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": 32,
+  "lora_dropout": 0.1,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "k_proj",
+    "v_proj",
+    "o_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}

adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c1677aac55780016aa5899bb8861521dd55e5822e46596889750149cff62ff3d
+size 13665592

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": 32,
+  "lora_dropout": 0.1,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "k_proj",
+    "v_proj",
+    "o_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}

checkpoint-1000/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5772181171b699a38742439a65343fea3c0d07127d705466164039af49cfa56f
+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 sha256:73655cd44c220b606f34c4b8a97c2f2f4f0d2e53e7853b0d51a512d71268baff
+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 sha256:9d2b531c24e8ab4583b7619ad87879b44782f610c515f0dbb1ec57bee8fc3536
+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:f9923bbd32fcee5a6a59ebaee726bbb969ccbbc3a2a0a15946e64b1da79bc797
+size 10229968

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 sha256:55422d1f8bf886ae6916aa3cd66f30fe0135a7c3fd5faa9355e190eb7ef8bbd6
+size 10229968

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:d6e4c1960b417fbc03f9eb51501b34b5d735600e0e88be1a9816a07da07cb6aa
+size 10229968

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 sha256:8c41b5dafa83796ef1ea4787beb395d018c601c2f49ce409a425eb41f97077a6
+size 10229968

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 sha256:60ff711983d46faef174522542ca03ed4c9e62a8cdebb770d47e41ed1c89ad53
+size 10229968

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:db5f2d56d9188a731bbdc37489e2c06f27abd5777aacb3d5336b3c66b56e7562
+size 10229968

checkpoint-1000/global_step1000/mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:78de1143b9c7f4ef36d3b294b5e19aac01912b1c9b3320d5c57c657e8a8d379d
+size 13740082

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:a1a71ef746d5605c01125c52c8992ee3510b53dee333bdcf6038404cc4d18369
+size 18032

checkpoint-1000/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:33c7e0bcd6392f33147cbe763b8ba2d23a5454e89548d59fbc6c27df91554ee5
+size 18032

checkpoint-1000/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f0feed945f36ceac4ad5c1581f6628e186e73f61311020fa7b54b9d862cf724
+size 18032

checkpoint-1000/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:48e8f54d6e56f4ecec8a7f5bc0ac31bbca49207f39eb837035e6556788a2a778
+size 18032

checkpoint-1000/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:46c64c3e1f93ef4f705341ce58f7c54a4cb9c25883788a04f1a7e46450e094f9
+size 18032

checkpoint-1000/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:96b084a3807172e1df4b2e1b655a5fe5318c69f12b4d3a90e206cfb66f5dccca
+size 18032

checkpoint-1000/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b42ad7b6efe463842f1ea4edf25c90bb719e313c9d7e8ae66340ac836cee85e5
+size 18032

checkpoint-1000/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a116663f3867f3cb87c1e28176d22d33fff4eb405bd7bfd6a5bd4cf1755ab28e
+size 18032

