Upload folder using huggingface_hub

checkpoint-100/README.md

@@ -0,0 +1,202 @@
+---
+base_model: openbmb/MiniCPM-V-2_6
+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 0.12.0

checkpoint-100/adapter_config.json

@@ -0,0 +1,33 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": {
+    "base_model_class": "MiniCPMV",
+    "parent_library": "transformers_modules.openbmb.MiniCPM-V-2_6.4e4be000cd81feda8b96d14b53f1791b4010b038.modeling_minicpmv"
+  },
+  "base_model_name_or_path": "openbmb/MiniCPM-V-2_6",
+  "bias": "none",
+  "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": 64,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": [
+    "embed_tokens",
+    "resampler",
+    "vpm"
+  ],
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": "llm\\..*layers\\.\\d+\\.self_attn\\.(q_proj|k_proj|v_proj|o_proj)",
+  "task_type": null,
+  "use_dora": false,
+  "use_rslora": false
+}

checkpoint-100/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4d6a6e48efacb468fb82b991d52f0d97757d5a27c10304181c6605c3ecfecb23
+size 2140796552

checkpoint-100/added_tokens.json

@@ -0,0 +1,25 @@
+{
+  "</box>": 151651,
+  "</image>": 151647,
+  "</image_id>": 151659,
+  "</point>": 151655,
+  "</quad>": 151653,
+  "</ref>": 151649,
+  "</slice>": 151657,
+  "<box>": 151650,
+  "<image>": 151646,
+  "<image_id>": 151658,
+  "<point>": 151654,
+  "<quad>": 151652,
+  "<ref>": 151648,
+  "<slice>": 151656,
+  "<|endoftext|>": 151643,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|reserved_special_token_0|>": 151660,
+  "<|reserved_special_token_1|>": 151661,
+  "<|reserved_special_token_2|>": 151662,
+  "<|reserved_special_token_3|>": 151663,
+  "<|reserved_special_token_4|>": 151664,
+  "<|reserved_special_token_5|>": 151665
+}

checkpoint-100/latest

@@ -0,0 +1 @@
+global_step100

checkpoint-100/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e2f9545d5c90c5fd71d917239c4d92b1e8ae9a74e2a10fd0b7bc72776bb56954
+size 15024

checkpoint-100/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:75ffc3a47f94a86086305c449bcc76363b72aff82f5b763912f23d0050a9b8bb
+size 15024

checkpoint-100/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4af7d189af29a7ab5235b27cfdbec052696e797b0d3c37c57d37c931742e6500
+size 15024

checkpoint-100/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b11562552c6ec2dc98c3a7929ce25263d8e5daec5ddf45022c1835c45a4a571a
+size 15024

checkpoint-100/special_tokens_map.json

@@ -0,0 +1,52 @@
+{
+  "additional_special_tokens": [
+    "<image>",
+    "</image>",
+    "<ref>",
+    "</ref>",
+    "<box>",
+    "</box>",
+    "<quad>",
+    "</quad>",
+    "<point>",
+    "</point>",
+    "<slice>",
+    "</slice>",
+    "<image_id>",
+    "</image_id>",
+    "<|reserved_special_token_0|>",
+    "<|reserved_special_token_1|>",
+    "<|reserved_special_token_2|>",
+    "<|reserved_special_token_3|>",
+    "<|reserved_special_token_4|>",
+    "<|reserved_special_token_5|>"
+  ],
+  "bos_token": {
+    "content": "<|im_start|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

checkpoint-100/tokenizer_config.json

@@ -0,0 +1,235 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "128244": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "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": "<image>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "</image>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<ref>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "</ref>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<box>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "</box>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<quad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "</quad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<point>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "</point>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<slice>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "</slice>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151658": {
+      "content": "<image_id>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151659": {
+      "content": "</image_id>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151660": {
+      "content": "<|reserved_special_token_0|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151661": {
+      "content": "<|reserved_special_token_1|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151662": {
+      "content": "<|reserved_special_token_2|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151663": {
+      "content": "<|reserved_special_token_3|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151664": {
+      "content": "<|reserved_special_token_4|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151665": {
+      "content": "<|reserved_special_token_5|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<image>",
+    "</image>",
+    "<ref>",
+    "</ref>",
+    "<box>",
+    "</box>",
+    "<quad>",
+    "</quad>",
+    "<point>",
+    "</point>",
+    "<slice>",
+    "</slice>",
+    "<image_id>",
+    "</image_id>",
+    "<|reserved_special_token_0|>",
+    "<|reserved_special_token_1|>",
+    "<|reserved_special_token_2|>",
+    "<|reserved_special_token_3|>",
+    "<|reserved_special_token_4|>",
+    "<|reserved_special_token_5|>"
+  ],
+  "auto_map": {
+    "AutoTokenizer": [
+      "openbmb/MiniCPM-V-2_6--tokenization_minicpmv_fast.MiniCPMVTokenizerFast",
+      null
+    ]
+  },
+  "bos_token": "<|im_start|>",
+  "chat_template": "{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n' }}{% endif %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "MiniCPMVTokenizer",
+  "unk_token": "<unk>"
+}

