Training in progress, step 500, checkpoint

checkpoint-500/README.md

@@ -0,0 +1,202 @@
+---
+base_model: meta-llama/CodeLlama-7b-Instruct-hf
+library_name: peft
+---
+
+# 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 0.14.0

checkpoint-500/adapter_config.json

@@ -0,0 +1,37 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "meta-llama/CodeLlama-7b-Instruct-hf",
+  "bias": "none",
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 16,
+  "lora_bias": false,
+  "lora_dropout": 0.0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "up_proj",
+    "gate_proj",
+    "k_proj",
+    "q_proj",
+    "o_proj",
+    "v_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}

checkpoint-500/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4de7bd14acccd57374f2a3b4d98a2018109523d7d21790a45c40a54730383744
+size 40036488

checkpoint-500/global_step500/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:53d5d908c773b69aa4b5c4d74a3f80266c774cd51bfbdbd9b592ccc825b01a32
+size 29992112

checkpoint-500/global_step500/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:663f78be436b0edd70686b93c90a27f2a3060d5f00fc9bae2ebad62d0514e883
+size 29992112

checkpoint-500/global_step500/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:f9a727887a44f53b7b4703eb757a5c89f6dec9eaa3cc45a4f6c89c72858ae1ab
+size 29992176

checkpoint-500/global_step500/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:d0e03d72af2c3933c2d3c948bf374df1593fa8504e667ebb29f2e726e4cfcdab
+size 29992176

checkpoint-500/global_step500/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ffc94f8549678ec6c4c520017ae298c571852ecc2f554828d2737f6397213dd7
+size 29992176

checkpoint-500/global_step500/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a90e0ff0fe9074c3dd521e504bf876167af84e602594dfca72e12ab3ec2b80a7
+size 29992176

checkpoint-500/global_step500/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:321df79aebb9ccb1e78406aa2cbeb3da0136a79600c55b27c4e9d1ec898f18ba
+size 29992176

checkpoint-500/global_step500/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bf84aaf700067831c7655dc6be98601b5be8c70a01233027e6e95928aa014783
+size 29992176

checkpoint-500/global_step500/mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0b188cd2042f2e5bf9cf413d8229a61620dc66404ac45e5e6c808067d7b44e39
+size 40324204

checkpoint-500/latest

@@ -0,0 +1 @@
+global_step500

checkpoint-500/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1a2fbcd26bac3ea7dc02fc9ede5b8a1914ca51611473722a11a969e1f26ac0ee
+size 15984

checkpoint-500/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:66d97b511d2fdb8061e5bf72c139923941c148260fac1caedd654028da6986c1
+size 15984

checkpoint-500/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3839473129eb8c438ab312370daa55eb10a0790f33d38fc5eaa24859b54b0d1f
+size 15984

checkpoint-500/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5088a0d34c7015afe60457fbb3f0a4740839369017a42ea4b3250322c2d63ceb
+size 15984

checkpoint-500/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f9cac0eb25286b75549fa2030810940adf357064a83facaf5c58ebe37190b6ac
+size 15984

checkpoint-500/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f0a57d29811122d52bd53f81af680412b91dde1cd2a12fa885d8a54388be8e2d
+size 15984

checkpoint-500/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2c90ab29b255eaf920ecc1cba0b586e426f8e2db67b44a65576693f84178a04f
+size 15984

checkpoint-500/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d4efbfa3cfb1bb8fb9c3380e65959a8b4eaf3bceb0507a26ffba1a3e4636ddb1
+size 15984

checkpoint-500/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cbc0bc6e52c2f77c3fb3fa71b6b0e38d8f04862f18c00efcd0284aba93ab1e11
+size 1064

checkpoint-500/special_tokens_map.json

@@ -0,0 +1,30 @@
+{
+  "additional_special_tokens": [
+    "▁<PRE>",
+    "▁<MID>",
+    "▁<SUF>",
+    "▁<EOT>"
+  ],
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "</s>",
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

checkpoint-500/tokenizer_config.json

@@ -0,0 +1,87 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "32007": {
+      "content": "▁<PRE>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "32008": {
+      "content": "▁<SUF>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "32009": {
+      "content": "▁<MID>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "32010": {
+      "content": "▁<EOT>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "▁<PRE>",
+    "▁<MID>",
+    "▁<SUF>",
+    "▁<EOT>"
+  ],
+  "bos_token": "<s>",
+  "chat_template": "{% if messages[0]['role'] == 'system' %}{% set system_message = messages[0]['content'] %}{% endif %}{% for message in messages %}{% set content = message['content'] %}{% if loop.index0 == 0 and system_message is defined %}{% set content = '<<SYS>>\n' + system_message + '\n<</SYS>>\n\n' + message['content'] %}{% endif %}{% if message['role'] == 'user' %}{{ '<s>' + '[INST] ' + content + ' [/INST]' }}{% elif message['role'] == 'assistant' %}{{ ' ' + content + ' ' + '</s>' }}{% endif %}{% endfor %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "</s>",
+  "eot_token": "▁<EOT>",
+  "extra_special_tokens": {},
+  "fill_token": "<FILL_ME>",
+  "legacy": null,
+  "middle_token": "▁<MID>",
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "</s>",
+  "padding_side": "right",
+  "prefix_token": "▁<PRE>",
+  "sp_model_kwargs": {},
+  "split_special_tokens": false,
+  "suffix_token": "▁<SUF>",
+  "tokenizer_class": "CodeLlamaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}

