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 500,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3413981795310974,
+      "learning_rate": 7.580472103004293e-05,
+      "loss": 1.0466,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 510,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3440905511379242,
+      "learning_rate": 7.526824034334764e-05,
+      "loss": 1.0503,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 520,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.367201030254364,
+      "learning_rate": 7.473175965665236e-05,
+      "loss": 1.0449,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 530,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.37777239084243774,
+      "learning_rate": 7.419527896995708e-05,
+      "loss": 1.0514,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 540,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.355307936668396,
+      "learning_rate": 7.36587982832618e-05,
+      "loss": 1.0458,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 550,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.43950700759887695,
+      "learning_rate": 7.312231759656653e-05,
+      "loss": 1.0426,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 560,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.31208208203315735,
+      "learning_rate": 7.258583690987125e-05,
+      "loss": 1.0432,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 570,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3358898460865021,
+      "learning_rate": 7.204935622317597e-05,
+      "loss": 1.0563,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 580,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3219963610172272,
+      "learning_rate": 7.15128755364807e-05,
+      "loss": 1.0402,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 590,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3409494161605835,
+      "learning_rate": 7.097639484978542e-05,
+      "loss": 1.0452,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 600,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3917517364025116,
+      "learning_rate": 7.043991416309014e-05,
+      "loss": 1.0358,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 610,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.36202436685562134,
+      "learning_rate": 6.990343347639486e-05,
+      "loss": 1.0421,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 620,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.34764185547828674,
+      "learning_rate": 6.936695278969958e-05,
+      "loss": 1.0414,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 630,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3368031084537506,
+      "learning_rate": 6.883047210300429e-05,
+      "loss": 1.0328,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 640,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3216454088687897,
+      "learning_rate": 6.829399141630901e-05,
+      "loss": 1.0368,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 650,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.36135369539260864,
+      "learning_rate": 6.775751072961373e-05,
+      "loss": 1.0375,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 660,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3772068917751312,
+      "learning_rate": 6.722103004291846e-05,
+      "loss": 1.0384,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 670,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.352990984916687,
+      "learning_rate": 6.668454935622318e-05,
+      "loss": 1.0347,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 680,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.38070163130760193,
+      "learning_rate": 6.61480686695279e-05,
+      "loss": 1.0414,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 690,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3262191414833069,
+      "learning_rate": 6.561158798283262e-05,
+      "loss": 1.0427,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 700,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.37472084164619446,
+      "learning_rate": 6.507510729613734e-05,
+      "loss": 1.035,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 710,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3830932676792145,
+      "learning_rate": 6.453862660944207e-05,
+      "loss": 1.0329,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 720,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.33351582288742065,
+      "learning_rate": 6.400214592274679e-05,
+      "loss": 1.0338,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 730,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.30762043595314026,
+      "learning_rate": 6.346566523605151e-05,
+      "loss": 1.0452,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 740,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3884871006011963,
+      "learning_rate": 6.292918454935622e-05,
+      "loss": 1.0387,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 750,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.34226515889167786,
+      "learning_rate": 6.239270386266095e-05,
+      "loss": 1.0282,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 760,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3282904624938965,
+      "learning_rate": 6.185622317596566e-05,
+      "loss": 1.0266,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 770,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.34865814447402954,
+      "learning_rate": 6.13197424892704e-05,
+      "loss": 1.0369,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 780,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.34097954630851746,
+      "learning_rate": 6.0783261802575106e-05,
+      "loss": 1.0241,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 790,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.342785120010376,
+      "learning_rate": 6.0246781115879835e-05,
+      "loss": 1.0289,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 800,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.33090144395828247,
+      "learning_rate": 5.971030042918455e-05,
+      "loss": 1.0335,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 810,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.38638409972190857,
+      "learning_rate": 5.917381974248928e-05,
+      "loss": 1.0272,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 820,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.34252798557281494,
+      "learning_rate": 5.8637339055793994e-05,
+      "loss": 1.0284,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 830,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3451687693595886,
+      "learning_rate": 5.810085836909872e-05,
+      "loss": 1.0212,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 840,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.384164422750473,
+      "learning_rate": 5.756437768240344e-05,
+      "loss": 1.0347,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 850,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3310836851596832,
+      "learning_rate": 5.7027896995708154e-05,
+      "loss": 1.0153,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 860,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3287970721721649,
+      "learning_rate": 5.6491416309012876e-05,
+      "loss": 1.0386,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 870,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3661152720451355,
+      "learning_rate": 5.59549356223176e-05,
+      "loss": 1.0195,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 880,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.39101266860961914,
+      "learning_rate": 5.541845493562232e-05,
+      "loss": 1.0285,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 890,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.38129398226737976,
+      "learning_rate": 5.4881974248927035e-05,
+      "loss": 1.0375,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 900,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.32832667231559753,
+      "learning_rate": 5.4345493562231764e-05,
+      "loss": 1.0229,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 910,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.35413920879364014,
+      "learning_rate": 5.380901287553648e-05,
+      "loss": 1.024,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 920,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3270743787288666,
+      "learning_rate": 5.327253218884121e-05,
+      "loss": 1.0319,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 930,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.33611926436424255,
+      "learning_rate": 5.273605150214592e-05,
+      "loss": 1.0198,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 940,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.34108686447143555,
+      "learning_rate": 5.219957081545065e-05,
+      "loss": 1.0294,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 950,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3304935395717621,
+      "learning_rate": 5.166309012875537e-05,
+      "loss": 1.0264,
+      "max_memory_allocated (GB)": 91.86,
+      "memory_allocated (GB)": 24.39,
+      "step": 960,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.3912031054496765,
+      "learning_rate": 5.112660944206009e-05,
+      "loss": 1.0192,
+      "max_memory_allocated (GB)": 91.87,
+      "memory_allocated (GB)": 24.39,
+      "step": 970,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.33931615948677063,
+      "learning_rate": 5.0590128755364804e-05,
+      "loss": 1.0242,
+      "max_memory_allocated (GB)": 91.87,
+      "memory_allocated (GB)": 24.39,
+      "step": 980,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.3690173923969269,
+      "learning_rate": 5.005364806866953e-05,
+      "loss": 1.0202,
+      "max_memory_allocated (GB)": 91.87,
+      "memory_allocated (GB)": 24.39,
+      "step": 990,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.35189080238342285,
+      "learning_rate": 4.951716738197425e-05,
+      "loss": 1.0206,
+      "max_memory_allocated (GB)": 91.87,
+      "memory_allocated (GB)": 24.39,
+      "step": 1000,
+      "total_memory_available (GB)": 94.62
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 1922,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 1000,
+  "total_flos": 1.119482194910249e+19,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-1000/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:93e2964507a7ee2a283815dc99898c3c891ee7684ce7926ce108452bc498151d
+size 5944

