Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -1,3 +1,208 @@
----
-license: apache-2.0
----
+---
+base_model: ByteDance-Seed/Seed-Coder-8B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- axolotl
+- base_model:adapter:ByteDance-Seed/Seed-Coder-8B-Instruct
+- lora
+- transformers
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.17.1

adapter_config.json

@@ -0,0 +1,42 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "ByteDance-Seed/Seed-Coder-8B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": null,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 64,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "qalora_group_size": 16,
+  "r": 64,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "k_proj",
+    "o_proj",
+    "up_proj",
+    "v_proj",
+    "q_proj",
+    "down_proj",
+    "gate_proj"
+  ],
+  "target_parameters": [],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ffd8ec095fa64253d1a6c14570edbe235af77f620aafe631c8ec38a491400e32
+size 335605144

chat_template.jinja

@@ -0,0 +1,6 @@
+{% set has_system = messages[0]['role'] == 'system' %}{% if not has_system %}{{ bos_token + 'system
+You are an AI programming assistant, utilizing the Seed-Coder model, developed by ByteDance Seed, and you only answer questions related to computer science. For politically sensitive questions, security and privacy issues, and other non-computer science questions, you will refuse to answer.
+
+' + eos_token }}{% endif %}{% for message in messages %}{{ bos_token + message['role'] + '
+' + message['content'] | trim + eos_token }}{% endfor %}{% if add_generation_prompt %}{{ bos_token + 'assistant
+'}}{% endif %}

global_step120/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f2376d646164e138e7ca3718dc7b7bc85a6beb4647ed13a2dc55ae6f3726621
+size 503332453

global_step120/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:22e62cef33911886693e5b0ed56c95c48669c3c4cc0d32fe374024e91c884727
+size 503332517

global_step120/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:a67b042430d77540fe75b2ac93d5919553a84f983683827d4deda19467581a04
+size 503332517

global_step120/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:a30354d9281f1feb3272465d5361aff36d833e5d773a0bc68d90b1493495c857
+size 503332517

global_step120/mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:43095e90f0ad7e9c6496941cad3d3fbd4b2ecdeb87ffaa9bcdb57536c6f2b903
+size 335887513

latest

@@ -0,0 +1 @@
+global_step120

rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cbd3ca1d9c0f9c37f1fb2f5e734486f6fee40ca3170ebc50735e5f40a8308cca
+size 15429

rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a7fdb00d15d449daf3744dc0427f59acbda6acf30622e04a5efa8cf6dfb563e
+size 15429

rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d007caa01fca0f5efc6ac34e538c4316ae15676e418ebe6f84ac7268f1b49fdb
+size 15429

rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c01f8eff2f00bc0b64994298d5b79a0f8c67e1d081075cc49bbea5efe9322c0a
+size 15429

scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f524ac6f6272cafebe421fd53779255bb9c1bdeb5510ba7ecced18033e1523f
+size 1465

special_tokens_map.json

@@ -0,0 +1,30 @@
+{
+  "bos_token": {
+    "content": "<[begin▁of▁sentence]>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<[end▁of▁sentence]>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<[PAD▁TOKEN]>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "sep_token": {
+    "content": "<[SEP▁TOKEN]>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:db6520146c388c495a98bbea62ff6d00c0a8935bed33622e33bb33ec71aaafed
+size 11891696

tokenizer_config.json

@@ -0,0 +1,1036 @@
+{
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<[begin▁of▁sentence]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<[PAD▁TOKEN]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<[end▁of▁sentence]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<[UNK_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<[CLS_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "5": {
+      "content": "<[MASK_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "6": {
+      "content": "<[SEP▁TOKEN]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "7": {
+      "content": "<[PLHD7_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "8": {
+      "content": "<[PLHD8_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "9": {
+      "content": "<[PLHD9_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "10": {
+      "content": "<[PLHD10_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "11": {
+      "content": "<[PLHD11_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "12": {
+      "content": "<[PLHD12_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "13": {
+      "content": "<[PLHD13_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "14": {
+      "content": "<[PLHD14_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "15": {
+      "content": "<[PLHD15_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "16": {
+      "content": "<[PLHD16_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "17": {
+      "content": "<[PLHD17_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "18": {
+      "content": "<[PLHD18_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "19": {
+      "content": "<[PLHD19_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "20": {
+      "content": "<[PLHD20_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "21": {
+      "content": "<[PLHD21_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "22": {
+      "content": "<[PLHD22_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "23": {
+      "content": "<[PLHD23_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "24": {
+      "content": "<[PLHD24_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "25": {
+      "content": "<[PLHD25_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "26": {
+      "content": "<[PLHD26_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "27": {
+      "content": "<[PLHD27_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "28": {
+      "content": "<[PLHD28_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "29": {
+      "content": "<[PLHD29_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "30": {
+      "content": "<[PLHD30_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "31": {
+      "content": "<[PLHD31_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "32": {
+      "content": "<[PLHD32_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "33": {
+      "content": "<[PLHD33_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "34": {
+      "content": "<[PLHD34_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "35": {
+      "content": "<[PLHD35_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "36": {
+      "content": "<[PLHD36_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "37": {
+      "content": "<[PLHD37_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "38": {
+      "content": "<[PLHD38_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "39": {
+      "content": "<[PLHD39_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "40": {
+      "content": "<[PLHD40_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "41": {
+      "content": "<[PLHD41_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "42": {
+      "content": "<[PLHD42_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "43": {
+      "content": "<[PLHD43_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "44": {
+      "content": "<[PLHD44_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "45": {
+      "content": "<[PLHD45_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "46": {
+      "content": "<[PLHD46_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "47": {
+      "content": "<[PLHD47_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "48": {
+      "content": "<[PLHD48_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "49": {
+      "content": "<[PLHD49_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50": {
+      "content": "<[PLHD50_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "51": {
+      "content": "<[PLHD51_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "52": {
+      "content": "<[PLHD52_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "53": {
+      "content": "<[PLHD53_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "54": {
+      "content": "<[PLHD54_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "55": {
+      "content": "<[PLHD55_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "56": {
+      "content": "<[PLHD56_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "57": {
+      "content": "<[PLHD57_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "58": {
+      "content": "<[PLHD58_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "59": {
+      "content": "<[PLHD59_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "60": {
+      "content": "<[PLHD60_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "61": {
+      "content": "<[PLHD61_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "62": {
+      "content": "<[PLHD62_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "63": {
+      "content": "<[PLHD63_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "64": {
+      "content": "<[PLHD64_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "65": {
+      "content": "<[PLHD65_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "66": {
+      "content": "<[PLHD66_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "67": {
+      "content": "<[PLHD67_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "68": {
+      "content": "<[PLHD68_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "69": {
+      "content": "<[PLHD69_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "70": {
+      "content": "<[PLHD70_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "71": {
+      "content": "<[PLHD71_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "72": {
+      "content": "<[PLHD72_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "73": {
+      "content": "<[PLHD73_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "74": {
+      "content": "<[PLHD74_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "75": {
+      "content": "<[PLHD75_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "76": {
+      "content": "<[PLHD76_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "77": {
+      "content": "<[PLHD77_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "78": {
+      "content": "<[PLHD78_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "79": {
+      "content": "<[PLHD79_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "80": {
+      "content": "<[PLHD80_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "81": {
+      "content": "<[PLHD81_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "82": {
+      "content": "<[PLHD82_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "83": {
+      "content": "<[PLHD83_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "84": {
+      "content": "<[PLHD84_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "85": {
+      "content": "<[PLHD85_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "86": {
+      "content": "<[PLHD86_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "87": {
+      "content": "<[PLHD87_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "88": {
+      "content": "<[PLHD88_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "89": {
+      "content": "<[PLHD89_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "90": {
+      "content": "<[PLHD90_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "91": {
+      "content": "<[PLHD91_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "92": {
+      "content": "<[PLHD92_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "93": {
+      "content": "<[PLHD93_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "94": {
+      "content": "<[PLHD94_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "95": {
+      "content": "<[PLHD95_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "96": {
+      "content": "<[PLHD96_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "97": {
+      "content": "<[PLHD97_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "98": {
+      "content": "<[PLHD98_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "99": {
+      "content": "<[PLHD99_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "100": {
+      "content": "<[PLHD100_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "101": {
+      "content": "<[PLHD101_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "102": {
+      "content": "<[PLHD102_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "103": {
+      "content": "<[PLHD103_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "104": {
+      "content": "<[PLHD104_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "105": {
+      "content": "<[PLHD105_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "106": {
+      "content": "<[PLHD106_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<[PLHD107_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "108": {
+      "content": "<[PLHD108_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "109": {
+      "content": "<[PLHD109_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "110": {
+      "content": "<[PLHD110_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "111": {
+      "content": "<[PLHD111_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "112": {
+      "content": "<[PLHD112_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "113": {
+      "content": "<[PLHD113_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "114": {
+      "content": "<[PLHD114_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "115": {
+      "content": "<[PLHD115_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "116": {
+      "content": "<[PLHD116_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "117": {
+      "content": "<[PLHD117_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "118": {
+      "content": "<[PLHD118_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "119": {
+      "content": "<[PLHD119_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "120": {
+      "content": "<[PLHD120_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "121": {
+      "content": "<[PLHD121_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "122": {
+      "content": "<[PLHD122_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "123": {
+      "content": "<[PLHD123_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "124": {
+      "content": "<[fim-prefix]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "125": {
+      "content": "<[fim-suffix]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126": {
+      "content": "<[fim-middle]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "127": {
+      "content": "<[PLHD127_never_used_51bce0c785ca2f68081bfa7d91973934]>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "bos_token": "<[begin▁of▁sentence]>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<[end▁of▁sentence]>",
+  "extra_special_tokens": {},
+  "model_max_length": 32768,
+  "pad_token": "<[PAD▁TOKEN]>",
+  "sep_token": "<[SEP▁TOKEN]>",
+  "tokenizer_class": "PreTrainedTokenizerFast"
+}