checkpoint-100/trainer_state.json

@@ -0,0 +1,733 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 50.0,
+  "eval_steps": 1000,
+  "global_step": 100,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.0,
+      "learning_rate": 0,
+      "loss": 1.4695,
+      "step": 1
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.0,
+      "learning_rate": 0,
+      "loss": 1.4475,
+      "step": 2
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.0,
+      "learning_rate": 0,
+      "loss": 1.5187,
+      "step": 3
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.0,
+      "learning_rate": 0,
+      "loss": 1.3848,
+      "step": 4
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.0,
+      "learning_rate": 0,
+      "loss": 1.3918,
+      "step": 5
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 5.2773966789245605,
+      "learning_rate": 0.0,
+      "loss": 1.4477,
+      "step": 6
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 6.704706192016602,
+      "learning_rate": 1.5051499783199055e-07,
+      "loss": 1.5294,
+      "step": 7
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 18.195552825927734,
+      "learning_rate": 2.385606273598312e-07,
+      "loss": 1.4874,
+      "step": 8
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 4.960075378417969,
+      "learning_rate": 3.010299956639811e-07,
+      "loss": 1.2974,
+      "step": 9
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 7.742049217224121,
+      "learning_rate": 3.494850021680093e-07,
+      "loss": 1.5687,
+      "step": 10
+    },
+    {
+      "epoch": 5.5,
+      "grad_norm": 8.554511070251465,
+      "learning_rate": 3.8907562519182173e-07,
+      "loss": 1.3764,
+      "step": 11
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 6.75316047668457,
+      "learning_rate": 4.2254902000712834e-07,
+      "loss": 1.509,
+      "step": 12
+    },
+    {
+      "epoch": 6.5,
+      "grad_norm": 7.616246700286865,
+      "learning_rate": 4.5154499349597166e-07,
+      "loss": 1.4578,
+      "step": 13
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 5.040093421936035,
+      "learning_rate": 4.771212547196623e-07,
+      "loss": 1.4536,
+      "step": 14
+    },
+    {
+      "epoch": 7.5,
+      "grad_norm": 7.984886646270752,
+      "learning_rate": 4.999999999999999e-07,
+      "loss": 1.4045,
+      "step": 15
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 7.743239879608154,
+      "learning_rate": 5.206963425791124e-07,
+      "loss": 1.4374,
+      "step": 16
+    },
+    {
+      "epoch": 8.5,
+      "grad_norm": 8.207304000854492,
+      "learning_rate": 5.395906230238123e-07,
+      "loss": 1.5724,
+      "step": 17
+    },
+    {
+      "epoch": 9.0,
+      "grad_norm": 6.423882961273193,
+      "learning_rate": 5.569716761534182e-07,
+      "loss": 1.4628,
+      "step": 18
+    },
+    {
+      "epoch": 9.5,
+      "grad_norm": 5.485311985015869,
+      "learning_rate": 5.730640178391189e-07,
+      "loss": 1.4199,
+      "step": 19
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 5.854134559631348,
+      "learning_rate": 5.880456295278405e-07,
+      "loss": 1.3805,
+      "step": 20
+    },
+    {
+      "epoch": 10.5,
+      "grad_norm": 6.40692663192749,
+      "learning_rate": 6.020599913279622e-07,
+      "loss": 1.5921,
+      "step": 21
+    },
+    {
+      "epoch": 11.0,
+      "grad_norm": 4.778651237487793,
+      "learning_rate": 6.15224460689137e-07,
+      "loss": 1.4138,
+      "step": 22
+    },
+    {
+      "epoch": 11.5,
+      "grad_norm": 6.353730201721191,
+      "learning_rate": 6.276362525516529e-07,
+      "loss": 1.421,
+      "step": 23
+    },
+    {
+      "epoch": 12.0,
+      "grad_norm": 3.764188051223755,
+      "learning_rate": 6.393768004764143e-07,
+      "loss": 1.272,
+      "step": 24
+    },
+    {
+      "epoch": 12.5,
+      "grad_norm": 4.659372806549072,
+      "learning_rate": 6.505149978319905e-07,
+      "loss": 1.5793,
+      "step": 25
+    },
+    {
+      "epoch": 13.0,
+      "grad_norm": 3.8088481426239014,
+      "learning_rate": 6.611096473669595e-07,
+      "loss": 1.2148,
+      "step": 26
+    },
+    {
+      "epoch": 13.5,
+      "grad_norm": 4.605009078979492,
+      "learning_rate": 6.712113404111031e-07,
+      "loss": 1.4131,
+      "step": 27
+    },
+    {
+      "epoch": 14.0,
+      "grad_norm": 4.647908687591553,
+      "learning_rate": 6.808639180087963e-07,
+      "loss": 1.2516,
+      "step": 28
+    },
+    {
+      "epoch": 14.5,
+      "grad_norm": 4.514801502227783,
+      "learning_rate": 6.901056208558029e-07,
+      "loss": 1.3228,
+      "step": 29
+    },
+    {
+      "epoch": 15.0,
+      "grad_norm": 4.330278396606445,
+      "learning_rate": 6.989700043360186e-07,
+      "loss": 1.3317,
+      "step": 30
+    },
+    {
+      "epoch": 15.5,
+      "grad_norm": 5.9373321533203125,
+      "learning_rate": 7.074866739854088e-07,
+      "loss": 1.4139,
+      "step": 31
+    },
+    {
+      "epoch": 16.0,
+      "grad_norm": 4.761633396148682,
+      "learning_rate": 7.156818820794935e-07,
+      "loss": 1.3568,
+      "step": 32
+    },
+    {
+      "epoch": 16.5,
+      "grad_norm": 4.071909427642822,
+      "learning_rate": 7.235790156711094e-07,
+      "loss": 1.314,
+      "step": 33
+    },
+    {
+      "epoch": 17.0,
+      "grad_norm": 4.649173259735107,
+      "learning_rate": 7.311989989494779e-07,
+      "loss": 1.4079,
+      "step": 34
+    },
+    {
+      "epoch": 17.5,
+      "grad_norm": 3.983167886734009,