checkpoint-500/trainer_state.json

@@ -0,0 +1,943 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.4103405826836274,
+  "eval_steps": 50,
+  "global_step": 500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.008206811653672548,
+      "grad_norm": 0.06318386644124985,
+      "learning_rate": 4.999451708687114e-06,
+      "logits/chosen": -2.1367907524108887,
+      "logits/rejected": -2.4948182106018066,
+      "logps/chosen": -0.291498601436615,
+      "logps/rejected": -0.3196522295475006,
+      "loss": 7.5728,
+      "rewards/accuracies": 0.5,
+      "rewards/chosen": -0.4372479021549225,
+      "rewards/margins": 0.04223042353987694,
+      "rewards/rejected": -0.47947829961776733,
+      "step": 10
+    },
+    {
+      "epoch": 0.016413623307345096,
+      "grad_norm": 0.07310314476490021,
+      "learning_rate": 4.997807075247147e-06,
+      "logits/chosen": -2.1456007957458496,
+      "logits/rejected": -2.4455342292785645,
+      "logps/chosen": -0.26213544607162476,
+      "logps/rejected": -0.32332050800323486,
+      "loss": 7.5298,
+      "rewards/accuracies": 0.5625,
+      "rewards/chosen": -0.3932031989097595,
+      "rewards/margins": 0.09177760779857635,
+      "rewards/rejected": -0.4849807620048523,
+      "step": 20
+    },
+    {
+      "epoch": 0.024620434961017644,
+      "grad_norm": 0.05936102196574211,
+      "learning_rate": 4.9950668210706795e-06,
+      "logits/chosen": -2.0765950679779053,
+      "logits/rejected": -2.485799789428711,
+      "logps/chosen": -0.26631081104278564,
+      "logps/rejected": -0.32647624611854553,
+      "loss": 7.5208,
+      "rewards/accuracies": 0.512499988079071,
+      "rewards/chosen": -0.39946624636650085,
+      "rewards/margins": 0.09024813771247864,
+      "rewards/rejected": -0.4897143840789795,
+      "step": 30
+    },
+    {
+      "epoch": 0.03282724661469019,
+      "grad_norm": 0.08499134331941605,
+      "learning_rate": 4.9912321481237616e-06,
+      "logits/chosen": -2.0753884315490723,
+      "logits/rejected": -2.441580295562744,
+      "logps/chosen": -0.2749950885772705,
+      "logps/rejected": -0.30180150270462036,
+      "loss": 7.4229,
+      "rewards/accuracies": 0.5375000238418579,
+      "rewards/chosen": -0.41249266266822815,
+      "rewards/margins": 0.04020959883928299,
+      "rewards/rejected": -0.45270222425460815,
+      "step": 40
+    },
+    {
+      "epoch": 0.04103405826836274,
+      "grad_norm": 0.07681389898061752,
+      "learning_rate": 4.986304738420684e-06,
+      "logits/chosen": -2.145660877227783,
+      "logits/rejected": -2.465946912765503,
+      "logps/chosen": -0.24909739196300507,
+      "logps/rejected": -0.2796121835708618,
+      "loss": 7.4811,
+      "rewards/accuracies": 0.48750001192092896,
+      "rewards/chosen": -0.373646080493927,
+      "rewards/margins": 0.045772187411785126,
+      "rewards/rejected": -0.41941824555397034,
+      "step": 50
+    },
+    {
+      "epoch": 0.04103405826836274,
+      "eval_logits/chosen": -2.012000799179077,
+      "eval_logits/rejected": -2.5381252765655518,
+      "eval_logps/chosen": -0.24157460033893585,
+      "eval_logps/rejected": -0.2957758605480194,
+      "eval_loss": 0.9317650198936462,
+      "eval_rewards/accuracies": 0.5252525210380554,
+      "eval_rewards/chosen": -0.3623619079589844,
+      "eval_rewards/margins": 0.08130191266536713,
+      "eval_rewards/rejected": -0.4436637759208679,
+      "eval_runtime": 26.0809,
+      "eval_samples_per_second": 30.214,
+      "eval_steps_per_second": 3.796,
+      "step": 50
+    },
+    {
+      "epoch": 0.04924086992203529,
+      "grad_norm": 0.06638535112142563,
+      "learning_rate": 4.980286753286196e-06,
+      "logits/chosen": -2.145846128463745,
+      "logits/rejected": -2.4077115058898926,
+      "logps/chosen": -0.22265203297138214,
+      "logps/rejected": -0.30774614214897156,
+      "loss": 7.4605,
+      "rewards/accuracies": 0.5625,
+      "rewards/chosen": -0.3339780271053314,
+      "rewards/margins": 0.1276412308216095,
+      "rewards/rejected": -0.4616192877292633,
+      "step": 60
+    },
+    {
+      "epoch": 0.057447681575707836,
+      "grad_norm": 0.057281140238046646,
+      "learning_rate": 4.973180832407471e-06,
+      "logits/chosen": -2.0021350383758545,
+      "logits/rejected": -2.4299912452697754,
+      "logps/chosen": -0.23488977551460266,
+      "logps/rejected": -0.33270469307899475,
+      "loss": 7.4257,
+      "rewards/accuracies": 0.574999988079071,
+      "rewards/chosen": -0.3523346781730652,
+      "rewards/margins": 0.14672236144542694,
+      "rewards/rejected": -0.4990570545196533,
+      "step": 70
+    },
+    {
+      "epoch": 0.06565449322938038,
+      "grad_norm": 0.07725922018289566,
+      "learning_rate": 4.964990092676263e-06,
+      "logits/chosen": -2.117995023727417,
+      "logits/rejected": -2.359265089035034,
+      "logps/chosen": -0.21598832309246063,
+      "logps/rejected": -0.300583153963089,
+      "loss": 7.4384,
+      "rewards/accuracies": 0.6000000238418579,
+      "rewards/chosen": -0.32398244738578796,
+      "rewards/margins": 0.12689228355884552,
+      "rewards/rejected": -0.4508747458457947,
+      "step": 80
+    },
+    {
+      "epoch": 0.07386130488305294,
+      "grad_norm": 0.0598183274269104,
+      "learning_rate": 4.9557181268217225e-06,
+      "logits/chosen": -2.282627582550049,
+      "logits/rejected": -2.441333532333374,
+      "logps/chosen": -0.23655852675437927,
+      "logps/rejected": -0.3246815800666809,
+      "loss": 7.4584,
+      "rewards/accuracies": 0.512499988079071,
+      "rewards/chosen": -0.3548378050327301,
+      "rewards/margins": 0.13218457996845245,