trainer_state.json

@@ -0,0 +1,1354 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.27522935779816515,
+  "eval_steps": 500,
+  "global_step": 120,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0022935779816513763,
+      "grad_norm": 0.12869106233119965,
+      "learning_rate": 0.0,
+      "loss": 0.1978,
+      "memory/device_reserved (GiB)": 50.77,
+      "memory/max_active (GiB)": 48.85,
+      "memory/max_allocated (GiB)": 48.85,
+      "step": 1,
+      "tokens_per_second_per_gpu": 354.96
+    },
+    {
+      "epoch": 0.0045871559633027525,
+      "grad_norm": 0.15667210519313812,
+      "learning_rate": 4.7619047619047615e-06,
+      "loss": 0.2353,
+      "memory/device_reserved (GiB)": 50.77,
+      "memory/max_active (GiB)": 48.85,
+      "memory/max_allocated (GiB)": 48.85,
+      "step": 2,
+      "tokens_per_second_per_gpu": 406.37
+    },
+    {
+      "epoch": 0.006880733944954129,
+      "grad_norm": 0.2217973917722702,
+      "learning_rate": 9.523809523809523e-06,
+      "loss": 0.2243,
+      "memory/device_reserved (GiB)": 50.87,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 3,
+      "tokens_per_second_per_gpu": 371.18
+    },
+    {
+      "epoch": 0.009174311926605505,
+      "grad_norm": 0.15948686003684998,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 0.2392,
+      "memory/device_reserved (GiB)": 50.87,
+      "memory/max_active (GiB)": 48.85,
+      "memory/max_allocated (GiB)": 48.85,
+      "step": 4,
+      "tokens_per_second_per_gpu": 414.48
+    },
+    {
+      "epoch": 0.011467889908256881,
+      "grad_norm": 0.153566375374794,
+      "learning_rate": 1.9047619047619046e-05,
+      "loss": 0.2182,
+      "memory/device_reserved (GiB)": 50.87,
+      "memory/max_active (GiB)": 48.93,
+      "memory/max_allocated (GiB)": 48.93,
+      "step": 5,
+      "tokens_per_second_per_gpu": 369.22
+    },
+    {
+      "epoch": 0.013761467889908258,
+      "grad_norm": 0.1521972268819809,
+      "learning_rate": 2.380952380952381e-05,
+      "loss": 0.2112,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 49.04,
+      "memory/max_allocated (GiB)": 49.04,
+      "step": 6,
+      "tokens_per_second_per_gpu": 429.31
+    },
+    {
+      "epoch": 0.016055045871559634,
+      "grad_norm": 0.168710395693779,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 0.226,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 7,
+      "tokens_per_second_per_gpu": 417.78
+    },
+    {
+      "epoch": 0.01834862385321101,
+      "grad_norm": 0.13864850997924805,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.1884,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 8,
+      "tokens_per_second_per_gpu": 439.56
+    },
+    {
+      "epoch": 0.020642201834862386,
+      "grad_norm": 0.15227903425693512,
+      "learning_rate": 3.809523809523809e-05,
+      "loss": 0.1996,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 48.93,
+      "memory/max_allocated (GiB)": 48.93,
+      "step": 9,
+      "tokens_per_second_per_gpu": 411.33
+    },
+    {
+      "epoch": 0.022935779816513763,
+      "grad_norm": 0.13421630859375,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 0.1599,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 10,
+      "tokens_per_second_per_gpu": 496.3
+    },
+    {
+      "epoch": 0.02522935779816514,
+      "grad_norm": 0.14955134689807892,
+      "learning_rate": 4.761904761904762e-05,
+      "loss": 0.1735,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 49.0,
+      "memory/max_allocated (GiB)": 49.0,
+      "step": 11,
+      "tokens_per_second_per_gpu": 372.95
+    },
+    {
+      "epoch": 0.027522935779816515,
+      "grad_norm": 0.1432778388261795,
+      "learning_rate": 5.2380952380952384e-05,
+      "loss": 0.1515,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 49.0,
+      "memory/max_allocated (GiB)": 49.0,
+      "step": 12,
+      "tokens_per_second_per_gpu": 398.65
+    },
+    {
+      "epoch": 0.02981651376146789,
+      "grad_norm": 0.14163611829280853,
+      "learning_rate": 5.714285714285714e-05,
+      "loss": 0.1517,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 13,
+      "tokens_per_second_per_gpu": 440.5
+    },
+    {
+      "epoch": 0.03211009174311927,
+      "grad_norm": 0.15477906167507172,
+      "learning_rate": 6.19047619047619e-05,
+      "loss": 0.1444,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 48.89,
+      "memory/max_allocated (GiB)": 48.89,
+      "step": 14,
+      "tokens_per_second_per_gpu": 385.32
+    },
+    {
+      "epoch": 0.034403669724770644,
+      "grad_norm": 0.1055532768368721,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 0.1292,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 48.93,
+      "memory/max_allocated (GiB)": 48.93,
+      "step": 15,
+      "tokens_per_second_per_gpu": 453.02
+    },
+    {
+      "epoch": 0.03669724770642202,
+      "grad_norm": 0.10180933028459549,
+      "learning_rate": 7.142857142857143e-05,
+      "loss": 0.1208,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 49.0,
+      "memory/max_allocated (GiB)": 49.0,
+      "step": 16,
+      "tokens_per_second_per_gpu": 474.27
+    },
+    {
+      "epoch": 0.0389908256880734,
+      "grad_norm": 0.07999677956104279,
+      "learning_rate": 7.619047619047618e-05,
+      "loss": 0.132,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 48.89,
+      "memory/max_allocated (GiB)": 48.89,
+      "step": 17,
+      "tokens_per_second_per_gpu": 382.05
+    },
+    {
+      "epoch": 0.04128440366972477,
+      "grad_norm": 0.09194924682378769,
+      "learning_rate": 8.095238095238096e-05,
+      "loss": 0.1067,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 48.81,
+      "memory/max_allocated (GiB)": 48.81,
+      "step": 18,
+      "tokens_per_second_per_gpu": 398.61
+    },
+    {
+      "epoch": 0.04357798165137615,
+      "grad_norm": 0.0931428000330925,
+      "learning_rate": 8.571428571428571e-05,
+      "loss": 0.1088,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 48.93,
+      "memory/max_allocated (GiB)": 48.93,
+      "step": 19,
+      "tokens_per_second_per_gpu": 447.07
+    },
+    {
+      "epoch": 0.045871559633027525,
+      "grad_norm": 0.06202042102813721,
+      "learning_rate": 9.047619047619048e-05,
+      "loss": 0.0962,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 49.0,
+      "memory/max_allocated (GiB)": 49.0,
+      "step": 20,
+      "tokens_per_second_per_gpu": 382.57
+    },
+    {
+      "epoch": 0.0481651376146789,
+      "grad_norm": 0.04220607504248619,
+      "learning_rate": 9.523809523809524e-05,
+      "loss": 0.0963,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 48.85,
+      "memory/max_allocated (GiB)": 48.85,
+      "step": 21,
+      "tokens_per_second_per_gpu": 423.29
+    },
+    {
+      "epoch": 0.05045871559633028,
+      "grad_norm": 0.050066106021404266,
+      "learning_rate": 0.0001,
+      "loss": 0.1032,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 48.89,
+      "memory/max_allocated (GiB)": 48.89,
+      "step": 22,
+      "tokens_per_second_per_gpu": 381.35
+    },
+    {
+      "epoch": 0.052752293577981654,
+      "grad_norm": 0.0557384118437767,
+      "learning_rate": 9.999856734543933e-05,
+      "loss": 0.1025,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 23,
+      "tokens_per_second_per_gpu": 393.62
+    },
+    {
+      "epoch": 0.05504587155963303,
+      "grad_norm": 0.04612402245402336,
+      "learning_rate": 9.999426946385727e-05,
+      "loss": 0.0985,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 48.89,