+      "learning_rate": 7.38560627359831e-07,
+      "loss": 1.2689,
+      "step": 35
+    },
+    {
+      "epoch": 18.0,
+      "grad_norm": 4.708627700805664,
+      "learning_rate": 7.456808469171361e-07,
+      "loss": 1.2591,
+      "step": 36
+    },
+    {
+      "epoch": 18.5,
+      "grad_norm": 4.3896260261535645,
+      "learning_rate": 7.525749891599529e-07,
+      "loss": 1.2727,
+      "step": 37
+    },
+    {
+      "epoch": 19.0,
+      "grad_norm": 3.192983627319336,
+      "learning_rate": 7.592569699389436e-07,
+      "loss": 1.2811,
+      "step": 38
+    },
+    {
+      "epoch": 19.5,
+      "grad_norm": 4.576615810394287,
+      "learning_rate": 7.657394585211274e-07,
+      "loss": 1.2692,
+      "step": 39
+    },
+    {
+      "epoch": 20.0,
+      "grad_norm": 6.580528736114502,
+      "learning_rate": 7.720340221751376e-07,
+      "loss": 1.2739,
+      "step": 40
+    },
+    {
+      "epoch": 20.5,
+      "grad_norm": 3.075108528137207,
+      "learning_rate": 7.781512503836435e-07,
+      "loss": 1.1164,
+      "step": 41
+    },
+    {
+      "epoch": 21.0,
+      "grad_norm": 4.4012041091918945,
+      "learning_rate": 7.841008620334974e-07,
+      "loss": 1.3188,
+      "step": 42
+    },
+    {
+      "epoch": 21.5,
+      "grad_norm": 4.855476379394531,
+      "learning_rate": 7.89891798308405e-07,
+      "loss": 1.113,
+      "step": 43
+    },
+    {
+      "epoch": 22.0,
+      "grad_norm": 5.26778507232666,
+      "learning_rate": 7.955323035132494e-07,
+      "loss": 1.3752,
+      "step": 44
+    },
+    {
+      "epoch": 22.5,
+      "grad_norm": 3.732664108276367,
+      "learning_rate": 8.01029995663981e-07,
+      "loss": 1.251,
+      "step": 45
+    },
+    {
+      "epoch": 23.0,
+      "grad_norm": 3.8093373775482178,
+      "learning_rate": 8.063919283598676e-07,
+      "loss": 1.1708,
+      "step": 46
+    },
+    {
+      "epoch": 23.5,
+      "grad_norm": 5.319915771484375,
+      "learning_rate": 8.116246451989502e-07,
+      "loss": 1.292,
+      "step": 47
+    },
+    {
+      "epoch": 24.0,
+      "grad_norm": 3.3320486545562744,
+      "learning_rate": 8.16734227789793e-07,
+      "loss": 1.1096,
+      "step": 48
+    },
+    {
+      "epoch": 24.5,
+      "grad_norm": 4.5503411293029785,
+      "learning_rate": 8.217263382430935e-07,
+      "loss": 1.0945,
+      "step": 49
+    },
+    {
+      "epoch": 25.0,
+      "grad_norm": 4.538294792175293,
+      "learning_rate": 8.266062568876716e-07,
+      "loss": 1.18,
+      "step": 50
+    },
+    {
+      "epoch": 25.5,
+      "grad_norm": 4.297428131103516,
+      "learning_rate": 8.313789158407869e-07,
+      "loss": 1.217,
+      "step": 51
+    },
+    {
+      "epoch": 26.0,
+      "grad_norm": 5.326357364654541,
+      "learning_rate": 8.360489289678585e-07,
+      "loss": 1.1089,
+      "step": 52
+    },
+    {
+      "epoch": 26.5,
+      "grad_norm": 22.330251693725586,
+      "learning_rate": 8.406206186877934e-07,
+      "loss": 1.2477,
+      "step": 53
+    },
+    {
+      "epoch": 27.0,
+      "grad_norm": 3.5450849533081055,
+      "learning_rate": 8.450980400142567e-07,
+      "loss": 1.0081,
+      "step": 54
+    },
+    {
+      "epoch": 27.5,
+      "grad_norm": 3.588421106338501,
+      "learning_rate": 8.494850021680092e-07,
+      "loss": 1.1557,
+      "step": 55
+    },
+    {
+      "epoch": 28.0,
+      "grad_norm": 5.4207987785339355,
+      "learning_rate": 8.53785088048968e-07,
+      "loss": 1.2518,
+      "step": 56
+    },
+    {
+      "epoch": 28.5,
+      "grad_norm": 9.896635055541992,
+      "learning_rate": 8.580016718173995e-07,
+      "loss": 1.0695,
+      "step": 57
+    },
+    {
+      "epoch": 29.0,
+      "grad_norm": 6.4676666259765625,
+      "learning_rate": 8.621379348003944e-07,
+      "loss": 1.2276,
+      "step": 58
+    },
+    {
+      "epoch": 29.5,
+      "grad_norm": 4.066528797149658,
+      "learning_rate": 8.661968799114842e-07,
+      "loss": 1.0339,
+      "step": 59
+    },
+    {
+      "epoch": 30.0,
+      "grad_norm": 4.463730812072754,
+      "learning_rate": 8.701813447471218e-07,
+      "loss": 1.0844,
+      "step": 60
+    },
+    {
+      "epoch": 30.5,
+      "grad_norm": 3.523402690887451,
+      "learning_rate": 8.740940135031001e-07,
+      "loss": 1.0907,
+      "step": 61
+    },
+    {
+      "epoch": 31.0,
+      "grad_norm": 4.35223913192749,
+      "learning_rate": 8.779374278362456e-07,
+      "loss": 1.1098,
+      "step": 62
+    },
+    {
+      "epoch": 31.5,
+      "grad_norm": 9.153762817382812,
+      "learning_rate": 8.817139967814684e-07,
+      "loss": 1.1932,
+      "step": 63
+    },
+    {
+      "epoch": 32.0,
+      "grad_norm": 3.6358563899993896,
+      "learning_rate": 8.854260058210719e-07,
+      "loss": 0.984,
+      "step": 64
+    },
+    {
+      "epoch": 32.5,
+      "grad_norm": 3.1282856464385986,
+      "learning_rate": 8.890756251918216e-07,
+      "loss": 1.004,
+      "step": 65
+    },
+    {
+      "epoch": 33.0,
+      "grad_norm": 3.8792643547058105,
+      "learning_rate": 8.926649175053833e-07,
+      "loss": 1.0181,
+      "step": 66
+    },
+    {
+      "epoch": 33.5,
+      "grad_norm": 3.481888771057129,
+      "learning_rate": 8.961958447491268e-07,
+      "loss": 1.0099,
+      "step": 67
+    },
+    {
+      "epoch": 34.0,
+      "grad_norm": 4.767885684967041,
+      "learning_rate": 8.996702747267907e-07,
+      "loss": 0.99,
+      "step": 68
+    },
+    {
+      "epoch": 34.5,
