Upload folder using huggingface_hub

README.md

@@ -0,0 +1,202 @@
+---
+base_model: meta-llama/Llama-3.2-11B-Vision-Instruct
+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 0.15.2

adapter_config.json

@@ -0,0 +1,31 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "/data1/chenjiazun/LLM/Llama-3.2-11B-Vision-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": [],
+  "peft_type": "LORA",
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": "^(language_model).*\\.(v_proj|o_proj|q_proj|gate_proj|k_proj|down_proj|up_proj)$",
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3564da42c58fdbed57b4705236b6ae8be0aca361a8526d231d431ee0eaffd1cb
+size 104941304

additional_config.json

@@ -0,0 +1 @@
+{"lora_dtype": null, "lorap_lr_ratio": null, "lorap_emb_lr": 1e-06}

args.json

@@ -0,0 +1,438 @@
+{
+  "model": "/data1/chenjiazun/LLM/Llama-3.2-11B-Vision-Instruct",
+  "model_type": "llama3_2_vision",
+  "model_revision": null,
+  "task_type": "causal_lm",
+  "torch_dtype": "bfloat16",
+  "attn_impl": null,
+  "num_labels": null,
+  "problem_type": null,
+  "rope_scaling": null,
+  "device_map": null,
+  "max_memory": {},
+  "local_repo_path": null,
+  "template": "llama3_2_vision",
+  "system": "You are a helpful assistant that truthfully answers user questions about the provided image.",
+  "max_length": 2048,
+  "truncation_strategy": "delete",
+  "max_pixels": null,
+  "agent_template": null,
+  "norm_bbox": null,
+  "response_prefix": null,
+  "padding_side": "right",
+  "loss_scale": "last_round",
+  "sequence_parallel_size": 1,
+  "use_chat_template": true,
+  "template_backend": "swift",
+  "dataset": [
+    "json_files/task1_dpo_absgt10_noidk__pari0.5_0602.json"
+  ],
+  "val_dataset": [],
+  "split_dataset_ratio": 0.01,
+  "data_seed": 42,
+  "dataset_num_proc": 4,
+  "dataset_shuffle": true,
+  "val_dataset_shuffle": false,
+  "streaming": false,
+  "interleave_prob": null,
+  "stopping_strategy": "first_exhausted",
+  "shuffle_buffer_size": 1000,
+  "enable_cache": false,
+  "download_mode": "reuse_dataset_if_exists",
+  "columns": {},
+  "strict": false,
+  "remove_unused_columns": true,
+  "model_name": [
+    null,
+    null
+  ],
+  "model_author": [
+    null,
+    null
+  ],
+  "custom_dataset_info": [],
+  "quant_method": null,
+  "quant_bits": null,
+  "hqq_axis": null,
+  "bnb_4bit_compute_dtype": "bfloat16",
+  "bnb_4bit_quant_type": "nf4",
+  "bnb_4bit_use_double_quant": true,
+  "bnb_4bit_quant_storage": null,
+  "max_new_tokens": 64,
+  "temperature": 0.9,
+  "top_k": 50,
+  "top_p": 0.9,
+  "repetition_penalty": 1.0,
+  "num_beams": 1,
+  "stream": false,
+  "stop_words": [],
+  "logprobs": false,
+  "top_logprobs": null,
+  "ckpt_dir": null,
+  "load_dataset_config": null,
+  "lora_modules": [],
+  "tuner_backend": "peft",
+  "train_type": "lora",
+  "adapters": [],
+  "external_plugins": [],
+  "seed": 42,
+  "model_kwargs": {},
+  "load_args": false,
+  "load_data_args": false,
+  "use_hf": false,
+  "hub_token": null,
+  "custom_register_path": [],
+  "ignore_args_error": false,
+  "use_swift_lora": false,
+  "output_dir": "/data2/kdd/crag/cjz/output/task1_dpo_absgt10_noidk__pari0.5_0602_p2prompt_r16/v0-20250604-235901",
+  "overwrite_output_dir": false,
+  "do_train": false,
+  "do_eval": false,
+  "do_predict": false,
+  "eval_strategy": "steps",
+  "prediction_loss_only": false,
+  "per_device_train_batch_size": 2,
+  "per_device_eval_batch_size": 2,
+  "per_gpu_train_batch_size": null,
+  "per_gpu_eval_batch_size": null,
+  "gradient_accumulation_steps": 16,
+  "eval_accumulation_steps": null,
+  "eval_delay": 0,
+  "torch_empty_cache_steps": null,
+  "learning_rate": 0.0002,
+  "weight_decay": 0.1,
+  "adam_beta1": 0.9,
+  "adam_beta2": 0.95,
+  "adam_epsilon": 1e-08,
+  "max_grad_norm": 1.0,
+  "num_train_epochs": 3.0,
+  "max_steps": -1,
+  "lr_scheduler_type": "cosine",
+  "lr_scheduler_kwargs": null,
+  "warmup_ratio": 0.05,
+  "warmup_steps": 0,
+  "log_level": "passive",
+  "log_level_replica": "warning",
+  "log_on_each_node": true,
+  "logging_dir": "/data2/kdd/crag/cjz/output/task1_dpo_absgt10_noidk__pari0.5_0602_p2prompt_r16/v0-20250604-235901/runs",
+  "logging_strategy": "steps",
+  "logging_first_step": true,
+  "logging_steps": 1,
+  "logging_nan_inf_filter": true,
+  "save_strategy": "steps",
+  "save_steps": 50.0,
+  "save_total_limit": null,
+  "save_safetensors": true,
+  "save_on_each_node": false,
+  "save_only_model": false,
+  "restore_callback_states_from_checkpoint": false,
+  "no_cuda": false,
+  "use_cpu": false,
+  "use_mps_device": false,
+  "jit_mode_eval": false,
+  "use_ipex": false,
+  "bf16": true,
+  "fp16": false,
+  "fp16_opt_level": "O1",
+  "half_precision_backend": "auto",
+  "bf16_full_eval": false,
+  "fp16_full_eval": false,
+  "tf32": null,
+  "local_rank": 0,
+  "ddp_backend": null,
+  "tpu_num_cores": null,
+  "tpu_metrics_debug": false,
+  "debug": null,
+  "dataloader_drop_last": false,
+  "eval_steps": 300.0,
+  "dataloader_num_workers": 4,
+  "dataloader_prefetch_factor": null,
+  "past_index": -1,
+  "run_name": null,
+  "disable_tqdm": null,
+  "label_names": null,
+  "load_best_model_at_end": false,
+  "metric_for_best_model": "loss",
+  "greater_is_better": false,
+  "ignore_data_skip": false,
+  "fsdp": "",
+  "fsdp_min_num_params": 0,
+  "fsdp_config": null,
+  "tp_size": 0,
+  "fsdp_transformer_layer_cls_to_wrap": null,
+  "accelerator_config": {
+    "dispatch_batches": false
+  },
+  "deepspeed": {
+    "fp16": {
+      "enabled": "auto",
+      "loss_scale": 0,
+      "loss_scale_window": 1000,
+      "initial_scale_power": 16,
+      "hysteresis": 2,
+      "min_loss_scale": 1
+    },
+    "bf16": {
+      "enabled": "auto"
+    },
+    "zero_optimization": {
+      "stage": 2,
+      "offload_optimizer": {
+        "device": "none",
+        "pin_memory": true
+      },
+      "allgather_partitions": true,
+      "allgather_bucket_size": 200000000.0,
+      "overlap_comm": false,
+      "reduce_scatter": true,
+      "reduce_bucket_size": 200000000.0,
+      "contiguous_gradients": true
+    },
+    "gradient_accumulation_steps": "auto",
+    "gradient_clipping": "auto",
+    "steps_per_print": 2000,
+    "train_batch_size": "auto",
+    "train_micro_batch_size_per_gpu": "auto",
+    "wall_clock_breakdown": false
+  },
+  "label_smoothing_factor": 0.0,
+  "optim": "adamw_torch",
+  "optim_args": null,
+  "adafactor": false,
+  "group_by_length": false,
+  "length_column_name": "length",
+  "report_to": [
+    "tensorboard"
+  ],
+  "ddp_find_unused_parameters": null,
+  "ddp_bucket_cap_mb": null,
+  "ddp_broadcast_buffers": null,
+  "dataloader_pin_memory": true,