+      "memory/max_allocated (GiB)": 48.89,
+      "step": 24,
+      "tokens_per_second_per_gpu": 515.46
+    },
+    {
+      "epoch": 0.05733944954128441,
+      "grad_norm": 0.09721734374761581,
+      "learning_rate": 9.998710660154898e-05,
+      "loss": 0.1062,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 48.81,
+      "memory/max_allocated (GiB)": 48.81,
+      "step": 25,
+      "tokens_per_second_per_gpu": 398.15
+    },
+    {
+      "epoch": 0.05963302752293578,
+      "grad_norm": 0.036745935678482056,
+      "learning_rate": 9.997707916899079e-05,
+      "loss": 0.1045,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 26,
+      "tokens_per_second_per_gpu": 422.42
+    },
+    {
+      "epoch": 0.06192660550458716,
+      "grad_norm": 0.04298936203122139,
+      "learning_rate": 9.996418774081658e-05,
+      "loss": 0.0923,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 48.89,
+      "memory/max_allocated (GiB)": 48.89,
+      "step": 27,
+      "tokens_per_second_per_gpu": 440.87
+    },
+    {
+      "epoch": 0.06422018348623854,
+      "grad_norm": 0.033536747097969055,
+      "learning_rate": 9.994843305578486e-05,
+      "loss": 0.096,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 48.89,
+      "memory/max_allocated (GiB)": 48.89,
+      "step": 28,
+      "tokens_per_second_per_gpu": 370.28
+    },
+    {
+      "epoch": 0.06651376146788991,
+      "grad_norm": 0.03256046772003174,
+      "learning_rate": 9.99298160167365e-05,
+      "loss": 0.0832,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 48.81,
+      "memory/max_allocated (GiB)": 48.81,
+      "step": 29,
+      "tokens_per_second_per_gpu": 357.19
+    },
+    {
+      "epoch": 0.06880733944954129,
+      "grad_norm": 0.042709868401288986,
+      "learning_rate": 9.990833769054293e-05,
+      "loss": 0.086,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 48.93,
+      "memory/max_allocated (GiB)": 48.93,
+      "step": 30,
+      "tokens_per_second_per_gpu": 441.89
+    },
+    {
+      "epoch": 0.07110091743119266,
+      "grad_norm": 0.04347776621580124,
+      "learning_rate": 9.988399930804504e-05,
+      "loss": 0.1,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 48.77,
+      "memory/max_allocated (GiB)": 48.77,
+      "step": 31,
+      "tokens_per_second_per_gpu": 348.66
+    },
+    {
+      "epoch": 0.07339449541284404,
+      "grad_norm": 0.030414681881666183,
+      "learning_rate": 9.985680226398261e-05,
+      "loss": 0.0811,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 49.0,
+      "memory/max_allocated (GiB)": 49.0,
+      "step": 32,
+      "tokens_per_second_per_gpu": 435.28
+    },
+    {
+      "epoch": 0.07568807339449542,
+      "grad_norm": 0.034023743122816086,
+      "learning_rate": 9.98267481169144e-05,
+      "loss": 0.0743,
+      "memory/device_reserved (GiB)": 50.93,
+      "memory/max_active (GiB)": 49.0,
+      "memory/max_allocated (GiB)": 49.0,
+      "step": 33,
+      "tokens_per_second_per_gpu": 482.51
+    },
+    {
+      "epoch": 0.0779816513761468,
+      "grad_norm": 0.03136487305164337,
+      "learning_rate": 9.979383858912885e-05,
+      "loss": 0.0739,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 49.08,
+      "memory/max_allocated (GiB)": 49.08,
+      "step": 34,
+      "tokens_per_second_per_gpu": 496.59
+    },
+    {
+      "epoch": 0.08027522935779817,
+      "grad_norm": 0.028108298778533936,
+      "learning_rate": 9.975807556654537e-05,
+      "loss": 0.077,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 35,
+      "tokens_per_second_per_gpu": 349.1
+    },
+    {
+      "epoch": 0.08256880733944955,
+      "grad_norm": 0.028020795434713364,
+      "learning_rate": 9.971946109860626e-05,
+      "loss": 0.0775,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.81,
+      "memory/max_allocated (GiB)": 48.81,
+      "step": 36,
+      "tokens_per_second_per_gpu": 351.02
+    },
+    {
+      "epoch": 0.08486238532110092,
+      "grad_norm": 0.028756650164723396,
+      "learning_rate": 9.967799739815925e-05,
+      "loss": 0.0788,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.93,
+      "memory/max_allocated (GiB)": 48.93,
+      "step": 37,
+      "tokens_per_second_per_gpu": 534.52
+    },
+    {
+      "epoch": 0.0871559633027523,
+      "grad_norm": 0.02806459739804268,
+      "learning_rate": 9.963368684133072e-05,
+      "loss": 0.0809,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 38,
+      "tokens_per_second_per_gpu": 367.94
+    },
+    {
+      "epoch": 0.08944954128440367,
+      "grad_norm": 0.02387731708586216,
+      "learning_rate": 9.958653196738954e-05,
+      "loss": 0.0642,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 49.04,
+      "memory/max_allocated (GiB)": 49.04,
+      "step": 39,
+      "tokens_per_second_per_gpu": 466.74
+    },
+    {
+      "epoch": 0.09174311926605505,
+      "grad_norm": 0.027889851480722427,
+      "learning_rate": 9.953653547860151e-05,
+      "loss": 0.0904,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.89,
+      "memory/max_allocated (GiB)": 48.89,
+      "step": 40,
+      "tokens_per_second_per_gpu": 371.51
+    },
+    {
+      "epoch": 0.09403669724770643,
+      "grad_norm": 0.031659577041864395,
+      "learning_rate": 9.948370024007454e-05,
+      "loss": 0.081,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.93,
+      "memory/max_allocated (GiB)": 48.93,
+      "step": 41,
+      "tokens_per_second_per_gpu": 479.04
+    },
+    {
+      "epoch": 0.0963302752293578,
+      "grad_norm": 0.03186093270778656,
+      "learning_rate": 9.942802927959443e-05,
+      "loss": 0.0881,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.93,
+      "memory/max_allocated (GiB)": 48.93,
+      "step": 42,
+      "tokens_per_second_per_gpu": 364.73
+    },
+    {
+      "epoch": 0.09862385321100918,
+      "grad_norm": 0.0313677079975605,
+      "learning_rate": 9.936952578745142e-05,
+      "loss": 0.0808,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.89,
+      "memory/max_allocated (GiB)": 48.89,
+      "step": 43,
+      "tokens_per_second_per_gpu": 418.0
+    },
+    {
+      "epoch": 0.10091743119266056,
+      "grad_norm": 0.0264989472925663,
+      "learning_rate": 9.93081931162573e-05,
+      "loss": 0.0664,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 49.0,
+      "memory/max_allocated (GiB)": 49.0,
+      "step": 44,
+      "tokens_per_second_per_gpu": 439.24
+    },
+    {
+      "epoch": 0.10321100917431193,
+      "grad_norm": 0.026272334158420563,
+      "learning_rate": 9.92440347807533e-05,
+      "loss": 0.0683,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 49.04,
+      "memory/max_allocated (GiB)": 49.04,
+      "step": 45,
+      "tokens_per_second_per_gpu": 482.81
+    },
+    {
+      "epoch": 0.10550458715596331,
+      "grad_norm": 0.029066840186715126,
+      "learning_rate": 9.91770544576087e-05,
+      "loss": 0.0737,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.93,
+      "memory/max_allocated (GiB)": 48.93,
+      "step": 46,
+      "tokens_per_second_per_gpu": 389.87
+    },
+    {
+      "epoch": 0.10779816513761468,
+      "grad_norm": 0.024542706087231636,
+      "learning_rate": 9.910725598521013e-05,
+      "loss": 0.0737,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.93,
+      "memory/max_allocated (GiB)": 48.93,
+      "step": 47,
+      "tokens_per_second_per_gpu": 473.12
+    },
+    {
+      "epoch": 0.11009174311926606,
+      "grad_norm": 0.042941153049468994,
+      "learning_rate": 9.90346433634416e-05,
+      "loss": 0.0951,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.81,