checkpoint-1000/special_tokens_map.json

@@ -0,0 +1,24 @@
+{
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "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": 0.9883864591055103,
+  "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": 1.2311009168624878,
+      "learning_rate": 5.6012058970266934e-05,
+      "loss": 1.6789,
+      "max_memory_allocated (GB)": 91.88,
+      "memory_allocated (GB)": 14.99,
+      "step": 10,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4872009754180908,
+      "learning_rate": 7.287336883921704e-05,
+      "loss": 1.3884,
+      "max_memory_allocated (GB)": 91.88,
+      "memory_allocated (GB)": 14.99,
+      "step": 20,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7868104577064514,
+      "learning_rate": 8.273660282559241e-05,
+      "loss": 1.2404,
+      "max_memory_allocated (GB)": 91.88,
+      "memory_allocated (GB)": 14.99,
+      "step": 30,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.35713163018226624,
+      "learning_rate": 8.973467870816715e-05,
+      "loss": 1.2036,
+      "max_memory_allocated (GB)": 91.88,
+      "memory_allocated (GB)": 14.99,
+      "step": 40,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3057297468185425,
+      "learning_rate": 9.516280807158375e-05,
+      "loss": 1.1788,
+      "max_memory_allocated (GB)": 91.88,
+      "memory_allocated (GB)": 14.99,
+      "step": 50,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3608625531196594,
+      "learning_rate": 9.959791269454252e-05,
+      "loss": 1.1525,
+      "max_memory_allocated (GB)": 91.88,
+      "memory_allocated (GB)": 14.99,
+      "step": 60,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3684042990207672,
+      "learning_rate": 9.959204487506375e-05,
+      "loss": 1.1261,
+      "max_memory_allocated (GB)": 91.88,
+      "memory_allocated (GB)": 14.99,
+      "step": 70,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4175470471382141,
+      "learning_rate": 9.908210096889343e-05,
+      "loss": 1.1214,
+      "max_memory_allocated (GB)": 91.88,
+      "memory_allocated (GB)": 14.99,
+      "step": 80,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4869045317173004,
+      "learning_rate": 9.85721570627231e-05,
+      "loss": 1.1043,
+      "max_memory_allocated (GB)": 91.88,
+      "memory_allocated (GB)": 14.99,
+      "step": 90,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6399329900741577,
+      "learning_rate": 9.806221315655279e-05,
+      "loss": 1.1059,
+      "max_memory_allocated (GB)": 91.88,
+      "memory_allocated (GB)": 14.99,
+      "step": 100,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5639649033546448,
+      "learning_rate": 9.755226925038246e-05,
+      "loss": 1.0711,
+      "max_memory_allocated (GB)": 91.91,
+      "memory_allocated (GB)": 14.99,
+      "step": 110,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.44946518540382385,
+      "learning_rate": 9.704232534421214e-05,
+      "loss": 1.0644,
+      "max_memory_allocated (GB)": 91.91,
+      "memory_allocated (GB)": 14.99,
+      "step": 120,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5573060512542725,
+      "learning_rate": 9.653238143804181e-05,
+      "loss": 1.0634,
+      "max_memory_allocated (GB)": 91.92,
+      "memory_allocated (GB)": 14.99,
+      "step": 130,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6105266213417053,
+      "learning_rate": 9.60224375318715e-05,
+      "loss": 1.0597,
+      "max_memory_allocated (GB)": 91.92,
+      "memory_allocated (GB)": 14.99,
+      "step": 140,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6286391615867615,
+      "learning_rate": 9.551249362570118e-05,
+      "loss": 1.0528,
+      "max_memory_allocated (GB)": 91.92,
+      "memory_allocated (GB)": 14.99,
+      "step": 150,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8733624815940857,
+      "learning_rate": 9.500254971953085e-05,
+      "loss": 1.0524,
+      "max_memory_allocated (GB)": 91.93,
+      "memory_allocated (GB)": 14.99,
+      "step": 160,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6268635392189026,
+      "learning_rate": 9.449260581336054e-05,
+      "loss": 1.0345,
+      "max_memory_allocated (GB)": 91.93,
+      "memory_allocated (GB)": 14.99,
+      "step": 170,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5832647681236267,
+      "learning_rate": 9.398266190719021e-05,
+      "loss": 1.0262,
+      "max_memory_allocated (GB)": 91.93,
+      "memory_allocated (GB)": 14.99,
+      "step": 180,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6518144011497498,
+      "learning_rate": 9.347271800101989e-05,
+      "loss": 1.0318,
+      "max_memory_allocated (GB)": 91.93,
+      "memory_allocated (GB)": 14.99,
+      "step": 190,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.49274373054504395,
+      "learning_rate": 9.296277409484956e-05,
+      "loss": 1.0298,
+      "max_memory_allocated (GB)": 91.93,
+      "memory_allocated (GB)": 14.99,
+      "step": 200,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5237769484519958,
+      "learning_rate": 9.245283018867925e-05,
+      "loss": 1.0176,
+      "max_memory_allocated (GB)": 91.93,
+      "memory_allocated (GB)": 14.99,
+      "step": 210,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.564319372177124,
+      "learning_rate": 9.194288628250894e-05,
+      "loss": 1.0172,
+      "max_memory_allocated (GB)": 91.93,
+      "memory_allocated (GB)": 14.99,
+      "step": 220,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4697343111038208,
+      "learning_rate": 9.14329423763386e-05,
+      "loss": 1.0262,
+      "max_memory_allocated (GB)": 91.93,
+      "memory_allocated (GB)": 14.99,
+      "step": 230,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.5207454562187195,
+      "learning_rate": 9.092299847016829e-05,
+      "loss": 1.024,
+      "max_memory_allocated (GB)": 91.93,
+      "memory_allocated (GB)": 14.99,
+      "step": 240,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4637609124183655,
+      "learning_rate": 9.041305456399796e-05,
+      "loss": 1.0069,
+      "max_memory_allocated (GB)": 91.93,
+      "memory_allocated (GB)": 14.99,
+      "step": 250,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.47436627745628357,
+      "learning_rate": 8.990311065782764e-05,
+      "loss": 1.0119,
+      "max_memory_allocated (GB)": 91.93,
+      "memory_allocated (GB)": 14.99,
+      "step": 260,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5096576809883118,
+      "learning_rate": 8.939316675165733e-05,
+      "loss": 1.0092,
+      "max_memory_allocated (GB)": 91.93,
+      "memory_allocated (GB)": 14.99,
+      "step": 270,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5780492424964905,
+      "learning_rate": 8.8883222845487e-05,
+      "loss": 1.0082,
+      "max_memory_allocated (GB)": 91.93,
+      "memory_allocated (GB)": 14.99,
+      "step": 280,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4528846740722656,
+      "learning_rate": 8.837327893931669e-05,
+      "loss": 1.0057,
+      "max_memory_allocated (GB)": 91.93,
+      "memory_allocated (GB)": 14.99,
+      "step": 290,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.5259899497032166,
+      "learning_rate": 8.786333503314635e-05,