+  "dataloader_persistent_workers": false,
+  "skip_memory_metrics": true,
+  "use_legacy_prediction_loop": false,
+  "push_to_hub": false,
+  "resume_from_checkpoint": null,
+  "hub_model_id": null,
+  "hub_strategy": "every_save",
+  "hub_private_repo": null,
+  "hub_always_push": false,
+  "gradient_checkpointing": true,
+  "gradient_checkpointing_kwargs": null,
+  "include_inputs_for_metrics": false,
+  "include_for_metrics": [],
+  "eval_do_concat_batches": true,
+  "fp16_backend": "auto",
+  "push_to_hub_model_id": null,
+  "push_to_hub_organization": null,
+  "push_to_hub_token": null,
+  "mp_parameters": "",
+  "auto_find_batch_size": false,
+  "full_determinism": false,
+  "torchdynamo": null,
+  "ray_scope": "last",
+  "ddp_timeout": 1800,
+  "torch_compile": false,
+  "torch_compile_backend": null,
+  "torch_compile_mode": null,
+  "include_tokens_per_second": false,
+  "include_num_input_tokens_seen": false,
+  "neftune_noise_alpha": null,
+  "optim_target_modules": null,
+  "batch_eval_metrics": false,
+  "eval_on_start": false,
+  "use_liger_kernel": false,
+  "eval_use_gather_object": false,
+  "average_tokens_across_devices": false,
+  "sortish_sampler": false,
+  "predict_with_generate": false,
+  "generation_max_length": null,
+  "generation_num_beams": null,
+  "generation_config": null,
+  "check_model": true,
+  "acc_strategy": "token",
+  "train_dataloader_shuffle": true,
+  "metric_warmup_step": 0,
+  "fsdp_num": 1,
+  "acc_steps": 1,
+  "eval_use_evalscope": false,
+  "eval_datasets": [],
+  "eval_limit": null,
+  "eval_datasets_args": null,
+  "eval_generation_config": null,
+  "freeze_parameters": [
+    "vision_model",
+    "multi_modal_projector"
+  ],
+  "freeze_parameters_ratio": 0.0,
+  "trainable_parameters": [],
+  "freeze_llm": false,
+  "freeze_vit": true,
+  "freeze_aligner": true,
+  "target_modules": [
+    "all-linear"
+  ],
+  "target_regex": null,
+  "modules_to_save": [],
+  "lora_rank": 16,
+  "lora_alpha": 32,
+  "lora_dropout": 0.05,
+  "lora_bias": "none",
+  "lora_dtype": null,
+  "lorap_lr_ratio": null,
+  "use_rslora": false,
+  "use_dora": false,
+  "lora_ga_batch_size": 2,
+  "lora_ga_iters": 2,
+  "lora_ga_max_length": 1024,
+  "lora_ga_direction": "ArB2r",
+  "lora_ga_scale": "stable",
+  "lora_ga_stable_gamma": 16,
+  "init_weights": true,
+  "fourier_n_frequency": 2000,
+  "fourier_scaling": 300.0,
+  "boft_block_size": 4,
+  "boft_block_num": 0,
+  "boft_n_butterfly_factor": 1,
+  "boft_dropout": 0.0,
+  "vera_rank": 256,
+  "vera_projection_prng_key": 0,
+  "vera_dropout": 0.0,
+  "vera_d_initial": 0.1,
+  "adapter_act": "gelu",
+  "adapter_length": 128,
+  "use_galore": false,
+  "galore_target_modules": null,
+  "galore_rank": 128,
+  "galore_update_proj_gap": 50,
+  "galore_scale": 1.0,
+  "galore_proj_type": "std",
+  "galore_optim_per_parameter": false,
+  "galore_with_embedding": false,
+  "galore_quantization": false,
+  "galore_proj_quant": false,
+  "galore_proj_bits": 4,
+  "galore_proj_group_size": 256,
+  "galore_cos_threshold": 0.4,
+  "galore_gamma_proj": 2,
+  "galore_queue_size": 5,
+  "adalora_target_r": 8,
+  "adalora_init_r": 12,
+  "adalora_tinit": 0,
+  "adalora_tfinal": 0,
+  "adalora_deltaT": 1,
+  "adalora_beta1": 0.85,
+  "adalora_beta2": 0.85,
+  "adalora_orth_reg_weight": 0.5,
+  "llamapro_num_new_blocks": 4,
+  "llamapro_num_groups": null,
+  "lisa_activated_layers": 0,
+  "lisa_step_interval": 20,
+  "reft_layer_key": null,
+  "reft_layers": null,
+  "reft_rank": 4,
+  "reft_intervention_type": "LoreftIntervention",
+  "reft_args": null,
+  "swanlab_token": null,
+  "swanlab_project": null,
+  "swanlab_workspace": null,
+  "swanlab_exp_name": null,
+  "swanlab_mode": "cloud",
+  "add_version": true,
+  "resume_only_model": false,
+  "create_checkpoint_symlink": false,
+  "packing": false,
+  "lazy_tokenize": true,
+  "loss_type": "sigmoid",
+  "optimizer": null,
+  "metric": null,
+  "zero_hpz_partition_size": null,
+  "reward_model": null,
+  "reward_adapters": [],
+  "reward_model_type": null,
+  "reward_model_revision": null,
+  "num_ppo_epochs": 4,
+  "whiten_rewards": false,
+  "kl_coef": 0.05,
+  "cliprange": 0.2,
+  "vf_coef": 0.1,
+  "cliprange_value": 0.2,
+  "gamma": 1.0,
+  "lam": 0.95,
+  "num_mini_batches": 1,
+  "local_rollout_forward_batch_size": 64,
+  "num_sample_generations": 10,
+  "response_length": 512,
+  "missing_eos_penalty": null,
+  "epsilon": 0.2,
+  "epsilon_high": null,
+  "num_infer_workers": 1,
+  "vllm_device": [
+    "auto"
+  ],
+  "vllm_gpu_memory_utilization": 0.9,
+  "vllm_max_model_len": null,
+  "vllm_max_num_seqs": 256,
+  "vllm_enforce_eager": false,
+  "vllm_limit_mm_per_prompt": null,
+  "vllm_enable_prefix_caching": true,
+  "cosine_min_len_value_wrong": -0.5,
+  "cosine_max_len_value_wrong": 0.0,
+  "cosine_min_len_value_correct": 1.0,
+  "cosine_max_len_value_correct": 0.5,
+  "cosine_max_len": null,
+  "repetition_n_grams": 3,
+  "repetition_max_penalty": -1.0,
+  "use_lmdeploy": false,
+  "lmdeploy_device": "auto",
+  "lmdeploy_session_len": null,
+  "lmdeploy_cache_max_entry_count": 0.8,
+  "async_generate": false,
+  "tensor_parallel_size": 1,
+  "sleep_level": 0,
+  "move_model_batches": null,
+  "offload_optimizer": false,
+  "offload_model": false,
+  "gc_collect_after_offload": false,
+  "multi_turn_func": null,
+  "dynamic_sample": false,
+  "max_resample_times": 3,
+  "overlong_filter": false,
+  "soft_max_length": null,
+  "soft_cache_length": null,
+  "scale_rewards": true,
+  "wandb_log_unique_prompts": null,
+  "vllm_server_host": null,
+  "vllm_server_port": 8000,
+  "vllm_server_timeout": 240.0,
+  "num_generations": 8,
+  "max_completion_length": 512,
+  "ds3_gather_for_generation": true,
+  "reward_funcs": [],
+  "reward_weights": null,
+  "log_completions": false,
+  "use_vllm": false,
+  "num_iterations": 1,
+  "rlhf_type": "dpo",
+  "ref_model": null,
+  "ref_model_type": null,
+  "ref_model_revision": null,
+  "beta": 0.1,
+  "label_smoothing": 0,
+  "rpo_alpha": 1.0,
+  "cpo_alpha": 1.0,
+  "simpo_gamma": 1,
+  "desirable_weight": 1.0,
+  "undesirable_weight": 1.0,
+  "center_rewards_coefficient": null,
+  "rank": 0,
+  "global_world_size": 1,
+  "local_world_size": 1,
+  "model_suffix": "Llama-3.2-11B-Vision-Instruct",
+  "model_info": "ModelInfo(model_type='llama3_2_vision', model_dir='/data1/chenjiazun/LLM/Llama-3.2-11B-Vision-Instruct', torch_dtype=torch.bfloat16, max_model_len=131072, quant_method=None, quant_bits=None, rope_scaling={'factor': 8.0, 'high_freq_factor': 4.0, 'low_freq_factor': 1.0, 'original_max_position_embeddings': 8192, 'rope_type': 'llama3'}, config=None, task_type='causal_lm', num_labels=None)",