+      "rewards/rejected": -0.48702239990234375,
+      "step": 90
+    },
+    {
+      "epoch": 0.08206811653672548,
+      "grad_norm": 0.058213479816913605,
+      "learning_rate": 4.9453690018345144e-06,
+      "logits/chosen": -2.1114468574523926,
+      "logits/rejected": -2.5035691261291504,
+      "logps/chosen": -0.23073866963386536,
+      "logps/rejected": -0.29445192217826843,
+      "loss": 7.4116,
+      "rewards/accuracies": 0.5375000238418579,
+      "rewards/chosen": -0.34610801935195923,
+      "rewards/margins": 0.09556989371776581,
+      "rewards/rejected": -0.44167789816856384,
+      "step": 100
+    },
+    {
+      "epoch": 0.08206811653672548,
+      "eval_logits/chosen": -2.0183491706848145,
+      "eval_logits/rejected": -2.5400593280792236,
+      "eval_logps/chosen": -0.20393377542495728,
+      "eval_logps/rejected": -0.2818409502506256,
+      "eval_loss": 0.9129964113235474,
+      "eval_rewards/accuracies": 0.5656565427780151,
+      "eval_rewards/chosen": -0.3059006631374359,
+      "eval_rewards/margins": 0.11686072498559952,
+      "eval_rewards/rejected": -0.4227614104747772,
+      "eval_runtime": 26.0825,
+      "eval_samples_per_second": 30.212,
+      "eval_steps_per_second": 3.796,
+      "step": 100
+    },
+    {
+      "epoch": 0.09027492819039803,
+      "grad_norm": 0.06249881908297539,
+      "learning_rate": 4.933947257182901e-06,
+      "logits/chosen": -2.1324548721313477,
+      "logits/rejected": -2.434319019317627,
+      "logps/chosen": -0.22180762887001038,
+      "logps/rejected": -0.28862181305885315,
+      "loss": 7.3604,
+      "rewards/accuracies": 0.512499988079071,
+      "rewards/chosen": -0.332711398601532,
+      "rewards/margins": 0.10022131353616714,
+      "rewards/rejected": -0.43293270468711853,
+      "step": 110
+    },
+    {
+      "epoch": 0.09848173984407058,
+      "grad_norm": 0.061758093535900116,
+      "learning_rate": 4.921457902821578e-06,
+      "logits/chosen": -2.0597169399261475,
+      "logits/rejected": -2.4386391639709473,
+      "logps/chosen": -0.22720107436180115,
+      "logps/rejected": -0.303659051656723,
+      "loss": 7.3624,
+      "rewards/accuracies": 0.6000000238418579,
+      "rewards/chosen": -0.3408016264438629,
+      "rewards/margins": 0.1146869882941246,
+      "rewards/rejected": -0.4554885923862457,
+      "step": 120
+    },
+    {
+      "epoch": 0.10668855149774313,
+      "grad_norm": 0.08368540555238724,
+      "learning_rate": 4.907906416994146e-06,
+      "logits/chosen": -2.0944437980651855,
+      "logits/rejected": -2.4157254695892334,
+      "logps/chosen": -0.19590887427330017,
+      "logps/rejected": -0.3365771770477295,
+      "loss": 7.3464,
+      "rewards/accuracies": 0.675000011920929,
+      "rewards/chosen": -0.29386329650878906,
+      "rewards/margins": 0.2110024392604828,
+      "rewards/rejected": -0.5048657655715942,
+      "step": 130
+    },
+    {
+      "epoch": 0.11489536315141567,
+      "grad_norm": 0.060954928398132324,
+      "learning_rate": 4.893298743830168e-06,
+      "logits/chosen": -2.1551766395568848,
+      "logits/rejected": -2.5695576667785645,
+      "logps/chosen": -0.19875812530517578,
+      "logps/rejected": -0.2967599928379059,
+      "loss": 7.3179,
+      "rewards/accuracies": 0.574999988079071,
+      "rewards/chosen": -0.29813718795776367,
+      "rewards/margins": 0.14700281620025635,
+      "rewards/rejected": -0.4451400339603424,
+      "step": 140
+    },
+    {
+      "epoch": 0.12310217480508823,
+      "grad_norm": 0.05665091797709465,
+      "learning_rate": 4.8776412907378845e-06,
+      "logits/chosen": -2.1410365104675293,
+      "logits/rejected": -2.4798667430877686,
+      "logps/chosen": -0.19316771626472473,
+      "logps/rejected": -0.2972142696380615,
+      "loss": 7.2384,
+      "rewards/accuracies": 0.574999988079071,
+      "rewards/chosen": -0.2897515296936035,
+      "rewards/margins": 0.15606984496116638,
+      "rewards/rejected": -0.4458213746547699,
+      "step": 150
+    },
+    {
+      "epoch": 0.12310217480508823,
+      "eval_logits/chosen": -2.0654072761535645,
+      "eval_logits/rejected": -2.596571207046509,
+      "eval_logps/chosen": -0.17970335483551025,
+      "eval_logps/rejected": -0.2767573893070221,
+      "eval_loss": 0.8982937335968018,
+      "eval_rewards/accuracies": 0.5858585834503174,
+      "eval_rewards/chosen": -0.269555002450943,
+      "eval_rewards/margins": 0.14558106660842896,
+      "eval_rewards/rejected": -0.41513609886169434,
+      "eval_runtime": 26.0741,
+      "eval_samples_per_second": 30.222,
+      "eval_steps_per_second": 3.797,
+      "step": 150
+    },
+    {
+      "epoch": 0.13130898645876077,
+      "grad_norm": 0.07328196614980698,
+      "learning_rate": 4.860940925593703e-06,
+      "logits/chosen": -2.202148914337158,
+      "logits/rejected": -2.5385117530822754,
+      "logps/chosen": -0.19814102351665497,
+      "logps/rejected": -0.3139093518257141,
+      "loss": 7.1309,
+      "rewards/accuracies": 0.6000000238418579,
+      "rewards/chosen": -0.29721152782440186,
+      "rewards/margins": 0.1736525148153305,
+      "rewards/rejected": -0.4708639979362488,
+      "step": 160
+    },
+    {
+      "epoch": 0.1395157981124333,
+      "grad_norm": 0.09789691120386124,
+      "learning_rate": 4.84320497372973e-06,
+      "logits/chosen": -2.153545618057251,
+      "logits/rejected": -2.532336950302124,
+      "logps/chosen": -0.1861150860786438,
+      "logps/rejected": -0.2787100672721863,
+      "loss": 7.2498,
+      "rewards/accuracies": 0.5874999761581421,
+      "rewards/chosen": -0.2791725993156433,
+      "rewards/margins": 0.13889247179031372,
+      "rewards/rejected": -0.4180651605129242,