+      "loss": 0.9989,
+      "max_memory_allocated (GB)": 91.93,
+      "memory_allocated (GB)": 14.99,
+      "step": 300,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5007658004760742,
+      "learning_rate": 8.735339112697604e-05,
+      "loss": 0.9997,
+      "max_memory_allocated (GB)": 91.93,
+      "memory_allocated (GB)": 14.99,
+      "step": 310,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4527484178543091,
+      "learning_rate": 8.684344722080571e-05,
+      "loss": 0.9949,
+      "max_memory_allocated (GB)": 91.93,
+      "memory_allocated (GB)": 14.99,
+      "step": 320,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4365575611591339,
+      "learning_rate": 8.633350331463539e-05,
+      "loss": 0.9943,
+      "max_memory_allocated (GB)": 91.93,
+      "memory_allocated (GB)": 14.99,
+      "step": 330,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.49428853392601013,
+      "learning_rate": 8.582355940846507e-05,
+      "loss": 1.0005,
+      "max_memory_allocated (GB)": 91.93,
+      "memory_allocated (GB)": 14.99,
+      "step": 340,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4611368775367737,
+      "learning_rate": 8.531361550229475e-05,
+      "loss": 0.9972,
+      "max_memory_allocated (GB)": 91.93,
+      "memory_allocated (GB)": 14.99,
+      "step": 350,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.43548157811164856,
+      "learning_rate": 8.480367159612444e-05,
+      "loss": 0.9833,
+      "max_memory_allocated (GB)": 91.93,
+      "memory_allocated (GB)": 14.99,
+      "step": 360,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4797479808330536,
+      "learning_rate": 8.42937276899541e-05,
+      "loss": 0.981,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 370,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.44958415627479553,
+      "learning_rate": 8.378378378378379e-05,
+      "loss": 0.9969,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 380,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4499351680278778,
+      "learning_rate": 8.327383987761347e-05,
+      "loss": 0.9847,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 390,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.45021358132362366,
+      "learning_rate": 8.276389597144315e-05,
+      "loss": 0.9874,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 400,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4754478335380554,
+      "learning_rate": 8.225395206527282e-05,
+      "loss": 0.9955,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 410,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.44393980503082275,
+      "learning_rate": 8.17440081591025e-05,
+      "loss": 0.9898,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 420,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.43429532647132874,
+      "learning_rate": 8.123406425293219e-05,
+      "loss": 0.9905,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 430,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4695710837841034,
+      "learning_rate": 8.072412034676186e-05,
+      "loss": 0.9702,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 440,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.40997833013534546,
+      "learning_rate": 8.021417644059154e-05,
+      "loss": 0.9825,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 450,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4330343008041382,
+      "learning_rate": 7.970423253442122e-05,
+      "loss": 0.9777,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 460,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.42674386501312256,
+      "learning_rate": 7.91942886282509e-05,
+      "loss": 0.9794,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 470,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4461188316345215,
+      "learning_rate": 7.868434472208057e-05,
+      "loss": 0.979,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 480,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4532679617404938,
+      "learning_rate": 7.817440081591025e-05,
+      "loss": 0.9764,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 490,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.42160096764564514,
+      "learning_rate": 7.766445690973994e-05,
+      "loss": 0.967,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 500,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.45428088307380676,
+      "learning_rate": 7.715451300356961e-05,
+      "loss": 0.975,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 510,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3874836564064026,
+      "learning_rate": 7.664456909739929e-05,
+      "loss": 0.9707,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 520,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4256057143211365,
+      "learning_rate": 7.613462519122897e-05,
+      "loss": 0.9775,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 530,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3986164331436157,
+      "learning_rate": 7.562468128505865e-05,
+      "loss": 0.972,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 540,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.4044497311115265,
+      "learning_rate": 7.511473737888832e-05,
+      "loss": 0.9725,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 550,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.439773291349411,
+      "learning_rate": 7.460479347271801e-05,
+      "loss": 0.9667,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 560,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.42315754294395447,
+      "learning_rate": 7.409484956654769e-05,
+      "loss": 0.9714,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 570,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.38059332966804504,
+      "learning_rate": 7.358490566037736e-05,
+      "loss": 0.979,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 580,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4393787086009979,
+      "learning_rate": 7.307496175420703e-05,
+      "loss": 0.9663,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 590,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4336337447166443,
+      "learning_rate": 7.256501784803672e-05,
+      "loss": 0.9661,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 600,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.41273126006126404,
+      "learning_rate": 7.20550739418664e-05,
+      "loss": 0.9672,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 610,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.45420095324516296,
+      "learning_rate": 7.154513003569607e-05,
+      "loss": 0.9668,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 620,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4193101227283478,
+      "learning_rate": 7.103518612952576e-05,
+      "loss": 0.9684,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 630,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4247801601886749,
+      "learning_rate": 7.052524222335543e-05,
+      "loss": 0.9652,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 640,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5118327140808105,
+      "learning_rate": 7.001529831718512e-05,
+      "loss": 0.9777,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 650,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.42659929394721985,