checkpoint-1000/zero_to_fp32.py

@@ -0,0 +1,592 @@
+#!/usr/bin/env python
+
+# Copyright (c) Microsoft Corporation.
+# SPDX-License-Identifier: Apache-2.0
+
+# DeepSpeed Team
+
+# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
+# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
+# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
+# application.
+#
+# example: python zero_to_fp32.py . pytorch_model.bin
+
+import argparse
+import torch
+import glob
+import math
+import os
+import re
+from collections import OrderedDict
+from dataclasses import dataclass
+
+# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
+# DeepSpeed data structures it has to be available in the current python environment.
+from deepspeed.utils import logger
+from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
+                                            FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
+                                            FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
+
+
+@dataclass
+class zero_model_state:
+    buffers: dict()
+    param_shapes: dict()
+    shared_params: list
+    ds_version: int
+    frozen_param_shapes: dict()
+    frozen_param_fragments: dict()
+
+
+debug = 0
+
+# load to cpu
+device = torch.device('cpu')
+
+
+def atoi(text):
+    return int(text) if text.isdigit() else text
+
+
+def natural_keys(text):
+    '''
+    alist.sort(key=natural_keys) sorts in human order
+    http://nedbatchelder.com/blog/200712/human_sorting.html
+    (See Toothy's implementation in the comments)
+    '''
+    return [atoi(c) for c in re.split(r'(\d+)', text)]
+
+
+def get_model_state_file(checkpoint_dir, zero_stage):
+    if not os.path.isdir(checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
+
+    # there should be only one file
+    if zero_stage <= 2:
+        file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
+    elif zero_stage == 3:
+        file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
+
+    if not os.path.exists(file):
+        raise FileNotFoundError(f"can't find model states file at '{file}'")
+
+    return file
+
+
+def get_checkpoint_files(checkpoint_dir, glob_pattern):
+    # XXX: need to test that this simple glob rule works for multi-node setup too
+    ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
+
+    if len(ckpt_files) == 0:
+        raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
+
+    return ckpt_files
+
+
+def get_optim_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
+
+
+def get_model_state_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
+
+
+def parse_model_states(files):
+    zero_model_states = []
+    for file in files:
+        state_dict = torch.load(file, map_location=device)
+
+        if BUFFER_NAMES not in state_dict:
+            raise ValueError(f"{file} is not a model state checkpoint")
+        buffer_names = state_dict[BUFFER_NAMES]
+        if debug:
+            print("Found buffers:", buffer_names)
+
+        # recover just the buffers while restoring them to fp32 if they were saved in fp16
+        buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
+        param_shapes = state_dict[PARAM_SHAPES]
+
+        # collect parameters that are included in param_shapes
+        param_names = []
+        for s in param_shapes:
+            for name in s.keys():
+                param_names.append(name)
+
+        # update with frozen parameters
+        frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
+        if frozen_param_shapes is not None:
+            if debug:
+                print(f"Found frozen_param_shapes: {frozen_param_shapes}")
+            param_names += list(frozen_param_shapes.keys())
+
+        # handle shared params
+        shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
+
+        ds_version = state_dict.get(DS_VERSION, None)
+
+        frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
+
+        z_model_state = zero_model_state(buffers=buffers,
+                                         param_shapes=param_shapes,
+                                         shared_params=shared_params,
+                                         ds_version=ds_version,
+                                         frozen_param_shapes=frozen_param_shapes,
+                                         frozen_param_fragments=frozen_param_fragments)
+        zero_model_states.append(z_model_state)
+
+    return zero_model_states
+
+
+def parse_optim_states(files, ds_checkpoint_dir):
+
+    total_files = len(files)
+    state_dicts = []
+    for f in files:
+        state_dict = torch.load(f, map_location=device)
+        # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
+        # and also handle the case where it was already removed by another helper script
+        state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
+        state_dicts.append(state_dict)
+
+    if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
+        raise ValueError(f"{files[0]} is not a zero checkpoint")
+    zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
+    world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
+
+    # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
+    # parameters can be different from data parallelism for non-expert parameters. So we can just
+    # use the max of the partition_count to get the dp world_size.
+
+    if type(world_size) is list:
+        world_size = max(world_size)
+
+    if world_size != total_files:
+        raise ValueError(
+            f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
+            "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
+        )
+
+    # the groups are named differently in each stage
+    if zero_stage <= 2:
+        fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
+    elif zero_stage == 3:
+        fp32_groups_key = FP32_FLAT_GROUPS
+    else:
+        raise ValueError(f"unknown zero stage {zero_stage}")
+
+    if zero_stage <= 2:
+        fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
+    elif zero_stage == 3:
+        # if there is more than one param group, there will be multiple flattened tensors - one
+        # flattened tensor per group - for simplicity merge them into a single tensor
+        #
+        # XXX: could make the script more memory efficient for when there are multiple groups - it
+        # will require matching the sub-lists of param_shapes for each param group flattened tensor
+
+        fp32_flat_groups = [
+            torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
+        ]
+
+    return zero_stage, world_size, fp32_flat_groups
+
+
+def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir):
+    """
+    Returns fp32 state_dict reconstructed from ds checkpoint
+
+    Args:
+        - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
+
+    """
+    print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
+
+    optim_files = get_optim_files(ds_checkpoint_dir)
+    zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