+      "step": 170
+    },
+    {
+      "epoch": 0.14772260976610588,
+      "grad_norm": 0.0829203873872757,
+      "learning_rate": 4.824441214720629e-06,
+      "logits/chosen": -2.254868268966675,
+      "logits/rejected": -2.5931999683380127,
+      "logps/chosen": -0.19771653413772583,
+      "logps/rejected": -0.2859548032283783,
+      "loss": 7.1449,
+      "rewards/accuracies": 0.5625,
+      "rewards/chosen": -0.29657480120658875,
+      "rewards/margins": 0.1323573738336563,
+      "rewards/rejected": -0.42893218994140625,
+      "step": 180
+    },
+    {
+      "epoch": 0.15592942141977842,
+      "grad_norm": 0.10499900579452515,
+      "learning_rate": 4.804657878971252e-06,
+      "logits/chosen": -2.2066543102264404,
+      "logits/rejected": -2.594515323638916,
+      "logps/chosen": -0.19944152235984802,
+      "logps/rejected": -0.26530706882476807,
+      "loss": 7.1717,
+      "rewards/accuracies": 0.5249999761581421,
+      "rewards/chosen": -0.29916223883628845,
+      "rewards/margins": 0.09879834204912186,
+      "rewards/rejected": -0.3979606032371521,
+      "step": 190
+    },
+    {
+      "epoch": 0.16413623307345096,
+      "grad_norm": 0.10522742569446564,
+      "learning_rate": 4.783863644106502e-06,
+      "logits/chosen": -2.282865047454834,
+      "logits/rejected": -2.651233196258545,
+      "logps/chosen": -0.1877971738576889,
+      "logps/rejected": -0.28373947739601135,
+      "loss": 7.1001,
+      "rewards/accuracies": 0.550000011920929,
+      "rewards/chosen": -0.28169578313827515,
+      "rewards/margins": 0.14391344785690308,
+      "rewards/rejected": -0.4256092607975006,
+      "step": 200
+    },
+    {
+      "epoch": 0.16413623307345096,
+      "eval_logits/chosen": -2.1934518814086914,
+      "eval_logits/rejected": -2.772143840789795,
+      "eval_logps/chosen": -0.1790025532245636,
+      "eval_logps/rejected": -0.3032245934009552,
+      "eval_loss": 0.8807509541511536,
+      "eval_rewards/accuracies": 0.5858585834503174,
+      "eval_rewards/chosen": -0.2685038149356842,
+      "eval_rewards/margins": 0.1863330751657486,
+      "eval_rewards/rejected": -0.454836905002594,
+      "eval_runtime": 26.0786,
+      "eval_samples_per_second": 30.216,
+      "eval_steps_per_second": 3.796,
+      "step": 200
+    },
+    {
+      "epoch": 0.1723430447271235,
+      "grad_norm": 0.13399213552474976,
+      "learning_rate": 4.762067631165049e-06,
+      "logits/chosen": -2.3500285148620605,
+      "logits/rejected": -2.6827149391174316,
+      "logps/chosen": -0.17780962586402893,
+      "logps/rejected": -0.2749634087085724,
+      "loss": 7.0722,
+      "rewards/accuracies": 0.5375000238418579,
+      "rewards/chosen": -0.266714483499527,
+      "rewards/margins": 0.145730659365654,
+      "rewards/rejected": -0.41244515776634216,
+      "step": 210
+    },
+    {
+      "epoch": 0.18054985638079607,
+      "grad_norm": 0.1727023422718048,
+      "learning_rate": 4.7392794005985324e-06,
+      "logits/chosen": -2.356797695159912,
+      "logits/rejected": -2.7276604175567627,
+      "logps/chosen": -0.20826168358325958,
+      "logps/rejected": -0.3760753273963928,
+      "loss": 7.0205,
+      "rewards/accuracies": 0.6499999761581421,
+      "rewards/chosen": -0.31239253282546997,
+      "rewards/margins": 0.2517204284667969,
+      "rewards/rejected": -0.5641129016876221,
+      "step": 220
+    },
+    {
+      "epoch": 0.1887566680344686,
+      "grad_norm": 0.14360135793685913,
+      "learning_rate": 4.715508948078037e-06,
+      "logits/chosen": -2.3526904582977295,
+      "logits/rejected": -2.785292148590088,
+      "logps/chosen": -0.22017621994018555,
+      "logps/rejected": -0.35291892290115356,
+      "loss": 6.9773,
+      "rewards/accuracies": 0.612500011920929,
+      "rewards/chosen": -0.3302643299102783,
+      "rewards/margins": 0.1991141140460968,
+      "rewards/rejected": -0.5293784737586975,
+      "step": 230
+    },
+    {
+      "epoch": 0.19696347968814115,
+      "grad_norm": 0.18886974453926086,
+      "learning_rate": 4.690766700109659e-06,
+      "logits/chosen": -2.3618314266204834,
+      "logits/rejected": -2.850059747695923,
+      "logps/chosen": -0.22531962394714355,
+      "logps/rejected": -0.38331112265586853,
+      "loss": 6.9879,
+      "rewards/accuracies": 0.637499988079071,
+      "rewards/chosen": -0.3379794657230377,
+      "rewards/margins": 0.23698726296424866,
+      "rewards/rejected": -0.5749667286872864,
+      "step": 240
+    },
+    {
+      "epoch": 0.2051702913418137,
+      "grad_norm": 0.2599099278450012,
+      "learning_rate": 4.665063509461098e-06,
+      "logits/chosen": -2.4776885509490967,
+      "logits/rejected": -2.8583390712738037,
+      "logps/chosen": -0.2538486123085022,
+      "logps/rejected": -0.42415136098861694,
+      "loss": 6.7357,
+      "rewards/accuracies": 0.6000000238418579,
+      "rewards/chosen": -0.3807729184627533,
+      "rewards/margins": 0.2554541230201721,
+      "rewards/rejected": -0.636227011680603,
+      "step": 250
+    },
+    {
+      "epoch": 0.2051702913418137,
+      "eval_logits/chosen": -2.3788387775421143,
+      "eval_logits/rejected": -2.958366632461548,
+      "eval_logps/chosen": -0.2162775695323944,
+      "eval_logps/rejected": -0.41066110134124756,
+      "eval_loss": 0.8405817747116089,
+      "eval_rewards/accuracies": 0.6060606241226196,
+      "eval_rewards/chosen": -0.3244163393974304,
+      "eval_rewards/margins": 0.2915753722190857,
+      "eval_rewards/rejected": -0.6159917116165161,
+      "eval_runtime": 26.0752,
+      "eval_samples_per_second": 30.22,
+      "eval_steps_per_second": 3.797,
+      "step": 250
+    },
+    {