+      "learning_rate": 6.950535441101478e-05,
+      "loss": 0.9624,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 660,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.45405977964401245,
+      "learning_rate": 6.899541050484447e-05,
+      "loss": 0.9585,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 670,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.40387892723083496,
+      "learning_rate": 6.848546659867415e-05,
+      "loss": 0.9633,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 680,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.39345934987068176,
+      "learning_rate": 6.797552269250382e-05,
+      "loss": 0.9788,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 690,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4187304675579071,
+      "learning_rate": 6.746557878633351e-05,
+      "loss": 0.9651,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 700,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4518466293811798,
+      "learning_rate": 6.695563488016318e-05,
+      "loss": 0.965,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 710,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4474211037158966,
+      "learning_rate": 6.644569097399287e-05,
+      "loss": 0.9714,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 720,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.38997548818588257,
+      "learning_rate": 6.593574706782255e-05,
+      "loss": 0.9528,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 730,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.42444851994514465,
+      "learning_rate": 6.542580316165222e-05,
+      "loss": 0.9638,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 740,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4098523259162903,
+      "learning_rate": 6.491585925548191e-05,
+      "loss": 0.9534,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 750,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.42875897884368896,
+      "learning_rate": 6.440591534931157e-05,
+      "loss": 0.9631,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 760,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4514971077442169,
+      "learning_rate": 6.389597144314126e-05,
+      "loss": 0.9623,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 770,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.43437400460243225,
+      "learning_rate": 6.338602753697093e-05,
+      "loss": 0.9657,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 780,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.40089166164398193,
+      "learning_rate": 6.287608363080062e-05,
+      "loss": 0.9607,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 790,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3970320224761963,
+      "learning_rate": 6.23661397246303e-05,
+      "loss": 0.9623,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 800,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5439819693565369,
+      "learning_rate": 6.185619581845997e-05,
+      "loss": 0.9585,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 810,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.37487202882766724,
+      "learning_rate": 6.134625191228966e-05,
+      "loss": 0.9509,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 820,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.42697498202323914,
+      "learning_rate": 6.0836308006119326e-05,
+      "loss": 0.9525,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 830,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.38735342025756836,
+      "learning_rate": 6.032636409994901e-05,
+      "loss": 0.954,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 840,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4077759087085724,
+      "learning_rate": 5.981642019377869e-05,
+      "loss": 0.9528,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 850,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.38065391778945923,
+      "learning_rate": 5.930647628760837e-05,
+      "loss": 0.9589,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 860,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3892410695552826,
+      "learning_rate": 5.879653238143804e-05,
+      "loss": 0.9538,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 870,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3951966166496277,
+      "learning_rate": 5.8286588475267726e-05,
+      "loss": 0.9518,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 880,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5943160057067871,
+      "learning_rate": 5.777664456909741e-05,
+      "loss": 0.9437,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 890,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.440020889043808,
+      "learning_rate": 5.7266700662927075e-05,
+      "loss": 0.9635,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 900,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.44286438822746277,
+      "learning_rate": 5.6756756756756757e-05,
+      "loss": 0.9511,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 910,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4247153699398041,
+      "learning_rate": 5.624681285058644e-05,
+      "loss": 0.9475,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 920,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.42017248272895813,
+      "learning_rate": 5.573686894441612e-05,
+      "loss": 0.9436,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 930,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4182833731174469,
+      "learning_rate": 5.5226925038245794e-05,
+      "loss": 0.9557,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 940,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.41079315543174744,
+      "learning_rate": 5.4716981132075475e-05,
+      "loss": 0.9585,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 950,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3954004943370819,
+      "learning_rate": 5.4207037225905157e-05,
+      "loss": 0.9478,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 960,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.39996138215065,
+      "learning_rate": 5.369709331973484e-05,
+      "loss": 0.947,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 970,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.39325571060180664,
+      "learning_rate": 5.3187149413564506e-05,
+      "loss": 0.9473,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 980,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.43744412064552307,
+      "learning_rate": 5.267720550739419e-05,
+      "loss": 0.9437,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 990,
+      "total_memory_available (GB)": 94.62
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.36840781569480896,
+      "learning_rate": 5.216726160122387e-05,
+      "loss": 0.9441,
+      "max_memory_allocated (GB)": 91.97,
+      "memory_allocated (GB)": 14.99,
+      "step": 1000,
+      "total_memory_available (GB)": 94.62
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 2022,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 500,
+  "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:dd4c180c2753de2a707d546fc3073ec95a26b4c16fb44ff8754e797399512c42
+size 5880

