Training in progress, step 195, checkpoint

checkpoint-195/README.md

@@ -0,0 +1,202 @@
+---
+base_model: Qwen/Qwen2.5-Coder-7B
+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 0.13.2

checkpoint-195/adapter_config.json

@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-Coder-7B",
+  "bias": "none",
+  "fan_in_fan_out": null,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 256,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 256,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "k_proj",
+    "v_proj",
+    "gate_proj",
+    "up_proj",
+    "q_proj",
+    "down_proj",
+    "o_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}

checkpoint-195/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c9c3ef46e27aeb515360bc06713aa2cf1dfeb89a4d861ec3e9c0924f088ed98e
+size 1291899552

checkpoint-195/added_tokens.json

@@ -0,0 +1,24 @@
+{
+  "</tool_call>": 151658,
+  "<tool_call>": 151657,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

checkpoint-195/global_step195/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:57b48d544e37cb8f6a9deb837cbf3f9f64e98e15e618932fa3d61d2ddb00e414
+size 486977360

checkpoint-195/global_step195/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dc8941969e8274b21584ac6d4ab710657ffaf65dc9d7baadc4a29da08f8b8d09
+size 486976656

checkpoint-195/global_step195/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:af50b8e5f90f1fa15bce38a0b4805cdf1cbe4c55646ea1ab437b0cb9e4c7d8d0
+size 486977360

checkpoint-195/global_step195/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3d64e2faca55af276b51a47ec1e0240cb5b98565bc5b3db3484df67cebe130cf
+size 486976720

checkpoint-195/global_step195/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b15d892f23de2115183d6374efc648b222eb737afa3d0a092a717627173937e7
+size 486977424

checkpoint-195/global_step195/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:14bbf5a653426282e9ba8bc775aa6bfc3f6cb31bc09e94089616c906976051e9
+size 486976720

checkpoint-195/global_step195/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:27d3f106c07b1649455bb4a9ae87c0e8b755069fb95bea378da19e60f5e25715
+size 486977424

checkpoint-195/global_step195/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61d4ffb539dd874a74341533ae23ff4edc522c3f42bde6d8e96fd97f375b2168
+size 486976720

checkpoint-195/global_step195/mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c929c2525d4cbf367be2d30318b7fe7a71ea0e24aef69819a278df79c213078
+size 1292148716

checkpoint-195/latest

@@ -0,0 +1 @@
+global_step195

checkpoint-195/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:03f72f90ab19248bdea4a4ae1953ed67bb5da9f39335ec1c6db89ef893d52777
+size 15920

checkpoint-195/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bf2a228c41baa84a37172e80aad237e8c7521d91e521a5a51a5f233a0e7df624
+size 15920

checkpoint-195/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bfa7133ecdeca5d779ffe50a45427618ecf4c861958d00b267df79d6256509fb
+size 15920

checkpoint-195/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:957be4c07ebdd63a314c05eab20942339322508db9d071b89aa0fe5f397ed247
+size 15920

checkpoint-195/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a5e6d3080b6840dbe1ca1ec94c47153763ee24dad93704902711661985b4df9d
+size 15920

checkpoint-195/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e356cfc56676798532a89a26df25eb5fb305dd26c86d77680b0d39a84153a6fa
+size 15920

checkpoint-195/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:afeb812dcd886cf460e913276e3bfa81d2292d8ef1a2579d15010f8eaaf85a58
+size 15920

checkpoint-195/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a1a9fb13a383ad1a208641f8c1c88b2ce07f8f1152158a56c61329effce29053
+size 15920

checkpoint-195/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:de04214c9d0b99f3fd922d6a66a496b9df631fd1fda8264ba530a12d8fb75e72
+size 1064

checkpoint-195/special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

checkpoint-195/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c5ae00e602b8860cbd784ba82a8aa14e8feecec692e7076590d014d7b7fdafa
+size 11421896

checkpoint-195/tokenizer_config.json

@@ -0,0 +1,207 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "chat_template": "{%- if tools %}\n    {{- '<|im_start|>system\\n' }}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- messages[0]['content'] }}\n    {%- else %}\n        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}\n    {%- endif %}\n    {{- \"\\n\\n# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n    {%- for tool in tools %}\n        {{- \"\\n\" }}\n        {{- tool | tojson }}\n    {%- endfor %}\n    {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- '<|im_start|>system\\n' + messages[0]['content'] + '<|im_end|>\\n' }}\n    {%- else %}\n        {{- '<|im_start|>system\\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- for message in messages %}\n    {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) or (message.role == \"assistant\" and not message.tool_calls) %}\n        {{- '<|im_start|>' + message.role + '\\n' + message.content + '<|im_end|>' + '\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {{- '<|im_start|>' + message.role }}\n        {%- if message.content %}\n            {{- '\\n' + message.content }}\n        {%- endif %}\n        {%- for tool_call in message.tool_calls %}\n            {%- if tool_call.function is defined %}\n                {%- set tool_call = tool_call.function %}\n            {%- endif %}\n            {{- '\\n<tool_call>\\n{\"name\": \"' }}\n            {{- tool_call.name }}\n            {{- '\", \"arguments\": ' }}\n            {{- tool_call.arguments | tojson }}\n            {{- '}\\n</tool_call>' }}\n        {%- endfor %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != \"tool\") %}\n            {{- '<|im_start|>user' }}\n        {%- endif %}\n        {{- '\\n<tool_response>\\n' }}\n        {{- message.content }}\n        {{- '\\n</tool_response>' }}\n        {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n            {{- '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