+      "epoch": 0.21337710299548626,
+      "grad_norm": 0.25840890407562256,
+      "learning_rate": 4.638410650401267e-06,
+      "logits/chosen": -2.5892271995544434,
+      "logits/rejected": -2.819650888442993,
+      "logps/chosen": -0.2257525473833084,
+      "logps/rejected": -0.46280306577682495,
+      "loss": 6.6048,
+      "rewards/accuracies": 0.574999988079071,
+      "rewards/chosen": -0.3386288285255432,
+      "rewards/margins": 0.3555757403373718,
+      "rewards/rejected": -0.6942045092582703,
+      "step": 260
+    },
+    {
+      "epoch": 0.2215839146491588,
+      "grad_norm": 0.2838613986968994,
+      "learning_rate": 4.610819813755038e-06,
+      "logits/chosen": -2.536020040512085,
+      "logits/rejected": -2.843383312225342,
+      "logps/chosen": -0.2739175856113434,
+      "logps/rejected": -0.5088076591491699,
+      "loss": 6.6403,
+      "rewards/accuracies": 0.6000000238418579,
+      "rewards/chosen": -0.41087642312049866,
+      "rewards/margins": 0.3523350656032562,
+      "rewards/rejected": -0.7632113695144653,
+      "step": 270
+    },
+    {
+      "epoch": 0.22979072630283134,
+      "grad_norm": 0.3575810194015503,
+      "learning_rate": 4.582303101775249e-06,
+      "logits/chosen": -2.4525866508483887,
+      "logits/rejected": -2.7756829261779785,
+      "logps/chosen": -0.27604570984840393,
+      "logps/rejected": -0.6104786992073059,
+      "loss": 6.4767,
+      "rewards/accuracies": 0.6875,
+      "rewards/chosen": -0.4140685498714447,
+      "rewards/margins": 0.5016494989395142,
+      "rewards/rejected": -0.9157179594039917,
+      "step": 280
+    },
+    {
+      "epoch": 0.23799753795650389,
+      "grad_norm": 0.36226338148117065,
+      "learning_rate": 4.55287302283426e-06,
+      "logits/chosen": -2.4639816284179688,
+      "logits/rejected": -2.865053415298462,
+      "logps/chosen": -0.35306140780448914,
+      "logps/rejected": -0.5840066075325012,
+      "loss": 6.2071,
+      "rewards/accuracies": 0.4749999940395355,
+      "rewards/chosen": -0.5295921564102173,
+      "rewards/margins": 0.3464178144931793,
+      "rewards/rejected": -0.8760099411010742,
+      "step": 290
+    },
+    {
+      "epoch": 0.24620434961017645,
+      "grad_norm": 0.38896313309669495,
+      "learning_rate": 4.522542485937369e-06,
+      "logits/chosen": -2.6630337238311768,
+      "logits/rejected": -2.7479195594787598,
+      "logps/chosen": -0.3706950545310974,
+      "logps/rejected": -0.7957242131233215,
+      "loss": 6.1801,
+      "rewards/accuracies": 0.5874999761581421,
+      "rewards/chosen": -0.5560425519943237,
+      "rewards/margins": 0.6375436782836914,
+      "rewards/rejected": -1.1935861110687256,
+      "step": 300
+    },
+    {
+      "epoch": 0.24620434961017645,
+      "eval_logits/chosen": -2.411334991455078,
+      "eval_logits/rejected": -2.820974588394165,
+      "eval_logps/chosen": -0.3725183308124542,
+      "eval_logps/rejected": -0.8138000965118408,
+      "eval_loss": 0.738965630531311,
+      "eval_rewards/accuracies": 0.6060606241226196,
+      "eval_rewards/chosen": -0.5587774515151978,
+      "eval_rewards/margins": 0.6619227528572083,
+      "eval_rewards/rejected": -1.2207001447677612,
+      "eval_runtime": 26.0764,
+      "eval_samples_per_second": 30.219,
+      "eval_steps_per_second": 3.797,
+      "step": 300
+    },
+    {
+      "epoch": 0.254411161263849,
+      "grad_norm": 0.5848517417907715,
+      "learning_rate": 4.491324795060491e-06,
+      "logits/chosen": -2.5622355937957764,
+      "logits/rejected": -2.7415950298309326,
+      "logps/chosen": -0.38525494933128357,
+      "logps/rejected": -0.8741232752799988,
+      "loss": 5.98,
+      "rewards/accuracies": 0.5375000238418579,
+      "rewards/chosen": -0.5778824687004089,
+      "rewards/margins": 0.7333025336265564,
+      "rewards/rejected": -1.3111850023269653,
+      "step": 310
+    },
+    {
+      "epoch": 0.26261797291752154,
+      "grad_norm": 0.38972222805023193,
+      "learning_rate": 4.4592336433146e-06,
+      "logits/chosen": -2.6293787956237793,
+      "logits/rejected": -2.7180721759796143,
+      "logps/chosen": -0.5181100964546204,
+      "logps/rejected": -0.97294682264328,
+      "loss": 5.608,
+      "rewards/accuracies": 0.612500011920929,
+      "rewards/chosen": -0.7771651744842529,
+      "rewards/margins": 0.6822551488876343,
+      "rewards/rejected": -1.4594202041625977,
+      "step": 320
+    },
+    {
+      "epoch": 0.2708247845711941,
+      "grad_norm": 0.5381959080696106,
+      "learning_rate": 4.426283106939474e-06,
+      "logits/chosen": -2.477749824523926,
+      "logits/rejected": -2.7682888507843018,
+      "logps/chosen": -0.47721824049949646,
+      "logps/rejected": -1.0577385425567627,
+      "loss": 5.5189,
+      "rewards/accuracies": 0.6000000238418579,
+      "rewards/chosen": -0.7158273458480835,
+      "rewards/margins": 0.8707805871963501,
+      "rewards/rejected": -1.5866079330444336,
+      "step": 330
+    },
+    {
+      "epoch": 0.2790315962248666,
+      "grad_norm": 0.5332415699958801,
+      "learning_rate": 4.3924876391293915e-06,
+      "logits/chosen": -2.5470075607299805,
+      "logits/rejected": -2.8264012336730957,
+      "logps/chosen": -0.5053269267082214,
+      "logps/rejected": -1.412097454071045,
+      "loss": 5.3992,
+      "rewards/accuracies": 0.675000011920929,
+      "rewards/chosen": -0.7579904198646545,
+      "rewards/margins": 1.360155701637268,
+      "rewards/rejected": -2.1181461811065674,
+      "step": 340
+    },
+    {
+      "epoch": 0.2872384078785392,
+      "grad_norm": 0.5841536521911621,
+      "learning_rate": 4.357862063693486e-06,
+      "logits/chosen": -2.5552780628204346,