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)

checkpoint-1500/README.md

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

checkpoint-1500/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": 32,
+  "lora_dropout": 0.1,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "k_proj",
+    "v_proj",
+    "o_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}

checkpoint-1500/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c2defbc903959f785a495b675953d393017047fe3cc7132eba2c9fc30a5f1c2
+size 13665592

checkpoint-1500/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-1500/global_step1500/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:a20ce447e78509e7b75fdd74cb94032f49d6c3ecff2abfa4c18cb3937ba24b35
+size 10229904

checkpoint-1500/global_step1500/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:a411f914f868edc6b297b568c88adfa2e57638006d153cb484e42650d958d8a0
+size 10229904

checkpoint-1500/global_step1500/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:305e24a00c2fda56c4853782833ae5575ffc0fae6665a631acaf389baf81299c
+size 10229968

checkpoint-1500/global_step1500/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:00c5e6f48f0575e61eb7f8cf2841f4766dd42f7885670191f32ddce054dbe624
+size 10229968

checkpoint-1500/global_step1500/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:936e784a83beeb00a8cdf3c008015494fc7365ad574f216f4628a3a61086c25c
+size 10229968

checkpoint-1500/global_step1500/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7677a1263aff0851c5544d4942d49198b6786735c02a1a4fbc8ae984b3ac679b
+size 10229968

checkpoint-1500/global_step1500/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:f8b8e21fdeebee5aead1e257a20fe9de80d3ba24541836696496b2c3087cad5d
+size 10229968

checkpoint-1500/global_step1500/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3a33c3d990ed1d8e6fb4fea9aff23309dd70d4e5e16dda37fad7d0e4e01e6e42
+size 10229968

checkpoint-1500/global_step1500/mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2cd7b89731690be67d9dbd7293d0c17e94171c9a1371b2ae65a59441c46db33c
+size 13740082

checkpoint-1500/latest

@@ -0,0 +1 @@
+global_step1500

checkpoint-1500/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ca21a878c0bb20065a7b92ab1e289e486e15ea4837b3a41b9116f480ec7aa724
+size 18032

checkpoint-1500/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e1902a238d9f23200cf562962dc21ab231af2270932085e2207b1e501534d30c
+size 18032

checkpoint-1500/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a1a997fd37e6ab137dc53b4b09331a9e0f028b8679a1614905e837f63ec94fb6
+size 18032

checkpoint-1500/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:099f92cc705087b82dff4bdc4ca07bad927f5d614c3b234c1340640a0c17da59
+size 18032