+  "model_meta": "ModelMeta(model_type='llama3_2_vision', model_groups=[ModelGroup(models=[Model(ms_model_id='LLM-Research/Llama-3.2-11B-Vision-Instruct', hf_model_id='meta-llama/Llama-3.2-11B-Vision-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='LLM-Research/Llama-3.2-90B-Vision-Instruct', hf_model_id='meta-llama/Llama-3.2-90B-Vision-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='LLM-Research/Llama-3.2-11B-Vision', hf_model_id='meta-llama/Llama-3.2-11B-Vision', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='LLM-Research/Llama-3.2-90B-Vision', hf_model_id='meta-llama/Llama-3.2-90B-Vision', model_path=None, ms_revision=None, hf_revision=None)], ignore_patterns=None, requires=None, tags=[])], template='llama3_2_vision', get_function=<function get_model_tokenizer_llama3_2_vision at 0x720c47d2ad40>, model_arch='llama3_2_vision', architectures=['MllamaForConditionalGeneration'], additional_saved_files=[], torch_dtype=None, is_multimodal=True, is_reward=False, task_type=None, ignore_patterns=None, requires=['transformers>=4.45'], tags=[])",
+  "model_dir": "/data1/chenjiazun/LLM/Llama-3.2-11B-Vision-Instruct",
+  "hub": "<class 'swift.hub.hub.MSHub'>",
+  "evaluation_strategy": "steps",
+  "training_args": "DPOConfig(output_dir='/data2/kdd/crag/cjz/output/task1_dpo_absgt10_noidk__pari0.5_0602_p2prompt_r16/v0-20250604-235901', overwrite_output_dir=False, do_train=False, do_eval=True, do_predict=False, eval_strategy=<IntervalStrategy.STEPS: 'steps'>, prediction_loss_only=False, per_device_train_batch_size=2, per_device_eval_batch_size=2, per_gpu_train_batch_size=None, per_gpu_eval_batch_size=None, gradient_accumulation_steps=16, eval_accumulation_steps=None, eval_delay=0, torch_empty_cache_steps=None, learning_rate=0.0002, weight_decay=0.1, adam_beta1=0.9, adam_beta2=0.95, adam_epsilon=1e-08, max_grad_norm=1.0, num_train_epochs=3.0, max_steps=-1, lr_scheduler_type=<SchedulerType.COSINE: 'cosine'>, lr_scheduler_kwargs=None, warmup_ratio=0.05, warmup_steps=0, log_level='passive', log_level_replica='warning', log_on_each_node=True, logging_dir='/data2/kdd/crag/cjz/output/task1_dpo_absgt10_noidk__pari0.5_0602_p2prompt_r16/v0-20250604-235901/runs', logging_strategy=<IntervalStrategy.STEPS: 'steps'>, logging_first_step=True, logging_steps=1, logging_nan_inf_filter=True, save_strategy=<SaveStrategy.STEPS: 'steps'>, save_steps=50, save_total_limit=None, save_safetensors=True, save_on_each_node=False, save_only_model=False, restore_callback_states_from_checkpoint=False, no_cuda=False, use_cpu=False, use_mps_device=False, seed=42, data_seed=42, jit_mode_eval=False, use_ipex=False, bf16=True, fp16=False, fp16_opt_level='O1', half_precision_backend='auto', bf16_full_eval=False, fp16_full_eval=False, tf32=None, local_rank=0, ddp_backend=None, tpu_num_cores=None, tpu_metrics_debug=False, debug=[], dataloader_drop_last=False, eval_steps=300, dataloader_num_workers=4, dataloader_prefetch_factor=10, past_index=-1, run_name='/data2/kdd/crag/cjz/output/task1_dpo_absgt10_noidk__pari0.5_0602_p2prompt_r16/v0-20250604-235901', disable_tqdm=False, remove_unused_columns=False, label_names=None, load_best_model_at_end=False, metric_for_best_model='loss', greater_is_better=False, ignore_data_skip=False, fsdp=[], fsdp_min_num_params=0, fsdp_config={'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, tp_size=0, fsdp_transformer_layer_cls_to_wrap=None, accelerator_config=AcceleratorConfig(split_batches=False, dispatch_batches=False, even_batches=True, use_seedable_sampler=True, non_blocking=False, gradient_accumulation_kwargs=None, use_configured_state=False), deepspeed={'fp16': {'enabled': 'auto', 'loss_scale': 0, 'loss_scale_window': 1000, 'initial_scale_power': 16, 'hysteresis': 2, 'min_loss_scale': 1}, 'bf16': {'enabled': 'auto'}, 'zero_optimization': {'stage': 2, 'offload_optimizer': {'device': 'none', 'pin_memory': True}, 'allgather_partitions': True, 'allgather_bucket_size': 200000000.0, 'overlap_comm': False, 'reduce_scatter': True, 'reduce_bucket_size': 200000000.0, 'contiguous_gradients': True}, 'gradient_accumulation_steps': 'auto', 'gradient_clipping': 'auto', 'steps_per_print': 2000, 'train_batch_size': 'auto', 'train_micro_batch_size_per_gpu': 'auto', 'wall_clock_breakdown': False}, label_smoothing_factor=0.0, optim=<OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, optim_args=None, adafactor=False, group_by_length=False, length_column_name='length', report_to=['tensorboard'], ddp_find_unused_parameters=None, ddp_bucket_cap_mb=None, ddp_broadcast_buffers=None, dataloader_pin_memory=True, dataloader_persistent_workers=False, skip_memory_metrics=True, use_legacy_prediction_loop=False, push_to_hub=False, resume_from_checkpoint=None, hub_model_id=None, hub_strategy=<HubStrategy.EVERY_SAVE: 'every_save'>, hub_token=None, hub_private_repo=None, hub_always_push=False, gradient_checkpointing=True, gradient_checkpointing_kwargs=None, include_inputs_for_metrics=False, include_for_metrics=[], eval_do_concat_batches=True, fp16_backend='auto', push_to_hub_model_id=None, push_to_hub_organization=None, push_to_hub_token=None, mp_parameters='', auto_find_batch_size=False, full_determinism=False, torchdynamo=None, ray_scope='last', ddp_timeout=1800, torch_compile=False, torch_compile_backend=None, torch_compile_mode=None, include_tokens_per_second=None, include_num_input_tokens_seen=None, neftune_noise_alpha=None, optim_target_modules=None, batch_eval_metrics=False, eval_on_start=False, use_liger_kernel=False, eval_use_gather_object=False, average_tokens_across_devices=None, model_init_kwargs=None, ref_model_init_kwargs=None, model_adapter_name=None, ref_adapter_name=None, force_use_ref_model=False, disable_dropout=True, use_logits_to_keep=False, dataset_num_proc=4, padding_value=None, label_pad_token_id=None, max_prompt_length=512, max_completion_length=512, max_length=2048, truncation_mode='keep_end', padding_free=False, precompute_ref_log_probs=False, precompute_ref_batch_size=None, tools=None, loss_type='sigmoid', beta=0.1, f_divergence_type=<FDivergenceType.REVERSE_KL: 'reverse_kl'>, f_alpha_divergence_coef=1.0, reference_free=False, label_smoothing=0, use_weighting=False, rpo_alpha=1.0, discopop_tau=0.05, sync_ref_model=False, ref_model_mixup_alpha=0.6, ref_model_sync_steps=512, generate_during_eval=False, check_model=True, acc_strategy='token', train_dataloader_shuffle=True, metric_warmup_step=0, fsdp_num=1, acc_steps=1, eval_use_evalscope=False, eval_datasets=[], eval_limit=None, eval_datasets_args=None, eval_generation_config=None, train_type='lora', optimizer=None, local_repo_path=None, galore_config=None)"
+}