+      "grad_norm": 6.9532952308654785,
+      "learning_rate": 9.030899869919433e-07,
+      "loss": 0.9372,
+      "step": 69
+    },
+    {
+      "epoch": 35.0,
+      "grad_norm": 5.427297592163086,
+      "learning_rate": 9.064566783214276e-07,
+      "loss": 1.023,
+      "step": 70
+    },
+    {
+      "epoch": 35.5,
+      "grad_norm": 3.7851903438568115,
+      "learning_rate": 9.097719677709341e-07,
+      "loss": 0.9984,
+      "step": 71
+    },
+    {
+      "epoch": 36.0,
+      "grad_norm": 3.750429630279541,
+      "learning_rate": 9.13037401350413e-07,
+      "loss": 0.9991,
+      "step": 72
+    },
+    {
+      "epoch": 36.5,
+      "grad_norm": 3.222670793533325,
+      "learning_rate": 9.162544563531181e-07,
+      "loss": 0.8962,
+      "step": 73
+    },
+    {
+      "epoch": 37.0,
+      "grad_norm": 2.7355740070343018,
+      "learning_rate": 9.194245453686276e-07,
+      "loss": 1.0381,
+      "step": 74
+    },
+    {
+      "epoch": 37.5,
+      "grad_norm": 6.130687236785889,
+      "learning_rate": 9.225490200071283e-07,
+      "loss": 0.943,
+      "step": 75
+    },
+    {
+      "epoch": 38.0,
+      "grad_norm": 4.569997310638428,
+      "learning_rate": 9.256291743595375e-07,
+      "loss": 1.0464,
+      "step": 76
+    },
+    {
+      "epoch": 38.5,
+      "grad_norm": 4.291537761688232,
+      "learning_rate": 9.28666248215634e-07,
+      "loss": 0.965,
+      "step": 77
+    },
+    {
+      "epoch": 39.0,
+      "grad_norm": 3.6804943084716797,
+      "learning_rate": 9.316614300602277e-07,
+      "loss": 0.9621,
+      "step": 78
+    },
+    {
+      "epoch": 39.5,
+      "grad_norm": 3.408571720123291,
+      "learning_rate": 9.346158598654879e-07,
+      "loss": 0.9144,
+      "step": 79
+    },
+    {
+      "epoch": 40.0,
+      "grad_norm": 3.565375566482544,
+      "learning_rate": 9.375306316958498e-07,
+      "loss": 0.8959,
+      "step": 80
+    },
+    {
+      "epoch": 40.5,
+      "grad_norm": 5.186458110809326,
+      "learning_rate": 9.404067961403955e-07,
+      "loss": 0.9522,
+      "step": 81
+    },
+    {
+      "epoch": 41.0,
+      "grad_norm": 3.754727602005005,
+      "learning_rate": 9.432453625862408e-07,
+      "loss": 0.8659,
+      "step": 82
+    },
+    {
+      "epoch": 41.5,
+      "grad_norm": 7.740540504455566,
+      "learning_rate": 9.4604730134524e-07,
+      "loss": 0.8567,
+      "step": 83
+    },
+    {
+      "epoch": 42.0,
+      "grad_norm": 5.675983905792236,
+      "learning_rate": 9.488135456452205e-07,
+      "loss": 0.9199,
+      "step": 84
+    },
+    {
+      "epoch": 42.5,
+      "grad_norm": 4.611757755279541,
+      "learning_rate": 9.515449934959715e-07,
+      "loss": 0.9635,
+      "step": 85
+    },
+    {
+      "epoch": 43.0,
+      "grad_norm": 2.9257593154907227,
+      "learning_rate": 9.542425094393247e-07,
+      "loss": 0.9204,
+      "step": 86
+    },
+    {
+      "epoch": 43.5,
+      "grad_norm": 3.5699961185455322,
+      "learning_rate": 9.569069261918583e-07,
+      "loss": 0.9794,
+      "step": 87
+    },
+    {
+      "epoch": 44.0,
+      "grad_norm": 4.43387508392334,
+      "learning_rate": 9.59539046188037e-07,
+      "loss": 0.8352,
+      "step": 88
+    },
+    {
+      "epoch": 44.5,
+      "grad_norm": 3.140352487564087,
+      "learning_rate": 9.621396430309406e-07,
+      "loss": 0.8887,
+      "step": 89
+    },
+    {
+      "epoch": 45.0,
+      "grad_norm": 5.941178798675537,
+      "learning_rate": 9.647094628571462e-07,
+      "loss": 0.8571,
+      "step": 90
+    },
+    {
+      "epoch": 45.5,
+      "grad_norm": 2.8487863540649414,
+      "learning_rate": 9.672492256217836e-07,
+      "loss": 0.8364,
+      "step": 91
+    },
+    {
+      "epoch": 46.0,
+      "grad_norm": 2.790127754211426,
+      "learning_rate": 9.69759626309309e-07,
+      "loss": 0.8898,
+      "step": 92
+    },
+    {
+      "epoch": 46.5,
+      "grad_norm": 2.601271629333496,
+      "learning_rate": 9.722413360750842e-07,
+      "loss": 0.8499,
+      "step": 93
+    },
+    {
+      "epoch": 47.0,
+      "grad_norm": 6.0029296875,
+      "learning_rate": 9.74695003322456e-07,
+      "loss": 0.7976,
+      "step": 94
+    },
+    {
+      "epoch": 47.5,
+      "grad_norm": 3.055148124694824,
+      "learning_rate": 9.771212547196622e-07,
+      "loss": 0.7946,
+      "step": 95
+    },
+    {
+      "epoch": 48.0,
+      "grad_norm": 5.841665744781494,
+      "learning_rate": 9.795206961605466e-07,
+      "loss": 0.8136,
+      "step": 96
+    },
+    {
+      "epoch": 48.5,
+      "grad_norm": 3.1043038368225098,
+      "learning_rate": 9.818939136727774e-07,
+      "loss": 0.7518,
+      "step": 97
+    },
+    {
+      "epoch": 49.0,
+      "grad_norm": 5.523963928222656,
+      "learning_rate": 9.842414742769674e-07,
+      "loss": 0.8265,
+      "step": 98
+    },
+    {
+      "epoch": 49.5,
+      "grad_norm": 2.8792777061462402,
+      "learning_rate": 9.865639267998492e-07,
+      "loss": 0.7474,
+      "step": 99
+    },
+    {
+      "epoch": 50.0,
+      "grad_norm": 2.840219497680664,
+      "learning_rate": 9.888618026444236e-07,
+      "loss": 0.784,
+      "step": 100
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5000,
+  "save_steps": 50,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.4015329495678976e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-100/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b36bec9488b445350964041059369935c653f243588cf528b415499fc00dcca2
+size 6968