+      "memory/max_allocated (GiB)": 48.81,
+      "step": 48,
+      "tokens_per_second_per_gpu": 325.12
+    },
+    {
+      "epoch": 0.11238532110091744,
+      "grad_norm": 0.029044413939118385,
+      "learning_rate": 9.89592207534552e-05,
+      "loss": 0.0745,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.73,
+      "memory/max_allocated (GiB)": 48.73,
+      "step": 49,
+      "tokens_per_second_per_gpu": 315.62
+    },
+    {
+      "epoch": 0.11467889908256881,
+      "grad_norm": 0.028920788317918777,
+      "learning_rate": 9.888099247743283e-05,
+      "loss": 0.0818,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 50,
+      "tokens_per_second_per_gpu": 441.3
+    },
+    {
+      "epoch": 0.11697247706422019,
+      "grad_norm": 0.026095205917954445,
+      "learning_rate": 9.879996301833833e-05,
+      "loss": 0.0688,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 51,
+      "tokens_per_second_per_gpu": 386.22
+    },
+    {
+      "epoch": 0.11926605504587157,
+      "grad_norm": 0.024823926389217377,
+      "learning_rate": 9.871613701966067e-05,
+      "loss": 0.0701,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 52,
+      "tokens_per_second_per_gpu": 511.32
+    },
+    {
+      "epoch": 0.12155963302752294,
+      "grad_norm": 0.036093298345804214,
+      "learning_rate": 9.862951928514782e-05,
+      "loss": 0.0823,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 49.0,
+      "memory/max_allocated (GiB)": 49.0,
+      "step": 53,
+      "tokens_per_second_per_gpu": 323.2
+    },
+    {
+      "epoch": 0.12385321100917432,
+      "grad_norm": 0.03257686272263527,
+      "learning_rate": 9.854011477853146e-05,
+      "loss": 0.0769,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 49.04,
+      "memory/max_allocated (GiB)": 49.04,
+      "step": 54,
+      "tokens_per_second_per_gpu": 447.62
+    },
+    {
+      "epoch": 0.12614678899082568,
+      "grad_norm": 0.03413158655166626,
+      "learning_rate": 9.844792862324258e-05,
+      "loss": 0.0728,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 55,
+      "tokens_per_second_per_gpu": 451.05
+    },
+    {
+      "epoch": 0.12844036697247707,
+      "grad_norm": 0.02947932481765747,
+      "learning_rate": 9.835296610211779e-05,
+      "loss": 0.0713,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 56,
+      "tokens_per_second_per_gpu": 457.44
+    },
+    {
+      "epoch": 0.13073394495412843,
+      "grad_norm": 0.0220651775598526,
+      "learning_rate": 9.825523265709666e-05,
+      "loss": 0.0607,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 49.0,
+      "memory/max_allocated (GiB)": 49.0,
+      "step": 57,
+      "tokens_per_second_per_gpu": 456.49
+    },
+    {
+      "epoch": 0.13302752293577982,
+      "grad_norm": 0.026394842192530632,
+      "learning_rate": 9.815473388890983e-05,
+      "loss": 0.0716,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.93,
+      "memory/max_allocated (GiB)": 48.93,
+      "step": 58,
+      "tokens_per_second_per_gpu": 393.95
+    },
+    {
+      "epoch": 0.1353211009174312,
+      "grad_norm": 0.027936838567256927,
+      "learning_rate": 9.805147555675805e-05,
+      "loss": 0.0738,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 49.0,
+      "memory/max_allocated (GiB)": 49.0,
+      "step": 59,
+      "tokens_per_second_per_gpu": 464.83
+    },
+    {
+      "epoch": 0.13761467889908258,
+      "grad_norm": 0.023982539772987366,
+      "learning_rate": 9.794546357798208e-05,
+      "loss": 0.0608,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 60,
+      "tokens_per_second_per_gpu": 450.66
+    },
+    {
+      "epoch": 0.13990825688073394,
+      "grad_norm": 0.027479754760861397,
+      "learning_rate": 9.783670402772379e-05,
+      "loss": 0.0672,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 49.0,
+      "memory/max_allocated (GiB)": 49.0,
+      "step": 61,
+      "tokens_per_second_per_gpu": 455.94
+    },
+    {
+      "epoch": 0.14220183486238533,
+      "grad_norm": 0.02617599070072174,
+      "learning_rate": 9.772520313857775e-05,
+      "loss": 0.0804,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 62,
+      "tokens_per_second_per_gpu": 394.85
+    },
+    {
+      "epoch": 0.1444954128440367,
+      "grad_norm": 0.030884992331266403,
+      "learning_rate": 9.761096730023432e-05,
+      "loss": 0.0768,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.89,
+      "memory/max_allocated (GiB)": 48.89,
+      "step": 63,
+      "tokens_per_second_per_gpu": 446.63
+    },
+    {
+      "epoch": 0.14678899082568808,
+      "grad_norm": 0.027579287067055702,
+      "learning_rate": 9.749400305911322e-05,
+      "loss": 0.0659,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.93,
+      "memory/max_allocated (GiB)": 48.93,
+      "step": 64,
+      "tokens_per_second_per_gpu": 484.34
+    },
+    {
+      "epoch": 0.14908256880733944,
+      "grad_norm": 0.030303625389933586,
+      "learning_rate": 9.737431711798864e-05,
+      "loss": 0.0645,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.85,
+      "memory/max_allocated (GiB)": 48.85,
+      "step": 65,
+      "tokens_per_second_per_gpu": 437.07
+    },
+    {
+      "epoch": 0.15137614678899083,
+      "grad_norm": 0.027446158230304718,
+      "learning_rate": 9.725191633560491e-05,
+      "loss": 0.08,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 66,
+      "tokens_per_second_per_gpu": 411.5
+    },
+    {
+      "epoch": 0.1536697247706422,
+      "grad_norm": 0.03177177160978317,
+      "learning_rate": 9.712680772628364e-05,
+      "loss": 0.0801,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.93,
+      "memory/max_allocated (GiB)": 48.93,
+      "step": 67,
+      "tokens_per_second_per_gpu": 429.18
+    },
+    {
+      "epoch": 0.1559633027522936,
+      "grad_norm": 0.0288909412920475,
+      "learning_rate": 9.69989984595216e-05,
+      "loss": 0.0707,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 49.04,
+      "memory/max_allocated (GiB)": 49.04,
+      "step": 68,
+      "tokens_per_second_per_gpu": 408.55
+    },
+    {
+      "epoch": 0.15825688073394495,
+      "grad_norm": 0.02751251310110092,
+      "learning_rate": 9.686849585957994e-05,
+      "loss": 0.0736,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.89,
+      "memory/max_allocated (GiB)": 48.89,
+      "step": 69,
+      "tokens_per_second_per_gpu": 420.0
+    },
+    {
+      "epoch": 0.16055045871559634,
+      "grad_norm": 0.023428168147802353,
+      "learning_rate": 9.673530740506447e-05,
+      "loss": 0.0648,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 70,
+      "tokens_per_second_per_gpu": 512.59
+    },
+    {
+      "epoch": 0.1628440366972477,
+      "grad_norm": 0.031534772366285324,
+      "learning_rate": 9.659944072849707e-05,
+      "loss": 0.0818,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.93,
+      "memory/max_allocated (GiB)": 48.93,
+      "step": 71,
+      "tokens_per_second_per_gpu": 456.9
+    },
+    {
+      "epoch": 0.1651376146788991,
+      "grad_norm": 0.027208171784877777,
+      "learning_rate": 9.646090361587827e-05,
+      "loss": 0.0709,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.93,
+      "memory/max_allocated (GiB)": 48.93,
+      "step": 72,
+      "tokens_per_second_per_gpu": 378.48
+    },
+    {
+      "epoch": 0.16743119266055045,
+      "grad_norm": 0.02961639314889908,
+      "learning_rate": 9.631970400624113e-05,
+      "loss": 0.0764,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.81,