latest

@@ -0,0 +1 @@
+global_step50

rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:835f869ea325fd6edf27b48b589309fb66641cb92b45f2fc13d1bb6e8814106c
+size 14244

scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:693a99ff35c0fc4ad522ce2332e9065f07ffbaaa13e849f824aac0c9b503a2d4
+size 1064

trainer_state.json

@@ -0,0 +1,934 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.12574662055957248,
+  "eval_steps": 300,
+  "global_step": 50,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.002514932411191449,
+      "grad_norm": 5.559185028076172,
+      "learning_rate": 3.3333333333333333e-06,
+      "logits/chosen": -0.5158984065055847,
+      "logits/rejected": -0.6433685421943665,
+      "logps/chosen": -8.879493713378906,
+      "logps/rejected": -18.298219680786133,
+      "loss": 1.2430726289749146,
+      "memory(GiB)": 30.69,
+      "nll_loss": 0.5499253869056702,
+      "rewards/accuracies": 0.0,
+      "rewards/chosen": 0.0,
+      "rewards/margins": 0.0,
+      "rewards/rejected": 0.0,
+      "step": 1,
+      "train_speed(iter/s)": 0.010477
+    },
+    {
+      "epoch": 0.005029864822382898,
+      "grad_norm": 5.400235176086426,
+      "learning_rate": 6.666666666666667e-06,
+      "logits/chosen": -0.5219721794128418,
+      "logits/rejected": -0.6263731718063354,
+      "logps/chosen": -4.77984094619751,
+      "logps/rejected": -15.150838851928711,
+      "loss": 1.0675525665283203,
+      "memory(GiB)": 33.9,
+      "nll_loss": 0.3744054436683655,
+      "rewards/accuracies": 0.0,
+      "rewards/chosen": 0.0,
+      "rewards/margins": 0.0,
+      "rewards/rejected": 0.0,
+      "step": 2,
+      "train_speed(iter/s)": 0.010866
+    },
+    {
+      "epoch": 0.007544797233574348,
+      "grad_norm": 7.852550983428955,
+      "learning_rate": 1e-05,
+      "logits/chosen": -0.48881009221076965,
+      "logits/rejected": -0.6540625691413879,
+      "logps/chosen": -4.714770793914795,
+      "logps/rejected": -17.99374008178711,
+      "loss": 1.0851404666900635,
+      "memory(GiB)": 33.9,
+      "nll_loss": 0.3894187808036804,
+      "rewards/accuracies": 0.375,
+      "rewards/chosen": -0.002094469266012311,
+      "rewards/margins": -0.004971538204699755,
+      "rewards/rejected": 0.0028770684730261564,
+      "step": 3,
+      "train_speed(iter/s)": 0.011002
+    },
+    {
+      "epoch": 0.010059729644765796,
+      "grad_norm": 10.150762557983398,
+      "learning_rate": 1.3333333333333333e-05,
+      "logits/chosen": -0.4506370425224304,
+      "logits/rejected": -0.6422093510627747,
+      "logps/chosen": -3.6849846839904785,
+      "logps/rejected": -21.201997756958008,
+      "loss": 1.092491865158081,
+      "memory(GiB)": 37.36,
+      "nll_loss": 0.4020901322364807,
+      "rewards/accuracies": 0.5,
+      "rewards/chosen": 0.007022961974143982,
+      "rewards/margins": 0.0057290042750537395,
+      "rewards/rejected": 0.001293957349844277,
+      "step": 4,
+      "train_speed(iter/s)": 0.011051
+    },
+    {
+      "epoch": 0.012574662055957246,
+      "grad_norm": 6.588510036468506,
+      "learning_rate": 1.6666666666666667e-05,
+      "logits/chosen": -0.4159216284751892,
+      "logits/rejected": -0.5884239673614502,
+      "logps/chosen": -8.611150741577148,
+      "logps/rejected": -19.931533813476562,
+      "loss": 1.1986628770828247,
+      "memory(GiB)": 37.36,
+      "nll_loss": 0.5155895948410034,
+      "rewards/accuracies": 0.71875,
+      "rewards/chosen": 0.005048489198088646,
+      "rewards/margins": 0.020845137536525726,
+      "rewards/rejected": -0.01579664833843708,
+      "step": 5,
+      "train_speed(iter/s)": 0.011086
+    },
+    {
+      "epoch": 0.015089594467148696,
+      "grad_norm": 3.965754508972168,
+      "learning_rate": 2e-05,
+      "logits/chosen": -0.47118157148361206,
+      "logits/rejected": -0.5790206789970398,
+      "logps/chosen": -9.572708129882812,
+      "logps/rejected": -19.066829681396484,
+      "loss": 1.1965497732162476,
+      "memory(GiB)": 37.36,
+      "nll_loss": 0.5279884338378906,
+      "rewards/accuracies": 0.71875,
+      "rewards/chosen": 0.006116065196692944,
+      "rewards/margins": 0.05174415558576584,
+      "rewards/rejected": -0.04562808945775032,
+      "step": 6,
+      "train_speed(iter/s)": 0.01111
+    },
+    {
+      "epoch": 0.017604526878340146,
+      "grad_norm": 2.719625234603882,
+      "learning_rate": 2.3333333333333336e-05,
+      "logits/chosen": -0.40740644931793213,
+      "logits/rejected": -0.5419386029243469,
+      "logps/chosen": -9.413152694702148,
+      "logps/rejected": -24.703401565551758,
+      "loss": 1.1549676656723022,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.5149396061897278,
+      "rewards/accuracies": 0.71875,
+      "rewards/chosen": -0.02250390499830246,
+      "rewards/margins": 0.12123934924602509,
+      "rewards/rejected": -0.14374326169490814,
+      "step": 7,
+      "train_speed(iter/s)": 0.011114
+    },
+    {
+      "epoch": 0.020119459289531592,
+      "grad_norm": 3.325852394104004,
+      "learning_rate": 2.6666666666666667e-05,
+      "logits/chosen": -0.4091758728027344,
+      "logits/rejected": -0.5692812204360962,
+      "logps/chosen": -5.904629230499268,
+      "logps/rejected": -20.752662658691406,
+      "loss": 1.0213468074798584,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.3945533037185669,
+      "rewards/accuracies": 0.6875,
+      "rewards/chosen": -0.041384391486644745,
+      "rewards/margins": 0.17802417278289795,
+      "rewards/rejected": -0.2194085717201233,
+      "step": 8,
+      "train_speed(iter/s)": 0.011127
+    },
+    {
+      "epoch": 0.022634391700723042,
+      "grad_norm": 3.5531411170959473,
+      "learning_rate": 3e-05,
+      "logits/chosen": -0.3281947374343872,
+      "logits/rejected": -0.5156837701797485,
+      "logps/chosen": -5.975313186645508,
+      "logps/rejected": -27.066158294677734,
+      "loss": 0.8695712089538574,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.3611774742603302,
+      "rewards/accuracies": 0.875,
+      "rewards/chosen": 0.0457351878285408,
+      "rewards/margins": 0.4895857870578766,
+      "rewards/rejected": -0.443850576877594,
+      "step": 9,
+      "train_speed(iter/s)": 0.011136
+    },
+    {
+      "epoch": 0.025149324111914492,
+      "grad_norm": 4.356088161468506,
+      "learning_rate": 3.3333333333333335e-05,
+      "logits/chosen": -0.43971675634384155,
+      "logits/rejected": -0.5702978372573853,
+      "logps/chosen": -6.376318454742432,
+      "logps/rejected": -19.42030143737793,
+      "loss": 1.0046417713165283,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.42133060097694397,
+      "rewards/accuracies": 0.6875,