checkpoint-100/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)

checkpoint-150/README.md

@@ -0,0 +1,202 @@
+---
+base_model: openbmb/MiniCPM-V-2_6
+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 0.12.0

checkpoint-150/adapter_config.json

@@ -0,0 +1,33 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": {
+    "base_model_class": "MiniCPMV",
+    "parent_library": "transformers_modules.openbmb.MiniCPM-V-2_6.4e4be000cd81feda8b96d14b53f1791b4010b038.modeling_minicpmv"
+  },
+  "base_model_name_or_path": "openbmb/MiniCPM-V-2_6",
+  "bias": "none",
+  "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": 64,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": [
+    "embed_tokens",
+    "resampler",
+    "vpm"
+  ],
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": "llm\\..*layers\\.\\d+\\.self_attn\\.(q_proj|k_proj|v_proj|o_proj)",
+  "task_type": null,
+  "use_dora": false,
+  "use_rslora": false
+}

checkpoint-150/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7794d72e3ff8325f2dbd3317faa641f1b65eaf1c66813a7489ede91479381bcc
+size 2140796552

checkpoint-150/added_tokens.json

@@ -0,0 +1,25 @@
+{
+  "</box>": 151651,
+  "</image>": 151647,
+  "</image_id>": 151659,
+  "</point>": 151655,
+  "</quad>": 151653,
+  "</ref>": 151649,
+  "</slice>": 151657,
+  "<box>": 151650,
+  "<image>": 151646,
+  "<image_id>": 151658,
+  "<point>": 151654,
+  "<quad>": 151652,
+  "<ref>": 151648,
+  "<slice>": 151656,
+  "<|endoftext|>": 151643,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|reserved_special_token_0|>": 151660,
+  "<|reserved_special_token_1|>": 151661,
+  "<|reserved_special_token_2|>": 151662,
+  "<|reserved_special_token_3|>": 151663,
+  "<|reserved_special_token_4|>": 151664,
+  "<|reserved_special_token_5|>": 151665
+}

checkpoint-150/latest

@@ -0,0 +1 @@
+global_step150

checkpoint-150/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c7a08f4e7ee1e4f025bb90ecd215d01a19c1932033153c3bfa18632d82f8c644
+size 15024

checkpoint-150/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5ebd49c8029a26c79ffc307273a9913225a2be23eac594826ebf368b5c731bb2
+size 15024

checkpoint-150/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8ec1bf915187c7cd290cc29bdf2b4cf1df5dfef3a44efdd2711b903a864cfd60
+size 15024

checkpoint-150/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4123c08c5b84f0dc1f5ec7b0bff5bdfdc97a5b8498331986e370d4fa7cc0ac6d
+size 15024

checkpoint-150/special_tokens_map.json

@@ -0,0 +1,52 @@
+{
+  "additional_special_tokens": [
+    "<image>",
+    "</image>",
+    "<ref>",
+    "</ref>",
+    "<box>",
+    "</box>",
+    "<quad>",
+    "</quad>",
+    "<point>",
+    "</point>",
+    "<slice>",
+    "</slice>",
+    "<image_id>",
+    "</image_id>",
+    "<|reserved_special_token_0|>",
+    "<|reserved_special_token_1|>",
+    "<|reserved_special_token_2|>",
+    "<|reserved_special_token_3|>",
+    "<|reserved_special_token_4|>",
+    "<|reserved_special_token_5|>"
+  ],
+  "bos_token": {
+    "content": "<|im_start|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

checkpoint-150/tokenizer_config.json

@@ -0,0 +1,235 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "128244": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "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": "<image>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "</image>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<ref>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "</ref>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<box>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "</box>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<quad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "</quad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<point>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "</point>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<slice>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "</slice>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151658": {
+      "content": "<image_id>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151659": {
+      "content": "</image_id>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151660": {
+      "content": "<|reserved_special_token_0|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151661": {
+      "content": "<|reserved_special_token_1|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151662": {
+      "content": "<|reserved_special_token_2|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151663": {
+      "content": "<|reserved_special_token_3|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151664": {
+      "content": "<|reserved_special_token_4|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151665": {
+      "content": "<|reserved_special_token_5|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<image>",
+    "</image>",
+    "<ref>",
+    "</ref>",
+    "<box>",
+    "</box>",
+    "<quad>",
+    "</quad>",
+    "<point>",
+    "</point>",
+    "<slice>",
+    "</slice>",
+    "<image_id>",
+    "</image_id>",
+    "<|reserved_special_token_0|>",
+    "<|reserved_special_token_1|>",
+    "<|reserved_special_token_2|>",
+    "<|reserved_special_token_3|>",
+    "<|reserved_special_token_4|>",
+    "<|reserved_special_token_5|>"
+  ],
+  "auto_map": {
+    "AutoTokenizer": [
+      "openbmb/MiniCPM-V-2_6--tokenization_minicpmv_fast.MiniCPMVTokenizerFast",
+      null
+    ]
+  },
+  "bos_token": "<|im_start|>",
+  "chat_template": "{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n' }}{% endif %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "MiniCPMVTokenizer",
+  "unk_token": "<unk>"
+}