checkpoint-195/trainer_state.json

@@ -0,0 +1,1398 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.2503209242618742,
+  "eval_steps": 500,
+  "global_step": 195,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0012836970474967907,
+      "grad_norm": 0.9255548715591431,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 2.8985,
+      "step": 1
+    },
+    {
+      "epoch": 0.0025673940949935813,
+      "grad_norm": 0.7692601680755615,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 2.9774,
+      "step": 2
+    },
+    {
+      "epoch": 0.0038510911424903724,
+      "grad_norm": 0.7884671092033386,
+      "learning_rate": 1.5e-06,
+      "loss": 2.9898,
+      "step": 3
+    },
+    {
+      "epoch": 0.005134788189987163,
+      "grad_norm": 0.8319393396377563,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 2.9573,
+      "step": 4
+    },
+    {
+      "epoch": 0.006418485237483954,
+      "grad_norm": 0.822285532951355,
+      "learning_rate": 2.5e-06,
+      "loss": 2.9316,
+      "step": 5
+    },
+    {
+      "epoch": 0.007702182284980745,
+      "grad_norm": 0.7564650774002075,
+      "learning_rate": 3e-06,
+      "loss": 2.9537,
+      "step": 6
+    },
+    {
+      "epoch": 0.008985879332477536,
+      "grad_norm": 0.9220781922340393,
+      "learning_rate": 3.5000000000000004e-06,
+      "loss": 2.9826,
+      "step": 7
+    },
+    {
+      "epoch": 0.010269576379974325,
+      "grad_norm": 0.7977064847946167,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 2.8548,
+      "step": 8
+    },
+    {
+      "epoch": 0.011553273427471117,
+      "grad_norm": 0.6889916658401489,
+      "learning_rate": 4.5e-06,
+      "loss": 2.9936,
+      "step": 9
+    },
+    {
+      "epoch": 0.012836970474967908,
+      "grad_norm": 0.9777728915214539,
+      "learning_rate": 5e-06,
+      "loss": 2.9578,
+      "step": 10
+    },
+    {
+      "epoch": 0.014120667522464698,
+      "grad_norm": 0.8187949061393738,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 2.9442,
+      "step": 11
+    },
+    {
+      "epoch": 0.01540436456996149,
+      "grad_norm": 0.7016908526420593,
+      "learning_rate": 6e-06,
+      "loss": 2.9299,
+      "step": 12
+    },
+    {
+      "epoch": 0.01668806161745828,
+      "grad_norm": 0.6974747180938721,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 2.9699,
+      "step": 13
+    },
+    {
+      "epoch": 0.01797175866495507,
+      "grad_norm": 0.8423139452934265,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 2.8706,
+      "step": 14
+    },
+    {
+      "epoch": 0.019255455712451863,
+      "grad_norm": 0.8017705082893372,
+      "learning_rate": 7.5e-06,
+      "loss": 2.8443,
+      "step": 15
+    },
+    {
+      "epoch": 0.02053915275994865,
+      "grad_norm": 0.9362208247184753,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 2.872,
+      "step": 16
+    },
+    {
+      "epoch": 0.021822849807445442,
+      "grad_norm": 1.013128638267517,
+      "learning_rate": 8.500000000000002e-06,
+      "loss": 2.797,
+      "step": 17
+    },
+    {
+      "epoch": 0.023106546854942234,
+      "grad_norm": 1.255325436592102,
+      "learning_rate": 9e-06,
+      "loss": 2.8109,
+      "step": 18
+    },
+    {
+      "epoch": 0.024390243902439025,
+      "grad_norm": 1.1081339120864868,
+      "learning_rate": 9.5e-06,
+      "loss": 2.6742,
+      "step": 19
+    },
+    {
+      "epoch": 0.025673940949935817,
+      "grad_norm": 0.8878622651100159,
+      "learning_rate": 1e-05,
+      "loss": 2.7251,
+      "step": 20
+    },
+    {
+      "epoch": 0.026957637997432605,
+      "grad_norm": 0.894791305065155,
+      "learning_rate": 1.05e-05,
+      "loss": 2.7127,
+      "step": 21
+    },
+    {
+      "epoch": 0.028241335044929396,
+      "grad_norm": 0.6742448806762695,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 2.6842,
+      "step": 22
+    },
+    {
+      "epoch": 0.029525032092426188,
+      "grad_norm": 0.6250098943710327,
+      "learning_rate": 1.1500000000000002e-05,
+      "loss": 2.6476,
+      "step": 23
+    },
+    {
+      "epoch": 0.03080872913992298,
+      "grad_norm": 0.6331678032875061,
+      "learning_rate": 1.2e-05,
+      "loss": 2.6339,
+      "step": 24
+    },
+    {
+      "epoch": 0.03209242618741977,
+      "grad_norm": 0.4726584255695343,
+      "learning_rate": 1.25e-05,
+      "loss": 2.5994,
+      "step": 25
+    },
+    {
+      "epoch": 0.03337612323491656,
+      "grad_norm": 0.46077489852905273,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 2.5545,
+      "step": 26
+    },
+    {
+      "epoch": 0.03465982028241335,
+      "grad_norm": 0.5746111273765564,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 2.5542,
+      "step": 27
+    },
+    {
+      "epoch": 0.03594351732991014,
+      "grad_norm": 0.47136253118515015,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 2.6138,
+      "step": 28
+    },
+    {
+      "epoch": 0.037227214377406934,
+      "grad_norm": 0.5951219797134399,
+      "learning_rate": 1.45e-05,
+      "loss": 2.5941,
+      "step": 29
+    },
+    {
+      "epoch": 0.038510911424903725,
+      "grad_norm": 0.48593708872795105,
+      "learning_rate": 1.5e-05,
+      "loss": 2.5751,
+      "step": 30
+    },
+    {
+      "epoch": 0.03979460847240052,
+      "grad_norm": 0.482264906167984,
+      "learning_rate": 1.55e-05,
+      "loss": 2.5689,
+      "step": 31
+    },
+    {
+      "epoch": 0.0410783055198973,
+      "grad_norm": 0.4937015175819397,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 2.5559,
+      "step": 32
+    },
+    {
+      "epoch": 0.04236200256739409,
+      "grad_norm": 0.5136978030204773,
+      "learning_rate": 1.65e-05,
+      "loss": 2.5274,