+      "rewards/chosen": -0.17984721064567566,
+      "rewards/margins": 0.43266722559928894,
+      "rewards/rejected": -0.6125144362449646,
+      "step": 10,
+      "train_speed(iter/s)": 0.01115
+    },
+    {
+      "epoch": 0.027664256523105942,
+      "grad_norm": 3.9732213020324707,
+      "learning_rate": 3.6666666666666666e-05,
+      "logits/chosen": -0.4561832845211029,
+      "logits/rejected": -0.6065505146980286,
+      "logps/chosen": -6.498957633972168,
+      "logps/rejected": -27.428552627563477,
+      "loss": 0.9565106630325317,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.40273764729499817,
+      "rewards/accuracies": 0.75,
+      "rewards/chosen": -0.12391924858093262,
+      "rewards/margins": 0.5706068277359009,
+      "rewards/rejected": -0.6945260763168335,
+      "step": 11,
+      "train_speed(iter/s)": 0.011161
+    },
+    {
+      "epoch": 0.030179188934297392,
+      "grad_norm": 2.2540578842163086,
+      "learning_rate": 4e-05,
+      "logits/chosen": -0.42088037729263306,
+      "logits/rejected": -0.6009025573730469,
+      "logps/chosen": -4.243446350097656,
+      "logps/rejected": -26.567686080932617,
+      "loss": 0.7599425911903381,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.28007903695106506,
+      "rewards/accuracies": 0.8125,
+      "rewards/chosen": -0.14110325276851654,
+      "rewards/margins": 0.6590132713317871,
+      "rewards/rejected": -0.8001165390014648,
+      "step": 12,
+      "train_speed(iter/s)": 0.011171
+    },
+    {
+      "epoch": 0.03269412134548884,
+      "grad_norm": 6.460833549499512,
+      "learning_rate": 4.3333333333333334e-05,
+      "logits/chosen": -0.4136754274368286,
+      "logits/rejected": -0.5452494621276855,
+      "logps/chosen": -7.585506439208984,
+      "logps/rejected": -21.19340705871582,
+      "loss": 1.223608136177063,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.6720148921012878,
+      "rewards/accuracies": 0.71875,
+      "rewards/chosen": -0.25483113527297974,
+      "rewards/margins": 0.49035438895225525,
+      "rewards/rejected": -0.7451854944229126,
+      "step": 13,
+      "train_speed(iter/s)": 0.011181
+    },
+    {
+      "epoch": 0.03520905375668029,
+      "grad_norm": 4.092947006225586,
+      "learning_rate": 4.666666666666667e-05,
+      "logits/chosen": -0.527854323387146,
+      "logits/rejected": -0.5994939804077148,
+      "logps/chosen": -8.185604095458984,
+      "logps/rejected": -17.946279525756836,
+      "loss": 1.2557107210159302,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.66472989320755,
+      "rewards/accuracies": 0.65625,
+      "rewards/chosen": -0.145107239484787,
+      "rewards/margins": 0.41058170795440674,
+      "rewards/rejected": -0.5556889772415161,
+      "step": 14,
+      "train_speed(iter/s)": 0.011194
+    },
+    {
+      "epoch": 0.03772398616787174,
+      "grad_norm": 3.926342248916626,
+      "learning_rate": 5e-05,
+      "logits/chosen": -0.407508909702301,
+      "logits/rejected": -0.5949459075927734,
+      "logps/chosen": -3.4148104190826416,
+      "logps/rejected": -22.69104766845703,
+      "loss": 0.8548452258110046,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.40506720542907715,
+      "rewards/accuracies": 0.8125,
+      "rewards/chosen": -0.025914989411830902,
+      "rewards/margins": 0.6952927708625793,
+      "rewards/rejected": -0.7212077379226685,
+      "step": 15,
+      "train_speed(iter/s)": 0.0112
+    },
+    {
+      "epoch": 0.040238918579063185,
+      "grad_norm": 2.4273853302001953,
+      "learning_rate": 5.333333333333333e-05,
+      "logits/chosen": -0.3462049961090088,
+      "logits/rejected": -0.58342045545578,
+      "logps/chosen": -3.8651039600372314,
+      "logps/rejected": -28.470134735107422,
+      "loss": 0.821665346622467,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.3309793472290039,
+      "rewards/accuracies": 0.84375,
+      "rewards/chosen": 0.04178892821073532,
+      "rewards/margins": 0.5164157152175903,
+      "rewards/rejected": -0.4746267795562744,
+      "step": 16,
+      "train_speed(iter/s)": 0.011202
+    },
+    {
+      "epoch": 0.04275385099025464,
+      "grad_norm": 1.8921568393707275,
+      "learning_rate": 5.666666666666667e-05,
+      "logits/chosen": -0.5054872632026672,
+      "logits/rejected": -0.6153637170791626,
+      "logps/chosen": -1.8910857439041138,
+      "logps/rejected": -16.781475067138672,
+      "loss": 0.7665270566940308,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.2412014603614807,
+      "rewards/accuracies": 0.875,
+      "rewards/chosen": 0.04062865674495697,
+      "rewards/margins": 0.41468262672424316,
+      "rewards/rejected": -0.3740540146827698,
+      "step": 17,
+      "train_speed(iter/s)": 0.011212
+    },
+    {
+      "epoch": 0.045268783401446085,
+      "grad_norm": 3.2724339962005615,
+      "learning_rate": 6e-05,
+      "logits/chosen": -0.45994284749031067,
+      "logits/rejected": -0.5327339768409729,
+      "logps/chosen": -8.705971717834473,
+      "logps/rejected": -17.33376121520996,
+      "loss": 1.2050995826721191,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.5643416047096252,
+      "rewards/accuracies": 0.625,
+      "rewards/chosen": -0.048418305814266205,
+      "rewards/margins": 0.16582812368869781,
+      "rewards/rejected": -0.21424642205238342,
+      "step": 18,
+      "train_speed(iter/s)": 0.011215
+    },
+    {
+      "epoch": 0.04778371581263754,
+      "grad_norm": 2.6650567054748535,
+      "learning_rate": 6.333333333333333e-05,
+      "logits/chosen": -0.48406994342803955,
+      "logits/rejected": -0.5535129308700562,
+      "logps/chosen": -10.087296485900879,
+      "logps/rejected": -17.584102630615234,
+      "loss": 1.1602699756622314,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.5509804487228394,
+      "rewards/accuracies": 0.6875,
+      "rewards/chosen": -0.005405411124229431,
+      "rewards/margins": 0.23162491619586945,
+      "rewards/rejected": -0.23703034222126007,
+      "step": 19,
+      "train_speed(iter/s)": 0.01122
+    },
+    {
+      "epoch": 0.050298648223828984,
+      "grad_norm": 2.393120527267456,
+      "learning_rate": 6.666666666666667e-05,
+      "logits/chosen": -0.4699333906173706,
+      "logits/rejected": -0.5745525360107422,
+      "logps/chosen": -5.482749938964844,
+      "logps/rejected": -15.294036865234375,
+      "loss": 1.0463979244232178,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.40886539220809937,
+      "rewards/accuracies": 0.71875,
+      "rewards/chosen": 0.030414806678891182,
+      "rewards/margins": 0.18470264971256256,
+      "rewards/rejected": -0.15428784489631653,
+      "step": 20,
+      "train_speed(iter/s)": 0.011224