+      "logits/rejected": -2.7644314765930176,
+      "logps/chosen": -0.6264504790306091,
+      "logps/rejected": -1.4451416730880737,
+      "loss": 5.0093,
+      "rewards/accuracies": 0.699999988079071,
+      "rewards/chosen": -0.9396758079528809,
+      "rewards/margins": 1.228036880493164,
+      "rewards/rejected": -2.167712450027466,
+      "step": 350
+    },
+    {
+      "epoch": 0.2872384078785392,
+      "eval_logits/chosen": -2.486525774002075,
+      "eval_logits/rejected": -2.809356451034546,
+      "eval_logps/chosen": -0.6259626746177673,
+      "eval_logps/rejected": -1.6826657056808472,
+      "eval_loss": 0.610858678817749,
+      "eval_rewards/accuracies": 0.6464646458625793,
+      "eval_rewards/chosen": -0.9389441013336182,
+      "eval_rewards/margins": 1.585054636001587,
+      "eval_rewards/rejected": -2.523998737335205,
+      "eval_runtime": 26.0792,
+      "eval_samples_per_second": 30.216,
+      "eval_steps_per_second": 3.796,
+      "step": 350
+    },
+    {
+      "epoch": 0.29544521953221176,
+      "grad_norm": 0.6259649395942688,
+      "learning_rate": 4.322421568553529e-06,
+      "logits/chosen": -2.5630745887756348,
+      "logits/rejected": -2.80169939994812,
+      "logps/chosen": -0.6148477792739868,
+      "logps/rejected": -1.7640241384506226,
+      "loss": 5.0832,
+      "rewards/accuracies": 0.6499999761581421,
+      "rewards/chosen": -0.9222715497016907,
+      "rewards/margins": 1.7237647771835327,
+      "rewards/rejected": -2.646036148071289,
+      "step": 360
+    },
+    {
+      "epoch": 0.3036520311858843,
+      "grad_norm": 0.5134413838386536,
+      "learning_rate": 4.286181699082008e-06,
+      "logits/chosen": -2.5688040256500244,
+      "logits/rejected": -2.823493242263794,
+      "logps/chosen": -0.7328687906265259,
+      "logps/rejected": -2.135953187942505,
+      "loss": 4.484,
+      "rewards/accuracies": 0.6875,
+      "rewards/chosen": -1.0993033647537231,
+      "rewards/margins": 2.104626178741455,
+      "rewards/rejected": -3.203929901123047,
+      "step": 370
+    },
+    {
+      "epoch": 0.31185884283955684,
+      "grad_norm": 0.5029065608978271,
+      "learning_rate": 4.249158351283414e-06,
+      "logits/chosen": -2.507948160171509,
+      "logits/rejected": -2.797893524169922,
+      "logps/chosen": -0.8517419695854187,
+      "logps/rejected": -2.6004090309143066,
+      "loss": 4.3033,
+      "rewards/accuracies": 0.675000011920929,
+      "rewards/chosen": -1.2776129245758057,
+      "rewards/margins": 2.6230006217956543,
+      "rewards/rejected": -3.900613307952881,
+      "step": 380
+    },
+    {
+      "epoch": 0.3200656544932294,
+      "grad_norm": 0.6171831488609314,
+      "learning_rate": 4.211367764821722e-06,
+      "logits/chosen": -2.709599018096924,
+      "logits/rejected": -2.8980605602264404,
+      "logps/chosen": -0.9357224702835083,
+      "logps/rejected": -2.464841842651367,
+      "loss": 4.6813,
+      "rewards/accuracies": 0.5874999761581421,
+      "rewards/chosen": -1.4035838842391968,
+      "rewards/margins": 2.2936789989471436,
+      "rewards/rejected": -3.6972625255584717,
+      "step": 390
+    },
+    {
+      "epoch": 0.3282724661469019,
+      "grad_norm": 1.2649667263031006,
+      "learning_rate": 4.172826515897146e-06,
+      "logits/chosen": -2.3837532997131348,
+      "logits/rejected": -2.6686861515045166,
+      "logps/chosen": -0.9314821362495422,
+      "logps/rejected": -2.562440872192383,
+      "loss": 4.3764,
+      "rewards/accuracies": 0.737500011920929,
+      "rewards/chosen": -1.3972232341766357,
+      "rewards/margins": 2.4464378356933594,
+      "rewards/rejected": -3.843661069869995,
+      "step": 400
+    },
+    {
+      "epoch": 0.3282724661469019,
+      "eval_logits/chosen": -2.3613698482513428,
+      "eval_logits/rejected": -2.7536535263061523,
+      "eval_logps/chosen": -0.8348632454872131,
+      "eval_logps/rejected": -2.6590662002563477,
+      "eval_loss": 0.5019229650497437,
+      "eval_rewards/accuracies": 0.6767676472663879,
+      "eval_rewards/chosen": -1.2522947788238525,
+      "eval_rewards/margins": 2.736304759979248,
+      "eval_rewards/rejected": -3.9885993003845215,
+      "eval_runtime": 26.0816,
+      "eval_samples_per_second": 30.213,
+      "eval_steps_per_second": 3.796,
+      "step": 400
+    },
+    {
+      "epoch": 0.33647927780057446,
+      "grad_norm": 0.509773313999176,
+      "learning_rate": 4.133551509975264e-06,
+      "logits/chosen": -2.4433794021606445,
+      "logits/rejected": -2.732313871383667,
+      "logps/chosen": -1.0649070739746094,
+      "logps/rejected": -3.0055129528045654,
+      "loss": 4.0575,
+      "rewards/accuracies": 0.7250000238418579,
+      "rewards/chosen": -1.597360372543335,
+      "rewards/margins": 2.9109084606170654,
+      "rewards/rejected": -4.508269309997559,
+      "step": 410
+    },
+    {
+      "epoch": 0.344686089454247,
+      "grad_norm": 1.3467975854873657,
+      "learning_rate": 4.093559974371725e-06,
+      "logits/chosen": -2.296419858932495,
+      "logits/rejected": -2.650160551071167,
+      "logps/chosen": -1.1074317693710327,
+      "logps/rejected": -3.8539538383483887,
+      "loss": 3.9082,
+      "rewards/accuracies": 0.762499988079071,
+      "rewards/chosen": -1.6611478328704834,
+      "rewards/margins": 4.119783401489258,
+      "rewards/rejected": -5.780930519104004,
+      "step": 420
+    },
+    {
+      "epoch": 0.3528929011079196,
+      "grad_norm": 0.576802134513855,
+      "learning_rate": 4.052869450695776e-06,
+      "logits/chosen": -2.4122748374938965,
+      "logits/rejected": -2.657745122909546,
+      "logps/chosen": -1.4457646608352661,