+      "memory/max_allocated (GiB)": 48.81,
+      "step": 73,
+      "tokens_per_second_per_gpu": 316.38
+    },
+    {
+      "epoch": 0.16972477064220184,
+      "grad_norm": 0.027367761358618736,
+      "learning_rate": 9.617584999119625e-05,
+      "loss": 0.0672,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.85,
+      "memory/max_allocated (GiB)": 48.85,
+      "step": 74,
+      "tokens_per_second_per_gpu": 402.44
+    },
+    {
+      "epoch": 0.1720183486238532,
+      "grad_norm": 0.030167503282427788,
+      "learning_rate": 9.602934981446803e-05,
+      "loss": 0.0743,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 75,
+      "tokens_per_second_per_gpu": 531.29
+    },
+    {
+      "epoch": 0.1743119266055046,
+      "grad_norm": 0.0387263149023056,
+      "learning_rate": 9.588021187142235e-05,
+      "loss": 0.083,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.81,
+      "memory/max_allocated (GiB)": 48.81,
+      "step": 76,
+      "tokens_per_second_per_gpu": 424.59
+    },
+    {
+      "epoch": 0.17660550458715596,
+      "grad_norm": 0.027617793530225754,
+      "learning_rate": 9.572844470858537e-05,
+      "loss": 0.0769,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 77,
+      "tokens_per_second_per_gpu": 461.9
+    },
+    {
+      "epoch": 0.17889908256880735,
+      "grad_norm": 0.029771512374281883,
+      "learning_rate": 9.557405702315381e-05,
+      "loss": 0.0658,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.93,
+      "memory/max_allocated (GiB)": 48.93,
+      "step": 78,
+      "tokens_per_second_per_gpu": 475.77
+    },
+    {
+      "epoch": 0.1811926605504587,
+      "grad_norm": 0.029358675703406334,
+      "learning_rate": 9.541705766249655e-05,
+      "loss": 0.066,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 49.0,
+      "memory/max_allocated (GiB)": 49.0,
+      "step": 79,
+      "tokens_per_second_per_gpu": 489.33
+    },
+    {
+      "epoch": 0.1834862385321101,
+      "grad_norm": 0.023111771792173386,
+      "learning_rate": 9.525745562364756e-05,
+      "loss": 0.066,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 80,
+      "tokens_per_second_per_gpu": 382.84
+    },
+    {
+      "epoch": 0.18577981651376146,
+      "grad_norm": 0.029448291286826134,
+      "learning_rate": 9.509526005279044e-05,
+      "loss": 0.0608,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.81,
+      "memory/max_allocated (GiB)": 48.81,
+      "step": 81,
+      "tokens_per_second_per_gpu": 415.81
+    },
+    {
+      "epoch": 0.18807339449541285,
+      "grad_norm": 0.02794116735458374,
+      "learning_rate": 9.493048024473412e-05,
+      "loss": 0.0736,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 49.0,
+      "memory/max_allocated (GiB)": 49.0,
+      "step": 82,
+      "tokens_per_second_per_gpu": 400.02
+    },
+    {
+      "epoch": 0.19036697247706422,
+      "grad_norm": 0.04534873738884926,
+      "learning_rate": 9.476312564238034e-05,
+      "loss": 0.0673,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.93,
+      "memory/max_allocated (GiB)": 48.93,
+      "step": 83,
+      "tokens_per_second_per_gpu": 369.1
+    },
+    {
+      "epoch": 0.1926605504587156,
+      "grad_norm": 0.026540853083133698,
+      "learning_rate": 9.459320583618252e-05,
+      "loss": 0.0558,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 49.04,
+      "memory/max_allocated (GiB)": 49.04,
+      "step": 84,
+      "tokens_per_second_per_gpu": 611.61
+    },
+    {
+      "epoch": 0.19495412844036697,
+      "grad_norm": 0.03129403293132782,
+      "learning_rate": 9.442073056359604e-05,
+      "loss": 0.0741,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.93,
+      "memory/max_allocated (GiB)": 48.93,
+      "step": 85,
+      "tokens_per_second_per_gpu": 492.16
+    },
+    {
+      "epoch": 0.19724770642201836,
+      "grad_norm": 0.027526071295142174,
+      "learning_rate": 9.424570970852034e-05,
+      "loss": 0.0733,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.85,
+      "memory/max_allocated (GiB)": 48.85,
+      "step": 86,
+      "tokens_per_second_per_gpu": 427.76
+    },
+    {
+      "epoch": 0.19954128440366972,
+      "grad_norm": 0.025468798354268074,
+      "learning_rate": 9.406815330073244e-05,
+      "loss": 0.0613,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.93,
+      "memory/max_allocated (GiB)": 48.93,
+      "step": 87,
+      "tokens_per_second_per_gpu": 462.82
+    },
+    {
+      "epoch": 0.2018348623853211,
+      "grad_norm": 0.029043635353446007,
+      "learning_rate": 9.388807151531229e-05,
+      "loss": 0.0758,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.89,
+      "memory/max_allocated (GiB)": 48.89,
+      "step": 88,
+      "tokens_per_second_per_gpu": 353.91
+    },
+    {
+      "epoch": 0.20412844036697247,
+      "grad_norm": 0.03196391835808754,
+      "learning_rate": 9.37054746720595e-05,
+      "loss": 0.0678,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 49.0,
+      "memory/max_allocated (GiB)": 49.0,
+      "step": 89,
+      "tokens_per_second_per_gpu": 411.71
+    },
+    {
+      "epoch": 0.20642201834862386,
+      "grad_norm": 0.033272091299295425,
+      "learning_rate": 9.352037323490208e-05,
+      "loss": 0.0722,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.85,
+      "memory/max_allocated (GiB)": 48.85,
+      "step": 90,
+      "tokens_per_second_per_gpu": 398.81
+    },
+    {
+      "epoch": 0.20871559633027523,
+      "grad_norm": 0.03096090629696846,
+      "learning_rate": 9.333277781129678e-05,
+      "loss": 0.0809,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.89,
+      "memory/max_allocated (GiB)": 48.89,
+      "step": 91,
+      "tokens_per_second_per_gpu": 393.81
+    },
+    {
+      "epoch": 0.21100917431192662,
+      "grad_norm": 0.026267440989613533,
+      "learning_rate": 9.314269915162114e-05,
+      "loss": 0.0604,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 49.0,
+      "memory/max_allocated (GiB)": 49.0,
+      "step": 92,
+      "tokens_per_second_per_gpu": 453.78
+    },
+    {
+      "epoch": 0.21330275229357798,
+      "grad_norm": 0.02608361840248108,
+      "learning_rate": 9.295014814855753e-05,
+      "loss": 0.0663,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.93,
+      "memory/max_allocated (GiB)": 48.93,
+      "step": 93,
+      "tokens_per_second_per_gpu": 430.47
+    },
+    {
+      "epoch": 0.21559633027522937,
+      "grad_norm": 0.024829065427184105,
+      "learning_rate": 9.275513583646884e-05,
+      "loss": 0.0598,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.81,
+      "memory/max_allocated (GiB)": 48.81,
+      "step": 94,
+      "tokens_per_second_per_gpu": 384.01
+    },
+    {
+      "epoch": 0.21788990825688073,
+      "grad_norm": 0.03385532647371292,
+      "learning_rate": 9.255767339076622e-05,
+      "loss": 0.0719,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.93,
+      "memory/max_allocated (GiB)": 48.93,
+      "step": 95,
+      "tokens_per_second_per_gpu": 440.35
+    },
+    {
+      "epoch": 0.22018348623853212,
+      "grad_norm": 0.029608217999339104,
+      "learning_rate": 9.23577721272686e-05,
+      "loss": 0.094,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 49.04,
+      "memory/max_allocated (GiB)": 49.04,
+      "step": 96,
+      "tokens_per_second_per_gpu": 485.56
+    },
+    {
+      "epoch": 0.22247706422018348,
+      "grad_norm": 0.02693762816488743,
+      "learning_rate": 9.215544350155422e-05,
+      "loss": 0.0755,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 97,