checkpoint-150/trainer_state.json

@@ -0,0 +1,1083 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 75.0,
+  "eval_steps": 1000,
+  "global_step": 150,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.0,
+      "learning_rate": 0,
+      "loss": 1.4695,
+      "step": 1
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.0,
+      "learning_rate": 0,
+      "loss": 1.4475,
+      "step": 2
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.0,
+      "learning_rate": 0,
+      "loss": 1.5187,
+      "step": 3
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.0,
+      "learning_rate": 0,
+      "loss": 1.3848,
+      "step": 4
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.0,
+      "learning_rate": 0,
+      "loss": 1.3918,
+      "step": 5
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 5.2773966789245605,
+      "learning_rate": 0.0,
+      "loss": 1.4477,
+      "step": 6
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 6.704706192016602,
+      "learning_rate": 1.5051499783199055e-07,
+      "loss": 1.5294,
+      "step": 7
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 18.195552825927734,
+      "learning_rate": 2.385606273598312e-07,
+      "loss": 1.4874,
+      "step": 8
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 4.960075378417969,
+      "learning_rate": 3.010299956639811e-07,
+      "loss": 1.2974,
+      "step": 9
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 7.742049217224121,
+      "learning_rate": 3.494850021680093e-07,
+      "loss": 1.5687,
+      "step": 10
+    },
+    {
+      "epoch": 5.5,
+      "grad_norm": 8.554511070251465,
+      "learning_rate": 3.8907562519182173e-07,
+      "loss": 1.3764,
+      "step": 11
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 6.75316047668457,
+      "learning_rate": 4.2254902000712834e-07,
+      "loss": 1.509,
+      "step": 12
+    },
+    {
+      "epoch": 6.5,
+      "grad_norm": 7.616246700286865,
+      "learning_rate": 4.5154499349597166e-07,
+      "loss": 1.4578,
+      "step": 13
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 5.040093421936035,
+      "learning_rate": 4.771212547196623e-07,
+      "loss": 1.4536,
+      "step": 14
+    },
+    {
+      "epoch": 7.5,
+      "grad_norm": 7.984886646270752,
+      "learning_rate": 4.999999999999999e-07,
+      "loss": 1.4045,
+      "step": 15
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 7.743239879608154,
+      "learning_rate": 5.206963425791124e-07,
+      "loss": 1.4374,
+      "step": 16
+    },
+    {
+      "epoch": 8.5,
+      "grad_norm": 8.207304000854492,
+      "learning_rate": 5.395906230238123e-07,
+      "loss": 1.5724,
+      "step": 17
+    },
+    {
+      "epoch": 9.0,
+      "grad_norm": 6.423882961273193,
+      "learning_rate": 5.569716761534182e-07,
+      "loss": 1.4628,
+      "step": 18
+    },
+    {
+      "epoch": 9.5,
+      "grad_norm": 5.485311985015869,
+      "learning_rate": 5.730640178391189e-07,
+      "loss": 1.4199,
+      "step": 19
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 5.854134559631348,
+      "learning_rate": 5.880456295278405e-07,
+      "loss": 1.3805,
+      "step": 20
+    },
+    {
+      "epoch": 10.5,
+      "grad_norm": 6.40692663192749,
+      "learning_rate": 6.020599913279622e-07,
+      "loss": 1.5921,
+      "step": 21
+    },
+    {
+      "epoch": 11.0,
+      "grad_norm": 4.778651237487793,
+      "learning_rate": 6.15224460689137e-07,
+      "loss": 1.4138,
+      "step": 22
+    },
+    {
+      "epoch": 11.5,
+      "grad_norm": 6.353730201721191,
+      "learning_rate": 6.276362525516529e-07,
+      "loss": 1.421,
+      "step": 23
+    },
+    {
+      "epoch": 12.0,
+      "grad_norm": 3.764188051223755,
+      "learning_rate": 6.393768004764143e-07,
+      "loss": 1.272,
+      "step": 24
+    },
+    {
+      "epoch": 12.5,
+      "grad_norm": 4.659372806549072,
+      "learning_rate": 6.505149978319905e-07,
+      "loss": 1.5793,
+      "step": 25
+    },
+    {
+      "epoch": 13.0,
+      "grad_norm": 3.8088481426239014,
+      "learning_rate": 6.611096473669595e-07,
+      "loss": 1.2148,
+      "step": 26
+    },
+    {
+      "epoch": 13.5,
+      "grad_norm": 4.605009078979492,
+      "learning_rate": 6.712113404111031e-07,
+      "loss": 1.4131,
+      "step": 27
+    },
+    {
+      "epoch": 14.0,
+      "grad_norm": 4.647908687591553,
+      "learning_rate": 6.808639180087963e-07,
+      "loss": 1.2516,
+      "step": 28
+    },
+    {
+      "epoch": 14.5,
+      "grad_norm": 4.514801502227783,
+      "learning_rate": 6.901056208558029e-07,
+      "loss": 1.3228,
+      "step": 29
+    },
+    {
+      "epoch": 15.0,
+      "grad_norm": 4.330278396606445,
+      "learning_rate": 6.989700043360186e-07,
+      "loss": 1.3317,
+      "step": 30
+    },
+    {
+      "epoch": 15.5,
+      "grad_norm": 5.9373321533203125,
+      "learning_rate": 7.074866739854088e-07,
+      "loss": 1.4139,
+      "step": 31
+    },
+    {
+      "epoch": 16.0,
+      "grad_norm": 4.761633396148682,
+      "learning_rate": 7.156818820794935e-07,
+      "loss": 1.3568,
+      "step": 32