+      "step": 33
+    },
+    {
+      "epoch": 0.043645699614890884,
+      "grad_norm": 0.40434661507606506,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 2.4391,
+      "step": 34
+    },
+    {
+      "epoch": 0.044929396662387676,
+      "grad_norm": 0.49542951583862305,
+      "learning_rate": 1.75e-05,
+      "loss": 2.5738,
+      "step": 35
+    },
+    {
+      "epoch": 0.04621309370988447,
+      "grad_norm": 0.4381186366081238,
+      "learning_rate": 1.8e-05,
+      "loss": 2.5095,
+      "step": 36
+    },
+    {
+      "epoch": 0.04749679075738126,
+      "grad_norm": 0.486103892326355,
+      "learning_rate": 1.85e-05,
+      "loss": 2.6173,
+      "step": 37
+    },
+    {
+      "epoch": 0.04878048780487805,
+      "grad_norm": 0.4313197135925293,
+      "learning_rate": 1.9e-05,
+      "loss": 2.4951,
+      "step": 38
+    },
+    {
+      "epoch": 0.05006418485237484,
+      "grad_norm": 0.4211004078388214,
+      "learning_rate": 1.9500000000000003e-05,
+      "loss": 2.5469,
+      "step": 39
+    },
+    {
+      "epoch": 0.051347881899871634,
+      "grad_norm": 0.44780072569847107,
+      "learning_rate": 2e-05,
+      "loss": 2.4437,
+      "step": 40
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 0.44611668586730957,
+      "learning_rate": 2.05e-05,
+      "loss": 2.4795,
+      "step": 41
+    },
+    {
+      "epoch": 0.05391527599486521,
+      "grad_norm": 0.4598286747932434,
+      "learning_rate": 2.1e-05,
+      "loss": 2.4786,
+      "step": 42
+    },
+    {
+      "epoch": 0.055198973042362,
+      "grad_norm": 0.4416978061199188,
+      "learning_rate": 2.15e-05,
+      "loss": 2.453,
+      "step": 43
+    },
+    {
+      "epoch": 0.05648267008985879,
+      "grad_norm": 0.4136359989643097,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 2.4975,
+      "step": 44
+    },
+    {
+      "epoch": 0.057766367137355584,
+      "grad_norm": 0.44032955169677734,
+      "learning_rate": 2.25e-05,
+      "loss": 2.411,
+      "step": 45
+    },
+    {
+      "epoch": 0.059050064184852376,
+      "grad_norm": 0.49505728483200073,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 2.5049,
+      "step": 46
+    },
+    {
+      "epoch": 0.06033376123234917,
+      "grad_norm": 0.43698814511299133,
+      "learning_rate": 2.35e-05,
+      "loss": 2.5828,
+      "step": 47
+    },
+    {
+      "epoch": 0.06161745827984596,
+      "grad_norm": 0.44550350308418274,
+      "learning_rate": 2.4e-05,
+      "loss": 2.4326,
+      "step": 48
+    },
+    {
+      "epoch": 0.06290115532734275,
+      "grad_norm": 0.38959425687789917,
+      "learning_rate": 2.45e-05,
+      "loss": 2.4684,
+      "step": 49
+    },
+    {
+      "epoch": 0.06418485237483953,
+      "grad_norm": 0.4324244260787964,
+      "learning_rate": 2.5e-05,
+      "loss": 2.4496,
+      "step": 50
+    },
+    {
+      "epoch": 0.06546854942233633,
+      "grad_norm": 0.4213118553161621,
+      "learning_rate": 2.5500000000000003e-05,
+      "loss": 2.4742,
+      "step": 51
+    },
+    {
+      "epoch": 0.06675224646983312,
+      "grad_norm": 0.5279268622398376,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 2.4222,
+      "step": 52
+    },
+    {
+      "epoch": 0.06803594351732992,
+      "grad_norm": 0.40476322174072266,
+      "learning_rate": 2.6500000000000004e-05,
+      "loss": 2.4224,
+      "step": 53
+    },
+    {
+      "epoch": 0.0693196405648267,
+      "grad_norm": 0.6938806176185608,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 2.4194,
+      "step": 54
+    },
+    {
+      "epoch": 0.07060333761232349,
+      "grad_norm": 0.43899399042129517,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 2.4288,
+      "step": 55
+    },
+    {
+      "epoch": 0.07188703465982028,
+      "grad_norm": 0.3968575596809387,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 2.3968,
+      "step": 56
+    },
+    {
+      "epoch": 0.07317073170731707,
+      "grad_norm": 0.6113290786743164,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 2.3924,
+      "step": 57
+    },
+    {
+      "epoch": 0.07445442875481387,
+      "grad_norm": 0.35704493522644043,
+      "learning_rate": 2.9e-05,
+      "loss": 2.4004,
+      "step": 58
+    },
+    {
+      "epoch": 0.07573812580231065,
+      "grad_norm": 0.3809000551700592,
+      "learning_rate": 2.95e-05,
+      "loss": 2.4303,
+      "step": 59
+    },
+    {
+      "epoch": 0.07702182284980745,
+      "grad_norm": 0.4394189715385437,
+      "learning_rate": 3e-05,
+      "loss": 2.4716,
+      "step": 60
+    },
+    {
+      "epoch": 0.07830551989730423,
+      "grad_norm": 0.4325893521308899,
+      "learning_rate": 3.05e-05,
+      "loss": 2.4715,
+      "step": 61
+    },
+    {
+      "epoch": 0.07958921694480103,
+      "grad_norm": 0.3560517728328705,
+      "learning_rate": 3.1e-05,
+      "loss": 2.4745,
+      "step": 62
+    },
+    {
+      "epoch": 0.08087291399229782,
+      "grad_norm": 0.37922918796539307,
+      "learning_rate": 3.15e-05,
+      "loss": 2.4789,
+      "step": 63
+    },
+    {
+      "epoch": 0.0821566110397946,
+      "grad_norm": 0.3665093779563904,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 2.4829,
+      "step": 64
+    },
+    {
+      "epoch": 0.0834403080872914,
+      "grad_norm": 0.38106483221054077,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 2.4203,
+      "step": 65
+    },
+    {
+      "epoch": 0.08472400513478819,
+      "grad_norm": 0.5111002326011658,
+      "learning_rate": 3.3e-05,
+      "loss": 2.3302,
+      "step": 66
+    },
+    {
+      "epoch": 0.08600770218228498,
+      "grad_norm": 0.37738922238349915,
+      "learning_rate": 3.35e-05,
+      "loss": 2.4991,
+      "step": 67
+    },
+    {