+      "tokens_per_second_per_gpu": 432.16
+    },
+    {
+      "epoch": 0.22477064220183487,
+      "grad_norm": 0.02771424688398838,
+      "learning_rate": 9.195069910830427e-05,
+      "loss": 0.0692,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.89,
+      "memory/max_allocated (GiB)": 48.89,
+      "step": 98,
+      "tokens_per_second_per_gpu": 412.93
+    },
+    {
+      "epoch": 0.22706422018348624,
+      "grad_norm": 0.02276022732257843,
+      "learning_rate": 9.174355068063828e-05,
+      "loss": 0.0637,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 49.0,
+      "memory/max_allocated (GiB)": 49.0,
+      "step": 99,
+      "tokens_per_second_per_gpu": 418.24
+    },
+    {
+      "epoch": 0.22935779816513763,
+      "grad_norm": 0.026155246421694756,
+      "learning_rate": 9.15340100894418e-05,
+      "loss": 0.0698,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 100,
+      "tokens_per_second_per_gpu": 403.6
+    },
+    {
+      "epoch": 0.231651376146789,
+      "grad_norm": 0.022778436541557312,
+      "learning_rate": 9.132208934268622e-05,
+      "loss": 0.0654,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 49.0,
+      "memory/max_allocated (GiB)": 49.0,
+      "step": 101,
+      "tokens_per_second_per_gpu": 491.32
+    },
+    {
+      "epoch": 0.23394495412844038,
+      "grad_norm": 0.04701945558190346,
+      "learning_rate": 9.110780058474052e-05,
+      "loss": 0.0741,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.85,
+      "memory/max_allocated (GiB)": 48.85,
+      "step": 102,
+      "tokens_per_second_per_gpu": 444.03
+    },
+    {
+      "epoch": 0.23623853211009174,
+      "grad_norm": 0.030211661010980606,
+      "learning_rate": 9.08911560956753e-05,
+      "loss": 0.0789,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 103,
+      "tokens_per_second_per_gpu": 514.87
+    },
+    {
+      "epoch": 0.23853211009174313,
+      "grad_norm": 0.026159459725022316,
+      "learning_rate": 9.067216829055922e-05,
+      "loss": 0.0637,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 104,
+      "tokens_per_second_per_gpu": 446.47
+    },
+    {
+      "epoch": 0.2408256880733945,
+      "grad_norm": 0.02918146923184395,
+      "learning_rate": 9.045084971874738e-05,
+      "loss": 0.0727,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.89,
+      "memory/max_allocated (GiB)": 48.89,
+      "step": 105,
+      "tokens_per_second_per_gpu": 425.37
+    },
+    {
+      "epoch": 0.24311926605504589,
+      "grad_norm": 0.03170175105333328,
+      "learning_rate": 9.022721306316222e-05,
+      "loss": 0.0857,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.85,
+      "memory/max_allocated (GiB)": 48.85,
+      "step": 106,
+      "tokens_per_second_per_gpu": 301.79
+    },
+    {
+      "epoch": 0.24541284403669725,
+      "grad_norm": 0.032674651592969894,
+      "learning_rate": 9.000127113956674e-05,
+      "loss": 0.0795,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.77,
+      "memory/max_allocated (GiB)": 48.77,
+      "step": 107,
+      "tokens_per_second_per_gpu": 338.41
+    },
+    {
+      "epoch": 0.24770642201834864,
+      "grad_norm": 0.026492780074477196,
+      "learning_rate": 8.977303689583e-05,
+      "loss": 0.0775,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.89,
+      "memory/max_allocated (GiB)": 48.89,
+      "step": 108,
+      "tokens_per_second_per_gpu": 383.35
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.0290480125695467,
+      "learning_rate": 8.954252341118523e-05,
+      "loss": 0.076,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.93,
+      "memory/max_allocated (GiB)": 48.93,
+      "step": 109,
+      "tokens_per_second_per_gpu": 382.78
+    },
+    {
+      "epoch": 0.25229357798165136,
+      "grad_norm": 0.030473977327346802,
+      "learning_rate": 8.930974389548023e-05,
+      "loss": 0.0761,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.93,
+      "memory/max_allocated (GiB)": 48.93,
+      "step": 110,
+      "tokens_per_second_per_gpu": 476.56
+    },
+    {
+      "epoch": 0.2545871559633027,
+      "grad_norm": 0.02930077351629734,
+      "learning_rate": 8.90747116884204e-05,
+      "loss": 0.0691,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.89,
+      "memory/max_allocated (GiB)": 48.89,
+      "step": 111,
+      "tokens_per_second_per_gpu": 441.2
+    },
+    {
+      "epoch": 0.25688073394495414,
+      "grad_norm": 0.02884151227772236,
+      "learning_rate": 8.883744025880428e-05,
+      "loss": 0.0806,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 49.0,
+      "memory/max_allocated (GiB)": 49.0,
+      "step": 112,
+      "tokens_per_second_per_gpu": 406.96
+    },
+    {
+      "epoch": 0.2591743119266055,
+      "grad_norm": 0.02618175558745861,
+      "learning_rate": 8.859794320375168e-05,
+      "loss": 0.0677,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.89,
+      "memory/max_allocated (GiB)": 48.89,
+      "step": 113,
+      "tokens_per_second_per_gpu": 430.04
+    },
+    {
+      "epoch": 0.26146788990825687,
+      "grad_norm": 0.026963548734784126,
+      "learning_rate": 8.835623424792452e-05,
+      "loss": 0.0694,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.89,
+      "memory/max_allocated (GiB)": 48.89,
+      "step": 114,
+      "tokens_per_second_per_gpu": 351.9
+    },
+    {
+      "epoch": 0.26376146788990823,
+      "grad_norm": 0.021544624119997025,
+      "learning_rate": 8.811232724274035e-05,
+      "loss": 0.0613,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 115,
+      "tokens_per_second_per_gpu": 480.22
+    },
+    {
+      "epoch": 0.26605504587155965,
+      "grad_norm": 0.03840009495615959,
+      "learning_rate": 8.786623616557847e-05,
+      "loss": 0.0723,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 49.0,
+      "memory/max_allocated (GiB)": 49.0,
+      "step": 116,
+      "tokens_per_second_per_gpu": 433.18
+    },
+    {
+      "epoch": 0.268348623853211,
+      "grad_norm": 0.022571468725800514,
+      "learning_rate": 8.761797511897906e-05,
+      "loss": 0.065,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 49.0,
+      "memory/max_allocated (GiB)": 49.0,
+      "step": 117,
+      "tokens_per_second_per_gpu": 421.92
+    },
+    {
+      "epoch": 0.2706422018348624,
+      "grad_norm": 0.02688576467335224,
+      "learning_rate": 8.736755832983497e-05,
+      "loss": 0.0772,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.81,
+      "memory/max_allocated (GiB)": 48.81,
+      "step": 118,
+      "tokens_per_second_per_gpu": 354.3
+    },
+    {
+      "epoch": 0.27293577981651373,
+      "grad_norm": 0.025858785957098007,
+      "learning_rate": 8.711500014857634e-05,
+      "loss": 0.0745,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.81,
+      "memory/max_allocated (GiB)": 48.81,
+      "step": 119,
+      "tokens_per_second_per_gpu": 365.46
+    },
+    {
+      "epoch": 0.27522935779816515,
+      "grad_norm": 0.02718079835176468,
+      "learning_rate": 8.686031504834843e-05,
+      "loss": 0.0759,
+      "memory/device_reserved (GiB)": 50.97,
+      "memory/max_active (GiB)": 48.97,
+      "memory/max_allocated (GiB)": 48.97,
+      "step": 120,
+      "tokens_per_second_per_gpu": 426.06
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 436,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 60,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.4689538053609882e+18,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5c707753c7c6de02188b65d6b093332951c4a62dfe324581b6feeb5efb9b2a64
+size 9233