+    },
+    {
+      "epoch": 16.5,
+      "grad_norm": 4.071909427642822,
+      "learning_rate": 7.235790156711094e-07,
+      "loss": 1.314,
+      "step": 33
+    },
+    {
+      "epoch": 17.0,
+      "grad_norm": 4.649173259735107,
+      "learning_rate": 7.311989989494779e-07,
+      "loss": 1.4079,
+      "step": 34
+    },
+    {
+      "epoch": 17.5,
+      "grad_norm": 3.983167886734009,
+      "learning_rate": 7.38560627359831e-07,
+      "loss": 1.2689,
+      "step": 35
+    },
+    {
+      "epoch": 18.0,
+      "grad_norm": 4.708627700805664,
+      "learning_rate": 7.456808469171361e-07,
+      "loss": 1.2591,
+      "step": 36
+    },
+    {
+      "epoch": 18.5,
+      "grad_norm": 4.3896260261535645,
+      "learning_rate": 7.525749891599529e-07,
+      "loss": 1.2727,
+      "step": 37
+    },
+    {
+      "epoch": 19.0,
+      "grad_norm": 3.192983627319336,
+      "learning_rate": 7.592569699389436e-07,
+      "loss": 1.2811,
+      "step": 38
+    },
+    {
+      "epoch": 19.5,
+      "grad_norm": 4.576615810394287,
+      "learning_rate": 7.657394585211274e-07,
+      "loss": 1.2692,
+      "step": 39
+    },
+    {
+      "epoch": 20.0,
+      "grad_norm": 6.580528736114502,
+      "learning_rate": 7.720340221751376e-07,
+      "loss": 1.2739,
+      "step": 40
+    },
+    {
+      "epoch": 20.5,
+      "grad_norm": 3.075108528137207,
+      "learning_rate": 7.781512503836435e-07,
+      "loss": 1.1164,
+      "step": 41
+    },
+    {
+      "epoch": 21.0,
+      "grad_norm": 4.4012041091918945,
+      "learning_rate": 7.841008620334974e-07,
+      "loss": 1.3188,
+      "step": 42
+    },
+    {
+      "epoch": 21.5,
+      "grad_norm": 4.855476379394531,
+      "learning_rate": 7.89891798308405e-07,
+      "loss": 1.113,
+      "step": 43
+    },
+    {
+      "epoch": 22.0,
+      "grad_norm": 5.26778507232666,
+      "learning_rate": 7.955323035132494e-07,
+      "loss": 1.3752,
+      "step": 44
+    },
+    {
+      "epoch": 22.5,
+      "grad_norm": 3.732664108276367,
+      "learning_rate": 8.01029995663981e-07,
+      "loss": 1.251,
+      "step": 45
+    },
+    {
+      "epoch": 23.0,
+      "grad_norm": 3.8093373775482178,
+      "learning_rate": 8.063919283598676e-07,
+      "loss": 1.1708,
+      "step": 46
+    },
+    {
+      "epoch": 23.5,
+      "grad_norm": 5.319915771484375,
+      "learning_rate": 8.116246451989502e-07,
+      "loss": 1.292,
+      "step": 47
+    },
+    {
+      "epoch": 24.0,
+      "grad_norm": 3.3320486545562744,
+      "learning_rate": 8.16734227789793e-07,
+      "loss": 1.1096,
+      "step": 48
+    },
+    {
+      "epoch": 24.5,
+      "grad_norm": 4.5503411293029785,
+      "learning_rate": 8.217263382430935e-07,
+      "loss": 1.0945,
+      "step": 49
+    },
+    {
+      "epoch": 25.0,
+      "grad_norm": 4.538294792175293,
+      "learning_rate": 8.266062568876716e-07,
+      "loss": 1.18,
+      "step": 50
+    },
+    {
+      "epoch": 25.5,
+      "grad_norm": 4.297428131103516,
+      "learning_rate": 8.313789158407869e-07,
+      "loss": 1.217,
+      "step": 51
+    },
+    {
+      "epoch": 26.0,
+      "grad_norm": 5.326357364654541,
+      "learning_rate": 8.360489289678585e-07,
+      "loss": 1.1089,
+      "step": 52
+    },
+    {
+      "epoch": 26.5,
+      "grad_norm": 22.330251693725586,
+      "learning_rate": 8.406206186877934e-07,
+      "loss": 1.2477,
+      "step": 53
+    },
+    {
+      "epoch": 27.0,
+      "grad_norm": 3.5450849533081055,
+      "learning_rate": 8.450980400142567e-07,
+      "loss": 1.0081,
+      "step": 54
+    },
+    {
+      "epoch": 27.5,
+      "grad_norm": 3.588421106338501,
+      "learning_rate": 8.494850021680092e-07,
+      "loss": 1.1557,
+      "step": 55
+    },
+    {
+      "epoch": 28.0,
+      "grad_norm": 5.4207987785339355,
+      "learning_rate": 8.53785088048968e-07,
+      "loss": 1.2518,
+      "step": 56
+    },
+    {
+      "epoch": 28.5,
+      "grad_norm": 9.896635055541992,
+      "learning_rate": 8.580016718173995e-07,
+      "loss": 1.0695,
+      "step": 57
+    },
+    {
+      "epoch": 29.0,
+      "grad_norm": 6.4676666259765625,
+      "learning_rate": 8.621379348003944e-07,
+      "loss": 1.2276,
+      "step": 58
+    },
+    {
+      "epoch": 29.5,
+      "grad_norm": 4.066528797149658,
+      "learning_rate": 8.661968799114842e-07,
+      "loss": 1.0339,
+      "step": 59
+    },
+    {
+      "epoch": 30.0,
+      "grad_norm": 4.463730812072754,
+      "learning_rate": 8.701813447471218e-07,
+      "loss": 1.0844,
+      "step": 60
+    },
+    {
+      "epoch": 30.5,
+      "grad_norm": 3.523402690887451,
+      "learning_rate": 8.740940135031001e-07,
+      "loss": 1.0907,
+      "step": 61
+    },
+    {
+      "epoch": 31.0,
+      "grad_norm": 4.35223913192749,
+      "learning_rate": 8.779374278362456e-07,
+      "loss": 1.1098,
+      "step": 62
+    },
+    {
+      "epoch": 31.5,
+      "grad_norm": 9.153762817382812,
+      "learning_rate": 8.817139967814684e-07,
+      "loss": 1.1932,
+      "step": 63
+    },
+    {
+      "epoch": 32.0,
+      "grad_norm": 3.6358563899993896,