+    },
+    {
+      "epoch": 0.05281358063502043,
+      "grad_norm": 1.7328065633773804,
+      "learning_rate": 7e-05,
+      "logits/chosen": -0.5130109190940857,
+      "logits/rejected": -0.6171243190765381,
+      "logps/chosen": -8.619268417358398,
+      "logps/rejected": -18.43971824645996,
+      "loss": 1.1801689863204956,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.6120539903640747,
+      "rewards/accuracies": 0.78125,
+      "rewards/chosen": 0.07324957102537155,
+      "rewards/margins": 0.3033149242401123,
+      "rewards/rejected": -0.23006536066532135,
+      "step": 21,
+      "train_speed(iter/s)": 0.011228
+    },
+    {
+      "epoch": 0.055328513046211884,
+      "grad_norm": 2.848578691482544,
+      "learning_rate": 7.333333333333333e-05,
+      "logits/chosen": -0.46683233976364136,
+      "logits/rejected": -0.5329924821853638,
+      "logps/chosen": -4.357663154602051,
+      "logps/rejected": -15.35399055480957,
+      "loss": 0.9406945705413818,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.3216635584831238,
+      "rewards/accuracies": 0.65625,
+      "rewards/chosen": -0.010179778560996056,
+      "rewards/margins": 0.2008945494890213,
+      "rewards/rejected": -0.21107429265975952,
+      "step": 22,
+      "train_speed(iter/s)": 0.011229
+    },
+    {
+      "epoch": 0.05784344545740333,
+      "grad_norm": 1.7426271438598633,
+      "learning_rate": 7.666666666666667e-05,
+      "logits/chosen": -0.5002316236495972,
+      "logits/rejected": -0.5488481521606445,
+      "logps/chosen": -5.690494537353516,
+      "logps/rejected": -15.590461730957031,
+      "loss": 0.9775924682617188,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.355554461479187,
+      "rewards/accuracies": 0.75,
+      "rewards/chosen": 0.009325886145234108,
+      "rewards/margins": 0.18430516123771667,
+      "rewards/rejected": -0.1749793142080307,
+      "step": 23,
+      "train_speed(iter/s)": 0.011234
+    },
+    {
+      "epoch": 0.060358377868594784,
+      "grad_norm": 1.8518409729003906,
+      "learning_rate": 8e-05,
+      "logits/chosen": -0.5258264541625977,
+      "logits/rejected": -0.5805689096450806,
+      "logps/chosen": -9.133979797363281,
+      "logps/rejected": -17.886051177978516,
+      "loss": 1.1810051202774048,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.6029603481292725,
+      "rewards/accuracies": 0.71875,
+      "rewards/chosen": 0.04269062727689743,
+      "rewards/margins": 0.2787190079689026,
+      "rewards/rejected": -0.23602835834026337,
+      "step": 24,
+      "train_speed(iter/s)": 0.011237
+    },
+    {
+      "epoch": 0.06287331027978624,
+      "grad_norm": 2.4842066764831543,
+      "learning_rate": 8.333333333333334e-05,
+      "logits/chosen": -0.44220152497291565,
+      "logits/rejected": -0.5788168907165527,
+      "logps/chosen": -4.989194869995117,
+      "logps/rejected": -18.047056198120117,
+      "loss": 1.0011777877807617,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.4066605269908905,
+      "rewards/accuracies": 0.71875,
+      "rewards/chosen": 0.021916719153523445,
+      "rewards/margins": 0.2587287425994873,
+      "rewards/rejected": -0.236812025308609,
+      "step": 25,
+      "train_speed(iter/s)": 0.011239
+    },
+    {
+      "epoch": 0.06538824269097768,
+      "grad_norm": 1.766322135925293,
+      "learning_rate": 8.666666666666667e-05,
+      "logits/chosen": -0.3389984369277954,
+      "logits/rejected": -0.543292224407196,
+      "logps/chosen": -11.735610961914062,
+      "logps/rejected": -26.25033950805664,
+      "loss": 1.0063844919204712,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.5172990560531616,
+      "rewards/accuracies": 0.84375,
+      "rewards/chosen": 0.07315510511398315,
+      "rewards/margins": 0.5734083652496338,
+      "rewards/rejected": -0.5002533197402954,
+      "step": 26,
+      "train_speed(iter/s)": 0.011238
+    },
+    {
+      "epoch": 0.06790317510216913,
+      "grad_norm": 3.3314130306243896,
+      "learning_rate": 9e-05,
+      "logits/chosen": -0.37741950154304504,
+      "logits/rejected": -0.5884826183319092,
+      "logps/chosen": -5.442512035369873,
+      "logps/rejected": -27.738323211669922,
+      "loss": 0.8821365237236023,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.4055330753326416,
+      "rewards/accuracies": 0.875,
+      "rewards/chosen": -0.03083166666328907,
+      "rewards/margins": 0.6056570410728455,
+      "rewards/rejected": -0.6364887952804565,
+      "step": 27,
+      "train_speed(iter/s)": 0.011238
+    },
+    {
+      "epoch": 0.07041810751336058,
+      "grad_norm": 2.849647283554077,
+      "learning_rate": 9.333333333333334e-05,
+      "logits/chosen": -0.4310716688632965,
+      "logits/rejected": -0.5529810190200806,
+      "logps/chosen": -8.224971771240234,
+      "logps/rejected": -26.79846954345703,
+      "loss": 0.8747767806053162,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.45313140749931335,
+      "rewards/accuracies": 0.875,
+      "rewards/chosen": -0.07478700578212738,
+      "rewards/margins": 0.8052245378494263,
+      "rewards/rejected": -0.8800115585327148,
+      "step": 28,
+      "train_speed(iter/s)": 0.011238
+    },
+    {
+      "epoch": 0.07293303992455202,
+      "grad_norm": 2.247368097305298,
+      "learning_rate": 9.666666666666667e-05,
+      "logits/chosen": -0.37296879291534424,
+      "logits/rejected": -0.5259015560150146,
+      "logps/chosen": -6.546780109405518,
+      "logps/rejected": -25.603477478027344,
+      "loss": 0.9804578423500061,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.4873877465724945,
+      "rewards/accuracies": 0.75,
+      "rewards/chosen": -0.046319298446178436,
+      "rewards/margins": 0.5485984086990356,
+      "rewards/rejected": -0.5949177145957947,
+      "step": 29,
+      "train_speed(iter/s)": 0.011239
+    },
+    {
+      "epoch": 0.07544797233574348,
+      "grad_norm": 1.7451585531234741,
+      "learning_rate": 0.0001,
+      "logits/chosen": -0.36992424726486206,
+      "logits/rejected": -0.45540890097618103,
+      "logps/chosen": -6.483048439025879,
+      "logps/rejected": -18.744525909423828,
+      "loss": 0.9453181624412537,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.37866073846817017,
+      "rewards/accuracies": 0.75,
+      "rewards/chosen": 0.0520947091281414,
+      "rewards/margins": 0.33587026596069336,
+      "rewards/rejected": -0.28377556800842285,
+      "step": 30,
+      "train_speed(iter/s)": 0.01124
+    },
+    {
+      "epoch": 0.07796290474693493,
+      "grad_norm": 2.8843863010406494,
+      "learning_rate": 0.00010333333333333334,
+      "logits/chosen": -0.3973531723022461,
+      "logits/rejected": -0.4880366325378418,