zero_to_fp32.py

@@ -0,0 +1,760 @@
+#!/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 . output_dir/
+#   or
+#   python zero_to_fp32.py . output_dir/ --safe_serialization
+
+import argparse
+import torch
+import glob
+import math
+import os
+import re
+import gc
+import json
+import numpy as np
+from tqdm import tqdm
+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, weights_only=False)
+
+        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 tqdm(files, desc='Loading checkpoint shards'):
+        state_dict = torch.load(f, map_location=device, mmap=True, weights_only=False)
+        # 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 ZERO_STAGE not 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}")
+
+    fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
+    return zero_stage, world_size, fp32_flat_groups
+
+
+def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
+    """
+    Returns fp32 state_dict reconstructed from ds checkpoint
+
+    Args:
+        - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
+
+    """
+    print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
+
+    optim_files = get_optim_files(ds_checkpoint_dir)
+    zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
+    print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
+
+    model_files = get_model_state_files(ds_checkpoint_dir)
+
+    zero_model_states = parse_model_states(model_files)
+    print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
+
+    if zero_stage <= 2:
+        return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+    elif zero_stage == 3:
+        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+
+
+def _zero2_merge_frozen_params(state_dict, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+    frozen_param_fragments = zero_model_states[0].frozen_param_fragments
+
+    if debug:
+        num_elem = sum(s.numel() for s in frozen_param_shapes.values())
+        print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        state_dict[name] = frozen_param_fragments[name]
+
+        if debug:
+            print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _has_callable(obj, fn):
+    attr = getattr(obj, fn, None)
+    return callable(attr)
+
+
+def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+
+    # Reconstruction protocol:
+    #
+    # XXX: document this
+
+    if debug:
+        for i in range(world_size):
+            for j in range(len(fp32_flat_groups[0])):
+                print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
+
+    # XXX: memory usage doubles here (zero2)
+    num_param_groups = len(fp32_flat_groups[0])
+    merged_single_partition_of_fp32_groups = []
+    for i in range(num_param_groups):
+        merged_partitions = [sd[i] for sd in fp32_flat_groups]
+        full_single_fp32_vector = torch.cat(merged_partitions, 0)
+        merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
+    avail_numel = sum(
+        [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
+
+    if debug:
+        wanted_params = sum([len(shapes) for shapes in param_shapes])
+        wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
+        # not asserting if there is a mismatch due to possible padding
+        print(f"Have {avail_numel} numels to process.")
+        print(f"Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    total_numel = 0
+    total_params = 0
+    for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
+        offset = 0
+        avail_numel = full_single_fp32_vector.numel()
+        for name, shape in shapes.items():
+
+            unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
+            total_numel += unpartitioned_numel
+            total_params += 1
+
+            if debug:
+                print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+            state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
+            offset += unpartitioned_numel
+
+        # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
+        # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
+        # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
+        # live optimizer object, so we are checking that the numbers are within the right range
+        align_to = 2 * world_size
+
+        def zero2_align(x):
+            return align_to * math.ceil(x / align_to)
+
+        if debug:
+            print(f"original offset={offset}, avail_numel={avail_numel}")
+
+        offset = zero2_align(offset)
+        avail_numel = zero2_align(avail_numel)
+
+        if debug:
+            print(f"aligned  offset={offset}, avail_numel={avail_numel}")
+
+        # Sanity check
+        if offset != avail_numel:
+            raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero2_merge_frozen_params(state_dict, zero_model_states)
+
+    _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def zero3_partitioned_param_info(unpartitioned_numel, world_size):
+    remainder = unpartitioned_numel % world_size
+    padding_numel = (world_size - remainder) if remainder else 0
+    partitioned_numel = math.ceil(unpartitioned_numel / world_size)
+    return partitioned_numel, padding_numel
+
+
+def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    if debug:
+        for i in range(world_size):
+            num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
+            print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in zero_model_states[0].frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
+        state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+class GatheredTensor:
+    """
+    A pseudo tensor that collects partitioned weights.
+    It is more memory efficient when there are multiple groups.
+    """
+
+    def __init__(self, flat_groups, flat_groups_offset, offset, partitioned_numel, shape):
+        self.flat_groups = flat_groups
+        self.flat_groups_offset = flat_groups_offset
+        self.offset = offset
+        self.partitioned_numel = partitioned_numel
+        self.shape = shape
+        self.dtype = self.flat_groups[0][0].dtype
+
+    def contiguous(self):
+        """
+        Merge partitioned weights from flat_groups into a single tensor.
+        """
+        end_idx = self.offset + self.partitioned_numel
+        world_size = len(self.flat_groups)
+        pad_flat_param_chunks = []
+
+        for rank_i in range(world_size):
+            # for each rank, we need to collect weights from related group/groups
+            flat_groups_at_rank_i = self.flat_groups[rank_i]
+            start_group_id = None
+            end_group_id = None
+            for group_id in range(len(self.flat_groups_offset)):
+                if self.flat_groups_offset[group_id] <= self.offset < self.flat_groups_offset[group_id + 1]:
+                    start_group_id = group_id
+                if self.flat_groups_offset[group_id] < end_idx <= self.flat_groups_offset[group_id + 1]:
+                    end_group_id = group_id
+                    break
+            # collect weights from related group/groups
+            for group_id in range(start_group_id, end_group_id + 1):
+                flat_tensor = flat_groups_at_rank_i[group_id]
+                start_offset = self.offset - self.flat_groups_offset[group_id]
+                end_offset = min(end_idx, self.flat_groups_offset[group_id + 1]) - self.flat_groups_offset[group_id]
+                pad_flat_param_chunks.append(flat_tensor[start_offset:end_offset])
+
+        # collect weights from all ranks
+        pad_flat_param = torch.cat(pad_flat_param_chunks, dim=0)
+        param = pad_flat_param[:self.shape.numel()].view(self.shape).contiguous()
+        return param
+
+
+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 = sum([flat_group.numel() for flat_group in fp32_flat_groups[0]]) * 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
+    flat_groups_offset = [0] + list(np.cumsum([flat_tensor.numel() for flat_tensor in fp32_flat_groups[0]]))
+    for name, shape in tqdm(param_shapes.items(), desc='Gathering sharded weights'):
+        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}"
+            )
+
+        # memory efficient tensor
+        tensor = GatheredTensor(fp32_flat_groups, flat_groups_offset, offset, partitioned_numel, shape)