+    print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
+
+    model_files = get_model_state_files(ds_checkpoint_dir)
+
+    zero_model_states = parse_model_states(model_files)
+    print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
+
+    if zero_stage <= 2:
+        return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states)
+    elif zero_stage == 3:
+        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states)
+
+
+def _zero2_merge_frozen_params(state_dict, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+    frozen_param_fragments = zero_model_states[0].frozen_param_fragments
+
+    if debug:
+        num_elem = sum(s.numel() for s in frozen_param_shapes.values())
+        print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        state_dict[name] = frozen_param_fragments[name]
+
+        if debug:
+            print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _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):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    _zero2_merge_frozen_params(state_dict, zero_model_states)
+
+    _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def zero3_partitioned_param_info(unpartitioned_numel, world_size):
+    remainder = unpartitioned_numel % world_size
+    padding_numel = (world_size - remainder) if remainder else 0
+    partitioned_numel = math.ceil(unpartitioned_numel / world_size)
+    return partitioned_numel, padding_numel
+
+
+def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    if debug:
+        for i in range(world_size):
+            num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
+            print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in zero_model_states[0].frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
+        state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+    avail_numel = fp32_flat_groups[0].numel() * world_size
+    # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
+    # param, re-consolidating each param, while dealing with padding if any
+
+    # merge list of dicts, preserving order
+    param_shapes = {k: v for d in param_shapes for k, v in d.items()}
+
+    if debug:
+        for i in range(world_size):
+            print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
+
+        wanted_params = len(param_shapes)
+        wanted_numel = sum(shape.numel() for shape in param_shapes.values())
+        # not asserting if there is a mismatch due to possible padding
+        avail_numel = fp32_flat_groups[0].numel() * world_size
+        print(f"Trainable params: Have {avail_numel} numels to process.")
+        print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    offset = 0
+    total_numel = 0
+    total_params = 0
+    for name, shape in param_shapes.items():
+
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+        total_params += 1
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+        # XXX: memory usage doubles here
+        state_dict[name] = torch.cat(
+            tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
+            0).narrow(0, 0, unpartitioned_numel).view(shape)
+        offset += partitioned_numel
+
+    offset *= world_size
+
+    # Sanity check
+    if offset != avail_numel:
+        raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
+
+    _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
+    ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
+    via a model hub.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
+
+    Returns:
+        - pytorch ``state_dict``
+
+    Note: this approach may not work if your application doesn't have sufficient free CPU memory and
+    you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
+    the checkpoint.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        # do the training and checkpoint saving
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
+        model = model.cpu() # move to cpu
+        model.load_state_dict(state_dict)
+        # submit to model hub or save the model to share with others
+
+    In this example the ``model`` will no longer be usable in the deepspeed context of the same
+    application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
+
+    """
+    if tag is None:
+        latest_path = os.path.join(checkpoint_dir, 'latest')
+        if os.path.isfile(latest_path):
+            with open(latest_path, 'r') as fd:
+                tag = fd.read().strip()
+        else:
+            raise ValueError(f"Unable to find 'latest' file at {latest_path}")
+
+    ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
+
+    if not os.path.isdir(ds_checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
+
+    return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir)
+
+
+def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
+    loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+    """
+
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+    print(f"Saving fp32 state dict to {output_file}")
+    torch.save(state_dict, output_file)
+
+
+def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
+    """
+    1. Put the provided model to cpu
+    2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
+    3. Load it into the provided model
+
+    Args:
+        - ``model``: the model object to update
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+
+    Returns:
+        - ``model`: modified model
+
+    Make sure you have plenty of CPU memory available before you call this function. If you don't
+    have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
+    conveniently placed for you in the checkpoint folder.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
+        model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
+        # submit to model hub or save the model to share with others
+
+    Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
+    of the same application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    """
+    logger.info(f"Extracting fp32 weights")
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+
+    logger.info(f"Overwriting model with fp32 weights")
+    model = model.cpu()
+    model.load_state_dict(state_dict, strict=False)
+
+    return model
+
+
+if __name__ == "__main__":
+
+    parser = argparse.ArgumentParser()
+    parser.add_argument("checkpoint_dir",
+                        type=str,
+                        help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
+    parser.add_argument(
+        "output_file",
+        type=str,
+        help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
+    parser.add_argument("-t",
+                        "--tag",
+                        type=str,
+                        default=None,
+                        help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
+    parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
+    args = parser.parse_args()
+
+    debug = args.debug
+
+    convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, args.output_file, tag=args.tag)