+      "epoch": 0.08729139922978177,
+      "grad_norm": 0.34703031182289124,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 2.4224,
+      "step": 68
+    },
+    {
+      "epoch": 0.08857509627727857,
+      "grad_norm": 0.48427700996398926,
+      "learning_rate": 3.45e-05,
+      "loss": 2.3946,
+      "step": 69
+    },
+    {
+      "epoch": 0.08985879332477535,
+      "grad_norm": 0.360221266746521,
+      "learning_rate": 3.5e-05,
+      "loss": 2.4853,
+      "step": 70
+    },
+    {
+      "epoch": 0.09114249037227215,
+      "grad_norm": 0.4006412625312805,
+      "learning_rate": 3.55e-05,
+      "loss": 2.4654,
+      "step": 71
+    },
+    {
+      "epoch": 0.09242618741976893,
+      "grad_norm": 0.3662618398666382,
+      "learning_rate": 3.6e-05,
+      "loss": 2.434,
+      "step": 72
+    },
+    {
+      "epoch": 0.09370988446726572,
+      "grad_norm": 0.3694933354854584,
+      "learning_rate": 3.65e-05,
+      "loss": 2.4194,
+      "step": 73
+    },
+    {
+      "epoch": 0.09499358151476252,
+      "grad_norm": 0.34268808364868164,
+      "learning_rate": 3.7e-05,
+      "loss": 2.4922,
+      "step": 74
+    },
+    {
+      "epoch": 0.0962772785622593,
+      "grad_norm": 0.3664718270301819,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 2.4083,
+      "step": 75
+    },
+    {
+      "epoch": 0.0975609756097561,
+      "grad_norm": 0.3909706473350525,
+      "learning_rate": 3.8e-05,
+      "loss": 2.4546,
+      "step": 76
+    },
+    {
+      "epoch": 0.09884467265725289,
+      "grad_norm": 0.36276674270629883,
+      "learning_rate": 3.85e-05,
+      "loss": 2.5313,
+      "step": 77
+    },
+    {
+      "epoch": 0.10012836970474968,
+      "grad_norm": 0.34822535514831543,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 2.4833,
+      "step": 78
+    },
+    {
+      "epoch": 0.10141206675224647,
+      "grad_norm": 0.37480583786964417,
+      "learning_rate": 3.9500000000000005e-05,
+      "loss": 2.449,
+      "step": 79
+    },
+    {
+      "epoch": 0.10269576379974327,
+      "grad_norm": 0.3415388762950897,
+      "learning_rate": 4e-05,
+      "loss": 2.3953,
+      "step": 80
+    },
+    {
+      "epoch": 0.10397946084724005,
+      "grad_norm": 0.3487205505371094,
+      "learning_rate": 4.05e-05,
+      "loss": 2.4469,
+      "step": 81
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 0.5083756446838379,
+      "learning_rate": 4.1e-05,
+      "loss": 2.3423,
+      "step": 82
+    },
+    {
+      "epoch": 0.10654685494223363,
+      "grad_norm": 0.3747817575931549,
+      "learning_rate": 4.15e-05,
+      "loss": 2.3919,
+      "step": 83
+    },
+    {
+      "epoch": 0.10783055198973042,
+      "grad_norm": 0.39472389221191406,
+      "learning_rate": 4.2e-05,
+      "loss": 2.5431,
+      "step": 84
+    },
+    {
+      "epoch": 0.10911424903722722,
+      "grad_norm": 0.36542952060699463,
+      "learning_rate": 4.25e-05,
+      "loss": 2.3858,
+      "step": 85
+    },
+    {
+      "epoch": 0.110397946084724,
+      "grad_norm": 0.34629878401756287,
+      "learning_rate": 4.3e-05,
+      "loss": 2.3528,
+      "step": 86
+    },
+    {
+      "epoch": 0.1116816431322208,
+      "grad_norm": 0.36755290627479553,
+      "learning_rate": 4.35e-05,
+      "loss": 2.5329,
+      "step": 87
+    },
+    {
+      "epoch": 0.11296534017971759,
+      "grad_norm": 0.3599033057689667,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 2.4971,
+      "step": 88
+    },
+    {
+      "epoch": 0.11424903722721438,
+      "grad_norm": 0.3730204701423645,
+      "learning_rate": 4.4500000000000004e-05,
+      "loss": 2.4082,
+      "step": 89
+    },
+    {
+      "epoch": 0.11553273427471117,
+      "grad_norm": 0.3773551881313324,
+      "learning_rate": 4.5e-05,
+      "loss": 2.3393,
+      "step": 90
+    },
+    {
+      "epoch": 0.11681643132220795,
+      "grad_norm": 0.36052408814430237,
+      "learning_rate": 4.55e-05,
+      "loss": 2.3799,
+      "step": 91
+    },
+    {
+      "epoch": 0.11810012836970475,
+      "grad_norm": 0.32685768604278564,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 2.4284,
+      "step": 92
+    },
+    {
+      "epoch": 0.11938382541720154,
+      "grad_norm": 0.36063292622566223,
+      "learning_rate": 4.6500000000000005e-05,
+      "loss": 2.3903,
+      "step": 93
+    },
+    {
+      "epoch": 0.12066752246469833,
+      "grad_norm": 0.3656150698661804,
+      "learning_rate": 4.7e-05,
+      "loss": 2.4378,
+      "step": 94
+    },
+    {
+      "epoch": 0.12195121951219512,
+      "grad_norm": 0.3563483655452728,
+      "learning_rate": 4.75e-05,
+      "loss": 2.4082,
+      "step": 95
+    },
+    {
+      "epoch": 0.12323491655969192,
+      "grad_norm": 0.35744163393974304,
+      "learning_rate": 4.8e-05,
+      "loss": 2.3979,
+      "step": 96
+    },
+    {
+      "epoch": 0.1245186136071887,
+      "grad_norm": 0.3400294780731201,
+      "learning_rate": 4.85e-05,
+      "loss": 2.4419,
+      "step": 97
+    },
+    {
+      "epoch": 0.1258023106546855,
+      "grad_norm": 0.3691268265247345,
+      "learning_rate": 4.9e-05,
+      "loss": 2.4671,
+      "step": 98
+    },
+    {
+      "epoch": 0.12708600770218229,
+      "grad_norm": 0.3483717739582062,
+      "learning_rate": 4.9500000000000004e-05,
+      "loss": 2.3947,
+      "step": 99
+    },
+    {
+      "epoch": 0.12836970474967907,
+      "grad_norm": 0.3494178354740143,
+      "learning_rate": 5e-05,
+      "loss": 2.4063,
+      "step": 100
+    },
+    {
+      "epoch": 0.12965340179717585,
+      "grad_norm": 0.3814634680747986,
+      "learning_rate": 4.9999986437272225e-05,
+      "loss": 2.4832,
+      "step": 101
+    },
+    {
+      "epoch": 0.13093709884467267,