+      "logps/rejected": -3.8541057109832764,
+      "loss": 3.6948,
+      "rewards/accuracies": 0.800000011920929,
+      "rewards/chosen": -2.168646812438965,
+      "rewards/margins": 3.612511396408081,
+      "rewards/rejected": -5.781157970428467,
+      "step": 430
+    },
+    {
+      "epoch": 0.36109971276159214,
+      "grad_norm": 1.8899520635604858,
+      "learning_rate": 4.011497787155938e-06,
+      "logits/chosen": -2.380706310272217,
+      "logits/rejected": -2.628760576248169,
+      "logps/chosen": -1.5100795030593872,
+      "logps/rejected": -4.016777515411377,
+      "loss": 3.7709,
+      "rewards/accuracies": 0.7875000238418579,
+      "rewards/chosen": -2.2651190757751465,
+      "rewards/margins": 3.760047197341919,
+      "rewards/rejected": -6.0251665115356445,
+      "step": 440
+    },
+    {
+      "epoch": 0.3693065244152647,
+      "grad_norm": 1.3820478916168213,
+      "learning_rate": 3.969463130731183e-06,
+      "logits/chosen": -2.3203773498535156,
+      "logits/rejected": -2.7199389934539795,
+      "logps/chosen": -1.7879537343978882,
+      "logps/rejected": -4.4142656326293945,
+      "loss": 3.6878,
+      "rewards/accuracies": 0.8374999761581421,
+      "rewards/chosen": -2.6819303035736084,
+      "rewards/margins": 3.9394683837890625,
+      "rewards/rejected": -6.621399879455566,
+      "step": 450
+    },
+    {
+      "epoch": 0.3693065244152647,
+      "eval_logits/chosen": -2.3454010486602783,
+      "eval_logits/rejected": -2.7509803771972656,
+      "eval_logps/chosen": -2.039729595184326,
+      "eval_logps/rejected": -4.705678462982178,
+      "eval_loss": 0.43350929021835327,
+      "eval_rewards/accuracies": 0.8383838534355164,
+      "eval_rewards/chosen": -3.05959415435791,
+      "eval_rewards/margins": 3.9989237785339355,
+      "eval_rewards/rejected": -7.058517932891846,
+      "eval_runtime": 26.0827,
+      "eval_samples_per_second": 30.212,
+      "eval_steps_per_second": 3.796,
+      "step": 450
+    },
+    {
+      "epoch": 0.3775133360689372,
+      "grad_norm": 1.2694813013076782,
+      "learning_rate": 3.92678391921108e-06,
+      "logits/chosen": -2.419553279876709,
+      "logits/rejected": -2.722438335418701,
+      "logps/chosen": -2.116955280303955,
+      "logps/rejected": -4.574510097503662,
+      "loss": 3.5762,
+      "rewards/accuracies": 0.8125,
+      "rewards/chosen": -3.1754324436187744,
+      "rewards/margins": 3.6863322257995605,
+      "rewards/rejected": -6.861765384674072,
+      "step": 460
+    },
+    {
+      "epoch": 0.38572014772260976,
+      "grad_norm": 2.1713900566101074,
+      "learning_rate": 3.88347887310836e-06,
+      "logits/chosen": -2.509795904159546,
+      "logits/rejected": -2.8286545276641846,
+      "logps/chosen": -3.043957233428955,
+      "logps/rejected": -5.701972007751465,
+      "loss": 3.3953,
+      "rewards/accuracies": 0.824999988079071,
+      "rewards/chosen": -4.565936088562012,
+      "rewards/margins": 3.9870212078094482,
+      "rewards/rejected": -8.552957534790039,
+      "step": 470
+    },
+    {
+      "epoch": 0.3939269593762823,
+      "grad_norm": 1.7294141054153442,
+      "learning_rate": 3.839566987447492e-06,
+      "logits/chosen": -2.5797224044799805,
+      "logits/rejected": -2.8154430389404297,
+      "logps/chosen": -3.6535918712615967,
+      "logps/rejected": -6.4350786209106445,
+      "loss": 2.8405,
+      "rewards/accuracies": 0.862500011920929,
+      "rewards/chosen": -5.4803876876831055,
+      "rewards/margins": 4.172229290008545,
+      "rewards/rejected": -9.652616500854492,
+      "step": 480
+    },
+    {
+      "epoch": 0.40213377102995485,
+      "grad_norm": 2.604421615600586,
+      "learning_rate": 3.795067523432826e-06,
+      "logits/chosen": -2.658399820327759,
+      "logits/rejected": -2.8255085945129395,
+      "logps/chosen": -5.536412239074707,
+      "logps/rejected": -7.725207328796387,
+      "loss": 2.5705,
+      "rewards/accuracies": 0.8999999761581421,
+      "rewards/chosen": -8.304617881774902,
+      "rewards/margins": 3.2831923961639404,
+      "rewards/rejected": -11.587809562683105,
+      "step": 490
+    },
+    {
+      "epoch": 0.4103405826836274,
+      "grad_norm": 3.06144118309021,
+      "learning_rate": 3.7500000000000005e-06,
+      "logits/chosen": -2.4845776557922363,
+      "logits/rejected": -2.7703354358673096,
+      "logps/chosen": -5.598423004150391,
+      "logps/rejected": -8.852571487426758,
+      "loss": 2.6946,
+      "rewards/accuracies": 0.887499988079071,
+      "rewards/chosen": -8.397635459899902,
+      "rewards/margins": 4.881222724914551,
+      "rewards/rejected": -13.27885913848877,
+      "step": 500
+    },
+    {
+      "epoch": 0.4103405826836274,
+      "eval_logits/chosen": -2.3113462924957275,
+      "eval_logits/rejected": -2.7164077758789062,
+      "eval_logps/chosen": -5.1130757331848145,
+      "eval_logps/rejected": -9.016916275024414,
+      "eval_loss": 0.28886228799819946,
+      "eval_rewards/accuracies": 0.9090909361839294,
+      "eval_rewards/chosen": -7.669614791870117,
+      "eval_rewards/margins": 5.8557610511779785,
+      "eval_rewards/rejected": -13.525375366210938,
+      "eval_runtime": 26.082,
+      "eval_samples_per_second": 30.212,
+      "eval_steps_per_second": 3.796,
+      "step": 500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 1500,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 50,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.8757331773271572e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-500/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b0a49fad07921611d3facd1cc353c724249c18c8e6de9b86b7e8565d2546ea37
+size 7224