+      "logps/chosen": -6.5046491622924805,
+      "logps/rejected": -18.793209075927734,
+      "loss": 1.0352380275726318,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.514582097530365,
+      "rewards/accuracies": 0.8125,
+      "rewards/chosen": 0.06333249807357788,
+      "rewards/margins": 0.5268359780311584,
+      "rewards/rejected": -0.4635034203529358,
+      "step": 31,
+      "train_speed(iter/s)": 0.011242
+    },
+    {
+      "epoch": 0.08047783715812637,
+      "grad_norm": 1.9934616088867188,
+      "learning_rate": 0.00010666666666666667,
+      "logits/chosen": -0.38435813784599304,
+      "logits/rejected": -0.4866703450679779,
+      "logps/chosen": -5.984093189239502,
+      "logps/rejected": -28.37721061706543,
+      "loss": 0.9228850603103638,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.3919341564178467,
+      "rewards/accuracies": 0.75,
+      "rewards/chosen": 0.08762034773826599,
+      "rewards/margins": 0.4990919232368469,
+      "rewards/rejected": -0.41147157549858093,
+      "step": 32,
+      "train_speed(iter/s)": 0.011242
+    },
+    {
+      "epoch": 0.08299276956931782,
+      "grad_norm": 2.605156421661377,
+      "learning_rate": 0.00011000000000000002,
+      "logits/chosen": -0.3683161437511444,
+      "logits/rejected": -0.4800977110862732,
+      "logps/chosen": -6.79887056350708,
+      "logps/rejected": -18.521530151367188,
+      "loss": 0.9287618398666382,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.4155515730381012,
+      "rewards/accuracies": 0.875,
+      "rewards/chosen": 0.062238909304142,
+      "rewards/margins": 0.4814634323120117,
+      "rewards/rejected": -0.41922450065612793,
+      "step": 33,
+      "train_speed(iter/s)": 0.011243
+    },
+    {
+      "epoch": 0.08550770198050928,
+      "grad_norm": 2.1593198776245117,
+      "learning_rate": 0.00011333333333333334,
+      "logits/chosen": -0.3462904989719391,
+      "logits/rejected": -0.4482634663581848,
+      "logps/chosen": -4.484028339385986,
+      "logps/rejected": -22.155223846435547,
+      "loss": 0.8287199139595032,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.3564324378967285,
+      "rewards/accuracies": 0.8125,
+      "rewards/chosen": -0.011370105668902397,
+      "rewards/margins": 0.8015316724777222,
+      "rewards/rejected": -0.8129018545150757,
+      "step": 34,
+      "train_speed(iter/s)": 0.011244
+    },
+    {
+      "epoch": 0.08802263439170073,
+      "grad_norm": 2.1223862171173096,
+      "learning_rate": 0.00011666666666666668,
+      "logits/chosen": -0.23746490478515625,
+      "logits/rejected": -0.4123544991016388,
+      "logps/chosen": -4.062009334564209,
+      "logps/rejected": -30.214771270751953,
+      "loss": 0.8381334543228149,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.45768678188323975,
+      "rewards/accuracies": 0.84375,
+      "rewards/chosen": 0.06588438153266907,
+      "rewards/margins": 1.1650195121765137,
+      "rewards/rejected": -1.0991352796554565,
+      "step": 35,
+      "train_speed(iter/s)": 0.011244
+    },
+    {
+      "epoch": 0.09053756680289217,
+      "grad_norm": 1.7519793510437012,
+      "learning_rate": 0.00012,
+      "logits/chosen": -0.274143248796463,
+      "logits/rejected": -0.4072067439556122,
+      "logps/chosen": -7.12678861618042,
+      "logps/rejected": -29.586929321289062,
+      "loss": 0.7818687558174133,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.38336753845214844,
+      "rewards/accuracies": 0.84375,
+      "rewards/chosen": 0.030900314450263977,
+      "rewards/margins": 1.0601840019226074,
+      "rewards/rejected": -1.0292836427688599,
+      "step": 36,
+      "train_speed(iter/s)": 0.011244
+    },
+    {
+      "epoch": 0.09305249921408362,
+      "grad_norm": 2.2621781826019287,
+      "learning_rate": 0.00012333333333333334,
+      "logits/chosen": -0.30612003803253174,
+      "logits/rejected": -0.44930776953697205,
+      "logps/chosen": -5.637537956237793,
+      "logps/rejected": -30.86564064025879,
+      "loss": 0.7518173456192017,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.4073091149330139,
+      "rewards/accuracies": 0.875,
+      "rewards/chosen": -0.1094760149717331,
+      "rewards/margins": 1.2509604692459106,
+      "rewards/rejected": -1.3604364395141602,
+      "step": 37,
+      "train_speed(iter/s)": 0.011245
+    },
+    {
+      "epoch": 0.09556743162527508,
+      "grad_norm": 2.6005046367645264,
+      "learning_rate": 0.00012666666666666666,
+      "logits/chosen": -0.36841830611228943,
+      "logits/rejected": -0.44501256942749023,
+      "logps/chosen": -8.786764144897461,
+      "logps/rejected": -21.693777084350586,
+      "loss": 0.967311441898346,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.5350124835968018,
+      "rewards/accuracies": 0.84375,
+      "rewards/chosen": -0.05166015028953552,
+      "rewards/margins": 0.9184513092041016,
+      "rewards/rejected": -0.9701113700866699,
+      "step": 38,
+      "train_speed(iter/s)": 0.011247
+    },
+    {
+      "epoch": 0.09808236403646652,
+      "grad_norm": 7.679723262786865,
+      "learning_rate": 0.00013000000000000002,
+      "logits/chosen": -0.3933795094490051,
+      "logits/rejected": -0.4303474426269531,
+      "logps/chosen": -9.630545616149902,
+      "logps/rejected": -25.0108642578125,
+      "loss": 1.1940371990203857,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.6422770023345947,
+      "rewards/accuracies": 0.65625,
+      "rewards/chosen": -0.19723013043403625,
+      "rewards/margins": 0.8150386214256287,
+      "rewards/rejected": -1.0122687816619873,
+      "step": 39,
+      "train_speed(iter/s)": 0.011247
+    },
+    {
+      "epoch": 0.10059729644765797,
+      "grad_norm": 2.497746229171753,
+      "learning_rate": 0.00013333333333333334,
+      "logits/chosen": -0.3561173677444458,
+      "logits/rejected": -0.4964173436164856,
+      "logps/chosen": -6.288295269012451,
+      "logps/rejected": -28.722339630126953,
+      "loss": 0.7734813690185547,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.35979557037353516,
+      "rewards/accuracies": 0.84375,
+      "rewards/chosen": 0.020053118467330933,
+      "rewards/margins": 1.0713183879852295,
+      "rewards/rejected": -1.0512654781341553,
+      "step": 40,
+      "train_speed(iter/s)": 0.011248
+    },
+    {
+      "epoch": 0.10311222885884942,
+      "grad_norm": 3.900786876678467,
+      "learning_rate": 0.00013666666666666666,
+      "logits/chosen": -0.30438148975372314,
+      "logits/rejected": -0.4660162031650543,
+      "logps/chosen": -9.80126953125,
+      "logps/rejected": -22.89883804321289,