+      "grad_norm": 0.378907710313797,
+      "learning_rate": 4.999994574910364e-05,
+      "loss": 2.3272,
+      "step": 102
+    },
+    {
+      "epoch": 0.13222079589216945,
+      "grad_norm": 0.3730032444000244,
+      "learning_rate": 4.999987793553836e-05,
+      "loss": 2.3965,
+      "step": 103
+    },
+    {
+      "epoch": 0.13350449293966624,
+      "grad_norm": 0.3670955002307892,
+      "learning_rate": 4.9999782996649994e-05,
+      "loss": 2.4565,
+      "step": 104
+    },
+    {
+      "epoch": 0.13478818998716302,
+      "grad_norm": 0.36450427770614624,
+      "learning_rate": 4.999966093254153e-05,
+      "loss": 2.5079,
+      "step": 105
+    },
+    {
+      "epoch": 0.13607188703465983,
+      "grad_norm": 0.38658779859542847,
+      "learning_rate": 4.9999511743345426e-05,
+      "loss": 2.4775,
+      "step": 106
+    },
+    {
+      "epoch": 0.13735558408215662,
+      "grad_norm": 0.33551573753356934,
+      "learning_rate": 4.999933542922354e-05,
+      "loss": 2.461,
+      "step": 107
+    },
+    {
+      "epoch": 0.1386392811296534,
+      "grad_norm": 0.3854399025440216,
+      "learning_rate": 4.999913199036719e-05,
+      "loss": 2.4075,
+      "step": 108
+    },
+    {
+      "epoch": 0.1399229781771502,
+      "grad_norm": 0.3932352364063263,
+      "learning_rate": 4.9998901426997104e-05,
+      "loss": 2.4311,
+      "step": 109
+    },
+    {
+      "epoch": 0.14120667522464697,
+      "grad_norm": 0.3379668593406677,
+      "learning_rate": 4.999864373936345e-05,
+      "loss": 2.5016,
+      "step": 110
+    },
+    {
+      "epoch": 0.14249037227214378,
+      "grad_norm": 0.32840296626091003,
+      "learning_rate": 4.9998358927745826e-05,
+      "loss": 2.3176,
+      "step": 111
+    },
+    {
+      "epoch": 0.14377406931964057,
+      "grad_norm": 0.3191027343273163,
+      "learning_rate": 4.999804699245325e-05,
+      "loss": 2.4007,
+      "step": 112
+    },
+    {
+      "epoch": 0.14505776636713735,
+      "grad_norm": 0.3358600437641144,
+      "learning_rate": 4.999770793382418e-05,
+      "loss": 2.3724,
+      "step": 113
+    },
+    {
+      "epoch": 0.14634146341463414,
+      "grad_norm": 0.335860937833786,
+      "learning_rate": 4.99973417522265e-05,
+      "loss": 2.3479,
+      "step": 114
+    },
+    {
+      "epoch": 0.14762516046213095,
+      "grad_norm": 0.34540730714797974,
+      "learning_rate": 4.999694844805753e-05,
+      "loss": 2.3675,
+      "step": 115
+    },
+    {
+      "epoch": 0.14890885750962773,
+      "grad_norm": 0.3298475742340088,
+      "learning_rate": 4.999652802174402e-05,
+      "loss": 2.3948,
+      "step": 116
+    },
+    {
+      "epoch": 0.15019255455712452,
+      "grad_norm": 0.3612127900123596,
+      "learning_rate": 4.999608047374211e-05,
+      "loss": 2.3855,
+      "step": 117
+    },
+    {
+      "epoch": 0.1514762516046213,
+      "grad_norm": 0.4185655415058136,
+      "learning_rate": 4.9995605804537426e-05,
+      "loss": 2.4015,
+      "step": 118
+    },
+    {
+      "epoch": 0.1527599486521181,
+      "grad_norm": 0.3759553134441376,
+      "learning_rate": 4.9995104014644986e-05,
+      "loss": 2.4483,
+      "step": 119
+    },
+    {
+      "epoch": 0.1540436456996149,
+      "grad_norm": 0.35989564657211304,
+      "learning_rate": 4.999457510460923e-05,
+      "loss": 2.4974,
+      "step": 120
+    },
+    {
+      "epoch": 0.15532734274711169,
+      "grad_norm": 0.3161202073097229,
+      "learning_rate": 4.999401907500405e-05,
+      "loss": 2.3712,
+      "step": 121
+    },
+    {
+      "epoch": 0.15661103979460847,
+      "grad_norm": 0.3105814456939697,
+      "learning_rate": 4.999343592643274e-05,
+      "loss": 2.4311,
+      "step": 122
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.3236968517303467,
+      "learning_rate": 4.9992825659528024e-05,
+      "loss": 2.5536,
+      "step": 123
+    },
+    {
+      "epoch": 0.15917843388960207,
+      "grad_norm": 0.3107609748840332,
+      "learning_rate": 4.9992188274952064e-05,
+      "loss": 2.3922,
+      "step": 124
+    },
+    {
+      "epoch": 0.16046213093709885,
+      "grad_norm": 0.39889928698539734,
+      "learning_rate": 4.999152377339642e-05,
+      "loss": 2.3488,
+      "step": 125
+    },
+    {
+      "epoch": 0.16174582798459564,
+      "grad_norm": 0.379323273897171,
+      "learning_rate": 4.99908321555821e-05,
+      "loss": 2.5278,
+      "step": 126
+    },
+    {
+      "epoch": 0.16302952503209242,
+      "grad_norm": 0.35020819306373596,
+      "learning_rate": 4.999011342225952e-05,
+      "loss": 2.3139,
+      "step": 127
+    },
+    {
+      "epoch": 0.1643132220795892,
+      "grad_norm": 0.34851884841918945,
+      "learning_rate": 4.998936757420851e-05,
+      "loss": 2.3495,
+      "step": 128
+    },
+    {
+      "epoch": 0.16559691912708602,
+      "grad_norm": 0.3453572392463684,
+      "learning_rate": 4.9988594612238336e-05,
+      "loss": 2.4128,
+      "step": 129
+    },
+    {
+      "epoch": 0.1668806161745828,
+      "grad_norm": 0.34227538108825684,
+      "learning_rate": 4.998779453718768e-05,
+      "loss": 2.3419,
+      "step": 130
+    },
+    {
+      "epoch": 0.1681643132220796,
+      "grad_norm": 0.3947238028049469,
+      "learning_rate": 4.998696734992462e-05,
+      "loss": 2.3941,
+      "step": 131
+    },
+    {
+      "epoch": 0.16944801026957637,
+      "grad_norm": 0.306533545255661,
+      "learning_rate": 4.998611305134669e-05,
+      "loss": 2.4645,
+      "step": 132
+    },
+    {
+      "epoch": 0.17073170731707318,
+      "grad_norm": 0.35172709822654724,
+      "learning_rate": 4.998523164238082e-05,
+      "loss": 2.4407,
+      "step": 133
+    },
+    {
+      "epoch": 0.17201540436456997,