eval_results.json

@@ -0,0 +1,13 @@
+{
+    "epoch": 2.0,
+    "eval_accuracy": 0.4599820966423157,
+    "eval_loss": 1.1161718368530273,
+    "eval_runtime": 166.5715,
+    "eval_samples": 25899,
+    "eval_samples_per_second": 160.844,
+    "eval_steps_per_second": 2.515,
+    "max_memory_allocated (GB)": 91.87,
+    "memory_allocated (GB)": 24.39,
+    "perplexity": 3.0531438702149534,
+    "total_memory_available (GB)": 94.62
+}

gaudi_config.json

@@ -0,0 +1,10 @@
+{
+  "autocast_bf16_ops": null,
+  "autocast_fp32_ops": null,
+  "optimum_version": "1.20.0",
+  "transformers_version": "4.38.2",
+  "use_dynamic_shapes": false,
+  "use_fused_adam": true,
+  "use_fused_clip_norm": true,
+  "use_torch_autocast": false
+}

runs/Jun05_07-58-02_9d340316bc35/events.out.tfevents.1717575602.9d340316bc35.8309.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6ae36d7bcc6640b59f3d4f68d7c9d3fd9e8aeca0d698e1c1395fcf20116990ba
+size 86156

runs/Jun05_07-58-02_9d340316bc35/events.out.tfevents.1717581859.9d340316bc35.8309.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1a04411d16d99535d15091abd35936d8b2137ef416514bccd267ef8fcba0f6e9
+size 571

special_tokens_map.json

@@ -0,0 +1,12 @@
+{
+  "bos_token": "<s>",
+  "eos_token": "</s>",
+  "pad_token": "</s>",
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.model

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

tokenizer_config.json

@@ -0,0 +1,42 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [],
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "</s>",
+  "legacy": true,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "</s>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}

train_results.json

@@ -0,0 +1,10 @@
+{
+    "epoch": 2.0,
+    "max_memory_allocated (GB)": 91.87,
+    "memory_allocated (GB)": 24.39,
+    "total_memory_available (GB)": 94.62,
+    "train_loss": 1.0469276263736165,
+    "train_runtime": 6085.8982,
+    "train_samples_per_second": 162.542,
+    "train_steps_per_second": 0.317
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:93e2964507a7ee2a283815dc99898c3c891ee7684ce7926ce108452bc498151d
+size 5944