+        state_dict[name] = tensor
+        offset += partitioned_numel
+
+    offset *= world_size
+
+    # Sanity check
+    if offset != avail_numel:
+        raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
+
+    _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def to_torch_tensor(state_dict, return_empty_tensor=False):
+    """
+    Convert state_dict of GatheredTensor to torch tensor
+    """
+    torch_state_dict = {}
+    converted_tensors = {}
+    for name, tensor in state_dict.items():
+        tensor_id = id(tensor)
+        if tensor_id in converted_tensors:  # shared tensors
+            shared_tensor = torch_state_dict[converted_tensors[tensor_id]]
+            torch_state_dict[name] = shared_tensor
+        else:
+            converted_tensors[tensor_id] = name
+            if return_empty_tensor:
+                torch_state_dict[name] = torch.empty(tensor.shape, dtype=tensor.dtype)
+            else:
+                torch_state_dict[name] = tensor.contiguous()
+    return torch_state_dict
+
+
+def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
+                                             tag=None,
+                                             exclude_frozen_parameters=False,
+                                             lazy_mode=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
+    ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
+    via a model hub.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+        - ``lazy_mode``: get state_dict in lazy mode. It returns a dict of pesduo tensor instead of torch tensor, which is more memory efficient.
+          Convert the pesduo tensor to torch tensor by ``.contiguous()``
+
+    Returns:
+        - pytorch ``state_dict``
+
+    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.
+
+    Note: the above usage may not work if your application doesn't have sufficient free CPU memory.
+    You may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
+    the checkpoint. Or you can load state_dict in lazy mode ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, lazy_mode=True) # not on cpu
+        for name, lazy_tensor in state_dict.item():
+            tensor = lazy_tensor.contiguous()  # to cpu
+            print(name, tensor)
+            # del tensor to release memory if it no longer in use
+    """
+    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")
+
+    state_dict = _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
+    if lazy_mode:
+        return state_dict
+    else:
+        return to_torch_tensor(state_dict)
+
+
+def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir,
+                                               output_dir,
+                                               max_shard_size="5GB",
+                                               safe_serialization=False,
+                                               tag=None,
+                                               exclude_frozen_parameters=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
+    loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``output_dir``: directory to the pytorch fp32 state_dict output files
+        - ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB
+        - ``safe_serialization``:  whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+    """
+
+    # Dependency pre-check
+    if safe_serialization:
+        try:
+            from safetensors.torch import save_file
+        except ImportError:
+            print('If you want to use `safe_serialization`, please `pip install safetensors`')
+            raise
+    if max_shard_size is not None:
+        try:
+            from huggingface_hub import split_torch_state_dict_into_shards
+        except ImportError:
+            print('If you want to use `max_shard_size`, please `pip install huggingface_hub`')
+            raise
+
+    # Convert zero checkpoint to state_dict
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
+                                                          tag,
+                                                          exclude_frozen_parameters,
+                                                          lazy_mode=True)
+
+    # Shard the model if it is too big.
+    weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin"
+    if max_shard_size is not None:
+        filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors")
+        # an memory-efficient approach for sharding
+        empty_state_dict = to_torch_tensor(state_dict, return_empty_tensor=True)
+        state_dict_split = split_torch_state_dict_into_shards(empty_state_dict,
+                                                              filename_pattern=filename_pattern,
+                                                              max_shard_size=max_shard_size)
+    else:
+        from collections import namedtuple
+        StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"])
+        state_dict_split = StateDictSplit(is_sharded=False,
+                                          filename_to_tensors={weights_name: list(state_dict.keys())})
+
+    # Save the model by shard
+    os.makedirs(output_dir, exist_ok=True)
+    filename_to_tensors = state_dict_split.filename_to_tensors.items()
+    for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"):
+        shard_state_dict = {tensor_name: state_dict[tensor_name] for tensor_name in tensors}
+        shard_state_dict = to_torch_tensor(shard_state_dict)
+        output_path = os.path.join(output_dir, shard_file)
+        if safe_serialization:
+            save_file(shard_state_dict, output_path, metadata={"format": "pt"})
+        else:
+            torch.save(shard_state_dict, output_path)
+        # release the memory of current shard
+        for tensor_name in list(shard_state_dict.keys()):
+            del state_dict[tensor_name]
+            del shard_state_dict[tensor_name]
+        del shard_state_dict
+        gc.collect()
+
+    # Save index if sharded
+    if state_dict_split.is_sharded:
+        index = {
+            "metadata": state_dict_split.metadata,
+            "weight_map": state_dict_split.tensor_to_filename,
+        }
+        save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json"
+        save_index_file = os.path.join(output_dir, save_index_file)
+        with open(save_index_file, "w", encoding="utf-8") as f:
+            content = json.dumps(index, indent=2, sort_keys=True) + "\n"
+            f.write(content)
+
+
+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("Extracting fp32 weights")
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+
+    logger.info("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_dir",
+                        type=str,
+                        help="directory to the pytorch fp32 state_dict output files"
+                        "(e.g. path/checkpoint-12-output/)")
+    parser.add_argument(
+        "--max_shard_size",
+        type=str,
+        default="5GB",
+        help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size"
+        "lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`"
+        "We default it to 5GB in order for models to be able to run easily on free-tier google colab instances"
+        "without CPU OOM issues.")
+    parser.add_argument(
+        "--safe_serialization",
+        default=False,
+        action='store_true',
+        help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).")
+    parser.add_argument("-t",
+                        "--tag",
+                        type=str,
+                        default=None,
+                        help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
+    parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
+    parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
+    args = parser.parse_args()
+
+    debug = args.debug
+
+    convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
+                                               args.output_dir,
+                                               max_shard_size=args.max_shard_size,
+                                               safe_serialization=args.safe_serialization,
+                                               tag=args.tag,
+                                               exclude_frozen_parameters=args.exclude_frozen_parameters)