+      "grad_norm": 0.34511688351631165,
+      "learning_rate": 4.9984323123983334e-05,
+      "loss": 2.3815,
+      "step": 134
+    },
+    {
+      "epoch": 0.17329910141206675,
+      "grad_norm": 0.33132505416870117,
+      "learning_rate": 4.9983387497140006e-05,
+      "loss": 2.3548,
+      "step": 135
+    },
+    {
+      "epoch": 0.17458279845956354,
+      "grad_norm": 0.32891082763671875,
+      "learning_rate": 4.998242476286601e-05,
+      "loss": 2.5308,
+      "step": 136
+    },
+    {
+      "epoch": 0.17586649550706032,
+      "grad_norm": 0.331152081489563,
+      "learning_rate": 4.998143492220592e-05,
+      "loss": 2.3858,
+      "step": 137
+    },
+    {
+      "epoch": 0.17715019255455713,
+      "grad_norm": 0.31813687086105347,
+      "learning_rate": 4.9980417976233735e-05,
+      "loss": 2.3136,
+      "step": 138
+    },
+    {
+      "epoch": 0.17843388960205392,
+      "grad_norm": 0.3268696069717407,
+      "learning_rate": 4.9979373926052865e-05,
+      "loss": 2.3133,
+      "step": 139
+    },
+    {
+      "epoch": 0.1797175866495507,
+      "grad_norm": 0.3389696180820465,
+      "learning_rate": 4.997830277279612e-05,
+      "loss": 2.3983,
+      "step": 140
+    },
+    {
+      "epoch": 0.1810012836970475,
+      "grad_norm": 0.3515508770942688,
+      "learning_rate": 4.997720451762572e-05,
+      "loss": 2.4848,
+      "step": 141
+    },
+    {
+      "epoch": 0.1822849807445443,
+      "grad_norm": 0.3302924335002899,
+      "learning_rate": 4.997607916173329e-05,
+      "loss": 2.3037,
+      "step": 142
+    },
+    {
+      "epoch": 0.18356867779204109,
+      "grad_norm": 0.3332863450050354,
+      "learning_rate": 4.997492670633987e-05,
+      "loss": 2.3563,
+      "step": 143
+    },
+    {
+      "epoch": 0.18485237483953787,
+      "grad_norm": 0.3398495614528656,
+      "learning_rate": 4.997374715269589e-05,
+      "loss": 2.4056,
+      "step": 144
+    },
+    {
+      "epoch": 0.18613607188703465,
+      "grad_norm": 0.3376169502735138,
+      "learning_rate": 4.9972540502081184e-05,
+      "loss": 2.3751,
+      "step": 145
+    },
+    {
+      "epoch": 0.18741976893453144,
+      "grad_norm": 0.33639830350875854,
+      "learning_rate": 4.9971306755804995e-05,
+      "loss": 2.432,
+      "step": 146
+    },
+    {
+      "epoch": 0.18870346598202825,
+      "grad_norm": 0.410265177488327,
+      "learning_rate": 4.9970045915205954e-05,
+      "loss": 2.3647,
+      "step": 147
+    },
+    {
+      "epoch": 0.18998716302952504,
+      "grad_norm": 0.31853362917900085,
+      "learning_rate": 4.99687579816521e-05,
+      "loss": 2.4224,
+      "step": 148
+    },
+    {
+      "epoch": 0.19127086007702182,
+      "grad_norm": 0.3495614230632782,
+      "learning_rate": 4.9967442956540863e-05,
+      "loss": 2.3961,
+      "step": 149
+    },
+    {
+      "epoch": 0.1925545571245186,
+      "grad_norm": 0.33301132917404175,
+      "learning_rate": 4.996610084129908e-05,
+      "loss": 2.359,
+      "step": 150
+    },
+    {
+      "epoch": 0.19383825417201542,
+      "grad_norm": 0.3186255395412445,
+      "learning_rate": 4.996473163738295e-05,
+      "loss": 2.4488,
+      "step": 151
+    },
+    {
+      "epoch": 0.1951219512195122,
+      "grad_norm": 0.3374113142490387,
+      "learning_rate": 4.996333534627809e-05,
+      "loss": 2.4019,
+      "step": 152
+    },
+    {
+      "epoch": 0.196405648267009,
+      "grad_norm": 0.33905014395713806,
+      "learning_rate": 4.996191196949952e-05,
+      "loss": 2.3272,
+      "step": 153
+    },
+    {
+      "epoch": 0.19768934531450577,
+      "grad_norm": 0.3546963036060333,
+      "learning_rate": 4.996046150859161e-05,
+      "loss": 2.4338,
+      "step": 154
+    },
+    {
+      "epoch": 0.19897304236200256,
+      "grad_norm": 0.35965755581855774,
+      "learning_rate": 4.9958983965128145e-05,
+      "loss": 2.3634,
+      "step": 155
+    },
+    {
+      "epoch": 0.20025673940949937,
+      "grad_norm": 0.3362196385860443,
+      "learning_rate": 4.995747934071229e-05,
+      "loss": 2.4457,
+      "step": 156
+    },
+    {
+      "epoch": 0.20154043645699615,
+      "grad_norm": 0.37214434146881104,
+      "learning_rate": 4.995594763697657e-05,
+      "loss": 2.3714,
+      "step": 157
+    },
+    {
+      "epoch": 0.20282413350449294,
+      "grad_norm": 0.3538600206375122,
+      "learning_rate": 4.995438885558294e-05,
+      "loss": 2.4338,
+      "step": 158
+    },
+    {
+      "epoch": 0.20410783055198972,
+      "grad_norm": 0.3853035569190979,
+      "learning_rate": 4.995280299822268e-05,
+      "loss": 2.4036,
+      "step": 159
+    },
+    {
+      "epoch": 0.20539152759948653,
+      "grad_norm": 0.35037294030189514,
+      "learning_rate": 4.9951190066616495e-05,
+      "loss": 2.3846,
+      "step": 160
+    },
+    {
+      "epoch": 0.20667522464698332,
+      "grad_norm": 0.3986567556858063,
+      "learning_rate": 4.994955006251443e-05,
+      "loss": 2.4405,
+      "step": 161
+    },
+    {
+      "epoch": 0.2079589216944801,
+      "grad_norm": 0.385637491941452,
+      "learning_rate": 4.994788298769593e-05,
+      "loss": 2.3237,
+      "step": 162
+    },
+    {
+      "epoch": 0.2092426187419769,
+      "grad_norm": 0.34437429904937744,
+      "learning_rate": 4.994618884396979e-05,
+      "loss": 2.3477,
+      "step": 163
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.34156107902526855,
+      "learning_rate": 4.99444676331742e-05,
+      "loss": 2.3576,
+      "step": 164
+    },
+    {
+      "epoch": 0.21181001283697048,
+      "grad_norm": 0.33888015151023865,
+      "learning_rate": 4.99427193571767e-05,
+      "loss": 2.311,
+      "step": 165
+    },
+    {
+      "epoch": 0.21309370988446727,