+      "loss": 1.0617499351501465,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.5348615646362305,
+      "rewards/accuracies": 0.71875,
+      "rewards/chosen": 0.037312038242816925,
+      "rewards/margins": 0.48836010694503784,
+      "rewards/rejected": -0.4510480761528015,
+      "step": 41,
+      "train_speed(iter/s)": 0.011248
+    },
+    {
+      "epoch": 0.10562716127004086,
+      "grad_norm": 4.212996006011963,
+      "learning_rate": 0.00014,
+      "logits/chosen": -0.37392380833625793,
+      "logits/rejected": -0.4811669588088989,
+      "logps/chosen": -8.54138469696045,
+      "logps/rejected": -24.88803482055664,
+      "loss": 0.9449828267097473,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.40984830260276794,
+      "rewards/accuracies": 0.71875,
+      "rewards/chosen": -0.078499935567379,
+      "rewards/margins": 0.6149317026138306,
+      "rewards/rejected": -0.6934316754341125,
+      "step": 42,
+      "train_speed(iter/s)": 0.011249
+    },
+    {
+      "epoch": 0.10814209368123232,
+      "grad_norm": 2.7297940254211426,
+      "learning_rate": 0.00014333333333333334,
+      "logits/chosen": -0.22235050797462463,
+      "logits/rejected": -0.44656193256378174,
+      "logps/chosen": -3.8628368377685547,
+      "logps/rejected": -29.070890426635742,
+      "loss": 0.6864454746246338,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.3404179513454437,
+      "rewards/accuracies": 0.9375,
+      "rewards/chosen": 0.10101112723350525,
+      "rewards/margins": 1.1784940958023071,
+      "rewards/rejected": -1.0774829387664795,
+      "step": 43,
+      "train_speed(iter/s)": 0.011247
+    },
+    {
+      "epoch": 0.11065702609242377,
+      "grad_norm": 4.357929229736328,
+      "learning_rate": 0.00014666666666666666,
+      "logits/chosen": -0.3152369558811188,
+      "logits/rejected": -0.4675106406211853,
+      "logps/chosen": -8.47473430633545,
+      "logps/rejected": -35.49666976928711,
+      "loss": 0.8841890096664429,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.543793261051178,
+      "rewards/accuracies": 0.8125,
+      "rewards/chosen": -0.2599925100803375,
+      "rewards/margins": 1.5472975969314575,
+      "rewards/rejected": -1.8072898387908936,
+      "step": 44,
+      "train_speed(iter/s)": 0.011248
+    },
+    {
+      "epoch": 0.11317195850361522,
+      "grad_norm": 4.053813934326172,
+      "learning_rate": 0.00015000000000000001,
+      "logits/chosen": -0.47932106256484985,
+      "logits/rejected": -0.5357992053031921,
+      "logps/chosen": -10.139131546020508,
+      "logps/rejected": -29.30261993408203,
+      "loss": 1.0090879201889038,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.6002873182296753,
+      "rewards/accuracies": 0.75,
+      "rewards/chosen": -0.2646394371986389,
+      "rewards/margins": 1.3857942819595337,
+      "rewards/rejected": -1.6504336595535278,
+      "step": 45,
+      "train_speed(iter/s)": 0.01125
+    },
+    {
+      "epoch": 0.11568689091480666,
+      "grad_norm": 3.4071342945098877,
+      "learning_rate": 0.00015333333333333334,
+      "logits/chosen": -0.35965389013290405,
+      "logits/rejected": -0.4886631965637207,
+      "logps/chosen": -6.492125988006592,
+      "logps/rejected": -33.90357971191406,
+      "loss": 0.9477238655090332,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.5519906878471375,
+      "rewards/accuracies": 0.8125,
+      "rewards/chosen": -0.13393068313598633,
+      "rewards/margins": 1.3498430252075195,
+      "rewards/rejected": -1.4837737083435059,
+      "step": 46,
+      "train_speed(iter/s)": 0.011251
+    },
+    {
+      "epoch": 0.11820182332599811,
+      "grad_norm": 1.5676898956298828,
+      "learning_rate": 0.00015666666666666666,
+      "logits/chosen": -0.3339039087295532,
+      "logits/rejected": -0.5405474305152893,
+      "logps/chosen": -5.18754768371582,
+      "logps/rejected": -34.61940383911133,
+      "loss": 0.5877978205680847,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.26233971118927,
+      "rewards/accuracies": 0.84375,
+      "rewards/chosen": 0.04858784377574921,
+      "rewards/margins": 1.4838428497314453,
+      "rewards/rejected": -1.4352549314498901,
+      "step": 47,
+      "train_speed(iter/s)": 0.011252
+    },
+    {
+      "epoch": 0.12071675573718957,
+      "grad_norm": 1.7600055932998657,
+      "learning_rate": 0.00016,
+      "logits/chosen": -0.38554954528808594,
+      "logits/rejected": -0.5349220633506775,
+      "logps/chosen": -5.35024356842041,
+      "logps/rejected": -30.936561584472656,
+      "loss": 0.7445188760757446,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.3484676480293274,
+      "rewards/accuracies": 0.84375,
+      "rewards/chosen": -0.04428477957844734,
+      "rewards/margins": 1.158116340637207,
+      "rewards/rejected": -1.2024011611938477,
+      "step": 48,
+      "train_speed(iter/s)": 0.011253
+    },
+    {
+      "epoch": 0.12323168814838101,
+      "grad_norm": 1.8286348581314087,
+      "learning_rate": 0.00016333333333333334,
+      "logits/chosen": -0.3060781955718994,
+      "logits/rejected": -0.5333737134933472,
+      "logps/chosen": -4.790343761444092,
+      "logps/rejected": -32.540409088134766,
+      "loss": 0.6686277985572815,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.3360018730163574,
+      "rewards/accuracies": 0.9375,
+      "rewards/chosen": 0.0967080146074295,
+      "rewards/margins": 1.285821795463562,
+      "rewards/rejected": -1.1891138553619385,
+      "step": 49,
+      "train_speed(iter/s)": 0.011254
+    },
+    {
+      "epoch": 0.12574662055957248,
+      "grad_norm": 1.8983184099197388,
+      "learning_rate": 0.0001666666666666667,
+      "logits/chosen": -0.22582519054412842,
+      "logits/rejected": -0.47035256028175354,
+      "logps/chosen": -6.48522424697876,
+      "logps/rejected": -32.73487091064453,
+      "loss": 0.8076327443122864,
+      "memory(GiB)": 40.85,
+      "nll_loss": 0.3926582336425781,
+      "rewards/accuracies": 0.78125,
+      "rewards/chosen": 0.01478707417845726,
+      "rewards/margins": 1.167083501815796,
+      "rewards/rejected": -1.1522964239120483,
+      "step": 50,
+      "train_speed(iter/s)": 0.011253
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1191,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "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": 2.7419712375750656e+16,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac2c834cf7926ef8f51413f25d539d89c2f75f1f045b50c056e31fed9717388d
+size 8696

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 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}")
+
+    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(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)