checkpoint-500/zero_to_fp32.py

@@ -0,0 +1,674 @@
+#!/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 json
+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)
+
+        if BUFFER_NAMES not in state_dict:
+            raise ValueError(f"{file} is not a model state checkpoint")
+        buffer_names = state_dict[BUFFER_NAMES]
+        if debug:
+            print("Found buffers:", buffer_names)
+
+        # recover just the buffers while restoring them to fp32 if they were saved in fp16
+        buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
+        param_shapes = state_dict[PARAM_SHAPES]
+
+        # collect parameters that are included in param_shapes
+        param_names = []
+        for s in param_shapes:
+            for name in s.keys():
+                param_names.append(name)
+
+        # update with frozen parameters
+        frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
+        if frozen_param_shapes is not None:
+            if debug:
+                print(f"Found frozen_param_shapes: {frozen_param_shapes}")
+            param_names += list(frozen_param_shapes.keys())
+
+        # handle shared params
+        shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
+
+        ds_version = state_dict.get(DS_VERSION, None)
+
+        frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
+
+        z_model_state = zero_model_state(buffers=buffers,
+                                         param_shapes=param_shapes,
+                                         shared_params=shared_params,
+                                         ds_version=ds_version,
+                                         frozen_param_shapes=frozen_param_shapes,
+                                         frozen_param_fragments=frozen_param_fragments)
+        zero_model_states.append(z_model_state)
+
+    return zero_model_states
+
+
+def parse_optim_states(files, ds_checkpoint_dir):
+    total_files = len(files)
+    state_dicts = []
+    for f in files:
+        state_dict = torch.load(f, map_location=device)
+        # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
+        # and also handle the case where it was already removed by another helper script
+        state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
+        state_dicts.append(state_dict)
+
+    if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
+        raise ValueError(f"{files[0]} is not a zero checkpoint")
+    zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
+    world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
+
+    # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
+    # parameters can be different from data parallelism for non-expert parameters. So we can just
+    # use the max of the partition_count to get the dp world_size.
+
+    if type(world_size) is list:
+        world_size = max(world_size)
+
+    if world_size != total_files:
+        raise ValueError(
+            f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
+            "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
+        )
+
+    # the groups are named differently in each stage
+    if zero_stage <= 2:
+        fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
+    elif zero_stage == 3:
+        fp32_groups_key = FP32_FLAT_GROUPS
+    else:
+        raise ValueError(f"unknown zero stage {zero_stage}")
+
+    if zero_stage <= 2:
+        fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
+    elif zero_stage == 3:
+        # if there is more than one param group, there will be multiple flattened tensors - one
+        # flattened tensor per group - for simplicity merge them into a single tensor
+        #
+        # XXX: could make the script more memory efficient for when there are multiple groups - it
+        # will require matching the sub-lists of param_shapes for each param group flattened tensor
+
+        fp32_flat_groups = [
+            torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
+        ]
+
+    return zero_stage, world_size, fp32_flat_groups
+
+
+def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
+    """
+    Returns fp32 state_dict reconstructed from ds checkpoint
+
+    Args:
+        - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
+
+    """
+    print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
+
+    optim_files = get_optim_files(ds_checkpoint_dir)
+    zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
+    print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
+
+    model_files = get_model_state_files(ds_checkpoint_dir)
+
+    zero_model_states = parse_model_states(model_files)
+    print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
+
+    if zero_stage <= 2:
+        return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+    elif zero_stage == 3:
+        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+
+
+def _zero2_merge_frozen_params(state_dict, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+    frozen_param_fragments = zero_model_states[0].frozen_param_fragments
+
+    if debug:
+        num_elem = sum(s.numel() for s in frozen_param_shapes.values())
+        print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        state_dict[name] = frozen_param_fragments[name]
+
+        if debug:
+            print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _has_callable(obj, fn):
+    attr = getattr(obj, fn, None)
+    return callable(attr)
+
+
+def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+
+    # Reconstruction protocol:
+    #
+    # XXX: document this
+
+    if debug:
+        for i in range(world_size):
+            for j in range(len(fp32_flat_groups[0])):
+                print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
+
+    # XXX: memory usage doubles here (zero2)
+    num_param_groups = len(fp32_flat_groups[0])
+    merged_single_partition_of_fp32_groups = []
+    for i in range(num_param_groups):
+        merged_partitions = [sd[i] for sd in fp32_flat_groups]
+        full_single_fp32_vector = torch.cat(merged_partitions, 0)
+        merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
+    avail_numel = sum(
+        [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
+
+    if debug:
+        wanted_params = sum([len(shapes) for shapes in param_shapes])
+        wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
+        # not asserting if there is a mismatch due to possible padding
+        print(f"Have {avail_numel} numels to process.")
+        print(f"Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    total_numel = 0
+    total_params = 0
+    for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
+        offset = 0
+        avail_numel = full_single_fp32_vector.numel()
+        for name, shape in shapes.items():
+
+            unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
+            total_numel += unpartitioned_numel
+            total_params += 1
+
+            if debug:
+                print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+            state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
+            offset += unpartitioned_numel
+
+        # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
+        # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
+        # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
+        # live optimizer object, so we are checking that the numbers are within the right range
+        align_to = 2 * world_size
+
+        def zero2_align(x):
+            return align_to * math.ceil(x / align_to)
+
+        if debug:
+            print(f"original offset={offset}, avail_numel={avail_numel}")
+
+        offset = zero2_align(offset)
+        avail_numel = zero2_align(avail_numel)
+
+        if debug:
+            print(f"aligned  offset={offset}, avail_numel={avail_numel}")
+
+        # Sanity check
+        if offset != avail_numel:
+            raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero2_merge_frozen_params(state_dict, zero_model_states)
+
+    _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def zero3_partitioned_param_info(unpartitioned_numel, world_size):
+    remainder = unpartitioned_numel % world_size
+    padding_numel = (world_size - remainder) if remainder else 0
+    partitioned_numel = math.ceil(unpartitioned_numel / world_size)
+    return partitioned_numel, padding_numel
+
+
+def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    if debug:
+        for i in range(world_size):
+            num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
+            print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in zero_model_states[0].frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
+        state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+    avail_numel = fp32_flat_groups[0].numel() * world_size
+    # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
+    # param, re-consolidating each param, while dealing with padding if any
+
+    # merge list of dicts, preserving order
+    param_shapes = {k: v for d in param_shapes for k, v in d.items()}
+
+    if debug:
+        for i in range(world_size):
+            print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
+
+        wanted_params = len(param_shapes)
+        wanted_numel = sum(shape.numel() for shape in param_shapes.values())
+        # not asserting if there is a mismatch due to possible padding
+        avail_numel = fp32_flat_groups[0].numel() * world_size
+        print(f"Trainable params: Have {avail_numel} numels to process.")
+        print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    offset = 0
+    total_numel = 0
+    total_params = 0
+    for name, shape in 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}"
+            )
+
+        # XXX: memory usage doubles here
+        state_dict[name] = torch.cat(
+            tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
+            0).narrow(0, 0, unpartitioned_numel).view(shape)
+        offset += partitioned_numel
+
+    offset *= world_size
+
+    # Sanity check
+    if offset != avail_numel:
+        raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
+
+    _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
+    ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
+    via a model hub.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+
+    Returns:
+        - pytorch ``state_dict``
+
+    Note: this approach may not work if your application doesn't have sufficient free CPU memory and
+    you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
+    the checkpoint.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        # do the training and checkpoint saving
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
+        model = model.cpu() # move to cpu
+        model.load_state_dict(state_dict)
+        # submit to model hub or save the model to share with others
+
+    In this example the ``model`` will no longer be usable in the deepspeed context of the same
+    application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
+
+    """
+    if tag is None:
+        latest_path = os.path.join(checkpoint_dir, 'latest')
+        if os.path.isfile(latest_path):
+            with open(latest_path, 'r') as fd:
+                tag = fd.read().strip()
+        else:
+            raise ValueError(f"Unable to find 'latest' file at {latest_path}")
+
+    ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
+
+    if not os.path.isdir(ds_checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
+
+    return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
+
+
+def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir,
+                                               output_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)
+
+    # 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")
+        state_dict_split = split_torch_state_dict_into_shards(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
+    filename_to_tensors = state_dict_split.filename_to_tensors.items()
+    for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"):
+        shard = {tensor: state_dict[tensor].contiguous() for tensor in tensors}
+        output_path = os.path.join(output_dir, shard_file)
+        if safe_serialization:
+            save_file(shard, output_path, metadata={"format": "pt"})
+        else:
+            torch.save(shard, output_path)
+
+    # 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(f"Extracting fp32 weights")
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+
+    logger.info(f"Overwriting model with fp32 weights")
+    model = model.cpu()
+    model.load_state_dict(state_dict, strict=False)
+
+    return model
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("checkpoint_dir",
+                        type=str,
+                        help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
+    parser.add_argument("output_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)