+      "grad_norm": 0.34711146354675293,
+      "learning_rate": 4.99409440178742e-05,
+      "loss": 2.293,
+      "step": 166
+    },
+    {
+      "epoch": 0.21437740693196405,
+      "grad_norm": 0.30586686730384827,
+      "learning_rate": 4.993914161719297e-05,
+      "loss": 2.479,
+      "step": 167
+    },
+    {
+      "epoch": 0.21566110397946084,
+      "grad_norm": 0.4145469069480896,
+      "learning_rate": 4.993731215708866e-05,
+      "loss": 2.2754,
+      "step": 168
+    },
+    {
+      "epoch": 0.21694480102695765,
+      "grad_norm": 0.6243994235992432,
+      "learning_rate": 4.993545563954626e-05,
+      "loss": 2.3751,
+      "step": 169
+    },
+    {
+      "epoch": 0.21822849807445444,
+      "grad_norm": 0.3301653265953064,
+      "learning_rate": 4.993357206658011e-05,
+      "loss": 2.6124,
+      "step": 170
+    },
+    {
+      "epoch": 0.21951219512195122,
+      "grad_norm": 0.337147980928421,
+      "learning_rate": 4.993166144023396e-05,
+      "loss": 2.3946,
+      "step": 171
+    },
+    {
+      "epoch": 0.220795892169448,
+      "grad_norm": 0.3796294033527374,
+      "learning_rate": 4.9929723762580835e-05,
+      "loss": 2.3314,
+      "step": 172
+    },
+    {
+      "epoch": 0.2220795892169448,
+      "grad_norm": 0.4021480679512024,
+      "learning_rate": 4.9927759035723175e-05,
+      "loss": 2.3035,
+      "step": 173
+    },
+    {
+      "epoch": 0.2233632862644416,
+      "grad_norm": 0.3196207582950592,
+      "learning_rate": 4.992576726179274e-05,
+      "loss": 2.4047,
+      "step": 174
+    },
+    {
+      "epoch": 0.2246469833119384,
+      "grad_norm": 0.31279096007347107,
+      "learning_rate": 4.992374844295064e-05,
+      "loss": 2.4409,
+      "step": 175
+    },
+    {
+      "epoch": 0.22593068035943517,
+      "grad_norm": 0.3068607747554779,
+      "learning_rate": 4.992170258138732e-05,
+      "loss": 2.4109,
+      "step": 176
+    },
+    {
+      "epoch": 0.22721437740693196,
+      "grad_norm": 0.35363098978996277,
+      "learning_rate": 4.991962967932258e-05,
+      "loss": 2.4083,
+      "step": 177
+    },
+    {
+      "epoch": 0.22849807445442877,
+      "grad_norm": 0.33908936381340027,
+      "learning_rate": 4.9917529739005574e-05,
+      "loss": 2.4129,
+      "step": 178
+    },
+    {
+      "epoch": 0.22978177150192555,
+      "grad_norm": 0.32510805130004883,
+      "learning_rate": 4.991540276271476e-05,
+      "loss": 2.3163,
+      "step": 179
+    },
+    {
+      "epoch": 0.23106546854942234,
+      "grad_norm": 0.32639047503471375,
+      "learning_rate": 4.991324875275794e-05,
+      "loss": 2.3567,
+      "step": 180
+    },
+    {
+      "epoch": 0.23234916559691912,
+      "grad_norm": 0.3410406708717346,
+      "learning_rate": 4.991106771147227e-05,
+      "loss": 2.4137,
+      "step": 181
+    },
+    {
+      "epoch": 0.2336328626444159,
+      "grad_norm": 0.3334997892379761,
+      "learning_rate": 4.990885964122421e-05,
+      "loss": 2.3994,
+      "step": 182
+    },
+    {
+      "epoch": 0.23491655969191272,
+      "grad_norm": 0.3203848600387573,
+      "learning_rate": 4.990662454440956e-05,
+      "loss": 2.4259,
+      "step": 183
+    },
+    {
+      "epoch": 0.2362002567394095,
+      "grad_norm": 0.34999069571495056,
+      "learning_rate": 4.9904362423453446e-05,
+      "loss": 2.413,
+      "step": 184
+    },
+    {
+      "epoch": 0.2374839537869063,
+      "grad_norm": 0.3413456678390503,
+      "learning_rate": 4.990207328081029e-05,
+      "loss": 2.3749,
+      "step": 185
+    },
+    {
+      "epoch": 0.23876765083440307,
+      "grad_norm": 0.3702191114425659,
+      "learning_rate": 4.989975711896388e-05,
+      "loss": 2.4309,
+      "step": 186
+    },
+    {
+      "epoch": 0.24005134788189988,
+      "grad_norm": 0.3365226089954376,
+      "learning_rate": 4.989741394042727e-05,
+      "loss": 2.3708,
+      "step": 187
+    },
+    {
+      "epoch": 0.24133504492939667,
+      "grad_norm": 0.36596447229385376,
+      "learning_rate": 4.989504374774288e-05,
+      "loss": 2.4387,
+      "step": 188
+    },
+    {
+      "epoch": 0.24261874197689345,
+      "grad_norm": 0.3280293941497803,
+      "learning_rate": 4.9892646543482377e-05,
+      "loss": 2.498,
+      "step": 189
+    },
+    {
+      "epoch": 0.24390243902439024,
+      "grad_norm": 0.35989147424697876,
+      "learning_rate": 4.989022233024681e-05,
+      "loss": 2.4993,
+      "step": 190
+    },
+    {
+      "epoch": 0.24518613607188702,
+      "grad_norm": 0.3124522268772125,
+      "learning_rate": 4.988777111066646e-05,
+      "loss": 2.3186,
+      "step": 191
+    },
+    {
+      "epoch": 0.24646983311938384,
+      "grad_norm": 0.32323285937309265,
+      "learning_rate": 4.988529288740096e-05,
+      "loss": 2.3859,
+      "step": 192
+    },
+    {
+      "epoch": 0.24775353016688062,
+      "grad_norm": 0.33156561851501465,
+      "learning_rate": 4.988278766313922e-05,
+      "loss": 2.3271,
+      "step": 193
+    },
+    {
+      "epoch": 0.2490372272143774,
+      "grad_norm": 0.32494014501571655,
+      "learning_rate": 4.9880255440599476e-05,
+      "loss": 2.4023,
+      "step": 194
+    },
+    {
+      "epoch": 0.2503209242618742,
+      "grad_norm": 0.3906209468841553,
+      "learning_rate": 4.987769622252921e-05,
+      "loss": 2.4245,
+      "step": 195
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 3116,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 195,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.366728800609239e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-195/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a98077399ec24ec1c4b81639fc3b1a9aa583b98269d8126c92d55b9add25889
+size 7928

checkpoint-195/zero_to_fp32.py

@@ -0,0 +1,604 @@
+#!/usr/bin/env python
+
+# Copyright (c) Microsoft Corporation.
+# SPDX-License-Identifier: Apache-2.0
+
+# DeepSpeed Team
+
+# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
+# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
+# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
+# application.
+#
+# example: python zero_to_fp32.py . pytorch_model.bin
+
+import argparse
+import torch
+import glob
+import math
+import os
+import re
+from collections import OrderedDict
+from dataclasses import dataclass
+
+# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
+# DeepSpeed data structures it has to be available in the current python environment.
+from deepspeed.utils import logger
+from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
+                                            FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
+                                            FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
+
+
+@dataclass
+class zero_model_state:
+    buffers: dict()
+    param_shapes: dict()
+    shared_params: list
+    ds_version: int
+    frozen_param_shapes: dict()
+    frozen_param_fragments: dict()
+
+
+debug = 0
+
+# load to cpu
+device = torch.device('cpu')
+
+
+def atoi(text):
+    return int(text) if text.isdigit() else text
+
+
+def natural_keys(text):
+    '''
+    alist.sort(key=natural_keys) sorts in human order
+    http://nedbatchelder.com/blog/200712/human_sorting.html
+    (See Toothy's implementation in the comments)
+    '''
+    return [atoi(c) for c in re.split(r'(\d+)', text)]
+
+
+def get_model_state_file(checkpoint_dir, zero_stage):
+    if not os.path.isdir(checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
+
+    # there should be only one file
+    if zero_stage <= 2:
+        file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
+    elif zero_stage == 3:
+        file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
+
+    if not os.path.exists(file):
+        raise FileNotFoundError(f"can't find model states file at '{file}'")
+
+    return file
+
+
+def get_checkpoint_files(checkpoint_dir, glob_pattern):
+    # XXX: need to test that this simple glob rule works for multi-node setup too
+    ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
+
+    if len(ckpt_files) == 0:
+        raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
+
+    return ckpt_files
+
+
+def get_optim_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
+
+
+def get_model_state_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
+
+
+def parse_model_states(files):
+    zero_model_states = []
+    for file in files:
+        state_dict = torch.load(file, map_location=device)
+
+        if BUFFER_NAMES not in state_dict:
+            raise ValueError(f"{file} is not a model state checkpoint")
+        buffer_names = state_dict[BUFFER_NAMES]
+        if debug:
+            print("Found buffers:", buffer_names)
+
+        # recover just the buffers while restoring them to fp32 if they were saved in fp16
+        buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
+        param_shapes = state_dict[PARAM_SHAPES]
+
+        # collect parameters that are included in param_shapes
+        param_names = []
+        for s in param_shapes:
+            for name in s.keys():
+                param_names.append(name)
+
+        # update with frozen parameters
+        frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
+        if frozen_param_shapes is not None:
+            if debug:
+                print(f"Found frozen_param_shapes: {frozen_param_shapes}")
+            param_names += list(frozen_param_shapes.keys())
+
+        # handle shared params
+        shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
+
+        ds_version = state_dict.get(DS_VERSION, None)
+
+        frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
+
+        z_model_state = zero_model_state(buffers=buffers,
+                                         param_shapes=param_shapes,
+                                         shared_params=shared_params,
+                                         ds_version=ds_version,
+                                         frozen_param_shapes=frozen_param_shapes,
+                                         frozen_param_fragments=frozen_param_fragments)
+        zero_model_states.append(z_model_state)
+
+    return zero_model_states
+
+
+def parse_optim_states(files, ds_checkpoint_dir):
+
+    total_files = len(files)
+    state_dicts = []
+    for f in files:
+        state_dict = torch.load(f, map_location=device)
+        # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
+        # and also handle the case where it was already removed by another helper script
+        state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
+        state_dicts.append(state_dict)
+
+    if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
+        raise ValueError(f"{files[0]} is not a zero checkpoint")
+    zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
+    world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
+
+    # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
+    # parameters can be different from data parallelism for non-expert parameters. So we can just
+    # use the max of the partition_count to get the dp world_size.
+
+    if type(world_size) is list:
+        world_size = max(world_size)
+
+    if world_size != total_files:
+        raise ValueError(
+            f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
+            "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
+        )
+
+    # the groups are named differently in each stage
+    if zero_stage <= 2:
+        fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
+    elif zero_stage == 3:
+        fp32_groups_key = FP32_FLAT_GROUPS
+    else:
+        raise ValueError(f"unknown zero stage {zero_stage}")
+
+    if zero_stage <= 2:
+        fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
+    elif zero_stage == 3:
+        # if there is more than one param group, there will be multiple flattened tensors - one
+        # flattened tensor per group - for simplicity merge them into a single tensor
+        #
+        # XXX: could make the script more memory efficient for when there are multiple groups - it
+        # will require matching the sub-lists of param_shapes for each param group flattened tensor
+
+        fp32_flat_groups = [
+            torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
+        ]
+
+    return zero_stage, world_size, fp32_flat_groups
+
+
+def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
+    """
+    Returns fp32 state_dict reconstructed from ds checkpoint
+
+    Args:
+        - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
+
+    """
+    print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
+
+    optim_files = get_optim_files(ds_checkpoint_dir)
+    zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
+    print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
+
+    model_files = get_model_state_files(ds_checkpoint_dir)
+
+    zero_model_states = parse_model_states(model_files)
+    print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
+
+    if zero_stage <= 2:
+        return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+    elif zero_stage == 3:
+        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+
+
+def _zero2_merge_frozen_params(state_dict, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+    frozen_param_fragments = zero_model_states[0].frozen_param_fragments
+
+    if debug:
+        num_elem = sum(s.numel() for s in frozen_param_shapes.values())
+        print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        state_dict[name] = frozen_param_fragments[name]
+
+        if debug:
+            print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _has_callable(obj, fn):
+    attr = getattr(obj, fn, None)
+    return callable(attr)
+
+
+def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+
+    # Reconstruction protocol:
+    #
+    # XXX: document this
+
+    if debug:
+        for i in range(world_size):
+            for j in range(len(fp32_flat_groups[0])):
+                print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
+
+    # XXX: memory usage doubles here (zero2)
+    num_param_groups = len(fp32_flat_groups[0])
+    merged_single_partition_of_fp32_groups = []
+    for i in range(num_param_groups):
+        merged_partitions = [sd[i] for sd in fp32_flat_groups]
+        full_single_fp32_vector = torch.cat(merged_partitions, 0)
+        merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
+    avail_numel = sum(
+        [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
+
+    if debug:
+        wanted_params = sum([len(shapes) for shapes in param_shapes])
+        wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
+        # not asserting if there is a mismatch due to possible padding
+        print(f"Have {avail_numel} numels to process.")
+        print(f"Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    total_numel = 0
+    total_params = 0
+    for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
+        offset = 0
+        avail_numel = full_single_fp32_vector.numel()
+        for name, shape in shapes.items():
+
+            unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
+            total_numel += unpartitioned_numel
+            total_params += 1
+
+            if debug:
+                print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+            state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
+            offset += unpartitioned_numel
+
+        # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
+        # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
+        # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
+        # live optimizer object, so we are checking that the numbers are within the right range
+        align_to = 2 * world_size
+
+        def zero2_align(x):
+            return align_to * math.ceil(x / align_to)
+
+        if debug:
+            print(f"original offset={offset}, avail_numel={avail_numel}")
+
+        offset = zero2_align(offset)
+        avail_numel = zero2_align(avail_numel)
+
+        if debug:
+            print(f"aligned  offset={offset}, avail_numel={avail_numel}")
+
+        # Sanity check
+        if offset != avail_numel:
+            raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero2_merge_frozen_params(state_dict, zero_model_states)
+
+    _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def zero3_partitioned_param_info(unpartitioned_numel, world_size):
+    remainder = unpartitioned_numel % world_size
+    padding_numel = (world_size - remainder) if remainder else 0
+    partitioned_numel = math.ceil(unpartitioned_numel / world_size)
+    return partitioned_numel, padding_numel
+
+
+def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    if debug:
+        for i in range(world_size):
+            num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
+            print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in zero_model_states[0].frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
+        state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+    avail_numel = fp32_flat_groups[0].numel() * world_size
+    # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
+    # param, re-consolidating each param, while dealing with padding if any
+
+    # merge list of dicts, preserving order
+    param_shapes = {k: v for d in param_shapes for k, v in d.items()}
+
+    if debug:
+        for i in range(world_size):
+            print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
+
+        wanted_params = len(param_shapes)
+        wanted_numel = sum(shape.numel() for shape in param_shapes.values())
+        # not asserting if there is a mismatch due to possible padding
+        avail_numel = fp32_flat_groups[0].numel() * world_size
+        print(f"Trainable params: Have {avail_numel} numels to process.")
+        print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    offset = 0
+    total_numel = 0
+    total_params = 0
+    for name, shape in param_shapes.items():
+
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+        total_params += 1
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+        # XXX: memory usage doubles here
+        state_dict[name] = torch.cat(
+            tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
+            0).narrow(0, 0, unpartitioned_numel).view(shape)
+        offset += partitioned_numel
+
+    offset *= world_size
+
+    # Sanity check
+    if offset != avail_numel:
+        raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
+
+    _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
+    ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
+    via a model hub.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+
+    Returns:
+        - pytorch ``state_dict``
+
+    Note: this approach may not work if your application doesn't have sufficient free CPU memory and
+    you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
+    the checkpoint.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        # do the training and checkpoint saving
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
+        model = model.cpu() # move to cpu
+        model.load_state_dict(state_dict)
+        # submit to model hub or save the model to share with others
+
+    In this example the ``model`` will no longer be usable in the deepspeed context of the same
+    application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
+
+    """
+    if tag is None:
+        latest_path = os.path.join(checkpoint_dir, 'latest')
+        if os.path.isfile(latest_path):
+            with open(latest_path, 'r') as fd:
+                tag = fd.read().strip()
+        else:
+            raise ValueError(f"Unable to find 'latest' file at {latest_path}")
+
+    ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
+
+    if not os.path.isdir(ds_checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
+
+    return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
+
+
+def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, 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_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+    """
+
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters)
+    print(f"Saving fp32 state dict to {output_file}")
+    torch.save(state_dict, output_file)
+
+
+def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
+    """
+    1. Put the provided model to cpu
+    2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
+    3. Load it into the provided model
+
+    Args:
+        - ``model``: the model object to update
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+
+    Returns:
+        - ``model`: modified model
+
+    Make sure you have plenty of CPU memory available before you call this function. If you don't
+    have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
+    conveniently placed for you in the checkpoint folder.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
+        model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
+        # submit to model hub or save the model to share with others
+
+    Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
+    of the same application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    """
+    logger.info(f"Extracting fp32 weights")
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+
+    logger.info(f"Overwriting model with fp32 weights")
+    model = model.cpu()
+    model.load_state_dict(state_dict, strict=False)
+
+    return model
+
+
+if __name__ == "__main__":
+
+    parser = argparse.ArgumentParser()
+    parser.add_argument("checkpoint_dir",
+                        type=str,
+                        help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
+    parser.add_argument(
+        "output_file",
+        type=str,
+        help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
+    parser.add_argument("-t",
+                        "--tag",
+                        type=str,
+                        default=None,
+                        help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
+    parser.add_argument("--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_file,
+                                               tag=args.tag,
+                                               exclude_frozen_parameters=args.exclude_frozen_parameters)