+      "learning_rate": 8.854260058210719e-07,
+      "loss": 0.984,
+      "step": 64
+    },
+    {
+      "epoch": 32.5,
+      "grad_norm": 3.1282856464385986,
+      "learning_rate": 8.890756251918216e-07,
+      "loss": 1.004,
+      "step": 65
+    },
+    {
+      "epoch": 33.0,
+      "grad_norm": 3.8792643547058105,
+      "learning_rate": 8.926649175053833e-07,
+      "loss": 1.0181,
+      "step": 66
+    },
+    {
+      "epoch": 33.5,
+      "grad_norm": 3.481888771057129,
+      "learning_rate": 8.961958447491268e-07,
+      "loss": 1.0099,
+      "step": 67
+    },
+    {
+      "epoch": 34.0,
+      "grad_norm": 4.767885684967041,
+      "learning_rate": 8.996702747267907e-07,
+      "loss": 0.99,
+      "step": 68
+    },
+    {
+      "epoch": 34.5,
+      "grad_norm": 6.9532952308654785,
+      "learning_rate": 9.030899869919433e-07,
+      "loss": 0.9372,
+      "step": 69
+    },
+    {
+      "epoch": 35.0,
+      "grad_norm": 5.427297592163086,
+      "learning_rate": 9.064566783214276e-07,
+      "loss": 1.023,
+      "step": 70
+    },
+    {
+      "epoch": 35.5,
+      "grad_norm": 3.7851903438568115,
+      "learning_rate": 9.097719677709341e-07,
+      "loss": 0.9984,
+      "step": 71
+    },
+    {
+      "epoch": 36.0,
+      "grad_norm": 3.750429630279541,
+      "learning_rate": 9.13037401350413e-07,
+      "loss": 0.9991,
+      "step": 72
+    },
+    {
+      "epoch": 36.5,
+      "grad_norm": 3.222670793533325,
+      "learning_rate": 9.162544563531181e-07,
+      "loss": 0.8962,
+      "step": 73
+    },
+    {
+      "epoch": 37.0,
+      "grad_norm": 2.7355740070343018,
+      "learning_rate": 9.194245453686276e-07,
+      "loss": 1.0381,
+      "step": 74
+    },
+    {
+      "epoch": 37.5,
+      "grad_norm": 6.130687236785889,
+      "learning_rate": 9.225490200071283e-07,
+      "loss": 0.943,
+      "step": 75
+    },
+    {
+      "epoch": 38.0,
+      "grad_norm": 4.569997310638428,
+      "learning_rate": 9.256291743595375e-07,
+      "loss": 1.0464,
+      "step": 76
+    },
+    {
+      "epoch": 38.5,
+      "grad_norm": 4.291537761688232,
+      "learning_rate": 9.28666248215634e-07,
+      "loss": 0.965,
+      "step": 77
+    },
+    {
+      "epoch": 39.0,
+      "grad_norm": 3.6804943084716797,
+      "learning_rate": 9.316614300602277e-07,
+      "loss": 0.9621,
+      "step": 78
+    },
+    {
+      "epoch": 39.5,
+      "grad_norm": 3.408571720123291,
+      "learning_rate": 9.346158598654879e-07,
+      "loss": 0.9144,
+      "step": 79
+    },
+    {
+      "epoch": 40.0,
+      "grad_norm": 3.565375566482544,
+      "learning_rate": 9.375306316958498e-07,
+      "loss": 0.8959,
+      "step": 80
+    },
+    {
+      "epoch": 40.5,
+      "grad_norm": 5.186458110809326,
+      "learning_rate": 9.404067961403955e-07,
+      "loss": 0.9522,
+      "step": 81
+    },
+    {
+      "epoch": 41.0,
+      "grad_norm": 3.754727602005005,
+      "learning_rate": 9.432453625862408e-07,
+      "loss": 0.8659,
+      "step": 82
+    },
+    {
+      "epoch": 41.5,
+      "grad_norm": 7.740540504455566,
+      "learning_rate": 9.4604730134524e-07,
+      "loss": 0.8567,
+      "step": 83
+    },
+    {
+      "epoch": 42.0,
+      "grad_norm": 5.675983905792236,
+      "learning_rate": 9.488135456452205e-07,
+      "loss": 0.9199,
+      "step": 84
+    },
+    {
+      "epoch": 42.5,
+      "grad_norm": 4.611757755279541,
+      "learning_rate": 9.515449934959715e-07,
+      "loss": 0.9635,
+      "step": 85
+    },
+    {
+      "epoch": 43.0,
+      "grad_norm": 2.9257593154907227,
+      "learning_rate": 9.542425094393247e-07,
+      "loss": 0.9204,
+      "step": 86
+    },
+    {
+      "epoch": 43.5,
+      "grad_norm": 3.5699961185455322,
+      "learning_rate": 9.569069261918583e-07,
+      "loss": 0.9794,
+      "step": 87
+    },
+    {
+      "epoch": 44.0,
+      "grad_norm": 4.43387508392334,
+      "learning_rate": 9.59539046188037e-07,
+      "loss": 0.8352,
+      "step": 88
+    },
+    {
+      "epoch": 44.5,
+      "grad_norm": 3.140352487564087,
+      "learning_rate": 9.621396430309406e-07,
+      "loss": 0.8887,
+      "step": 89
+    },
+    {
+      "epoch": 45.0,
+      "grad_norm": 5.941178798675537,
+      "learning_rate": 9.647094628571462e-07,
+      "loss": 0.8571,
+      "step": 90
+    },
+    {
+      "epoch": 45.5,
+      "grad_norm": 2.8487863540649414,
+      "learning_rate": 9.672492256217836e-07,
+      "loss": 0.8364,
+      "step": 91
+    },
+    {
+      "epoch": 46.0,
+      "grad_norm": 2.790127754211426,
+      "learning_rate": 9.69759626309309e-07,
+      "loss": 0.8898,
+      "step": 92
+    },
+    {
+      "epoch": 46.5,
+      "grad_norm": 2.601271629333496,
+      "learning_rate": 9.722413360750842e-07,
+      "loss": 0.8499,
+      "step": 93
+    },
+    {
+      "epoch": 47.0,
+      "grad_norm": 6.0029296875,
+      "learning_rate": 9.74695003322456e-07,
+      "loss": 0.7976,
+      "step": 94
+    },
+    {
+      "epoch": 47.5,
+      "grad_norm": 3.055148124694824,
+      "learning_rate": 9.771212547196622e-07,
+      "loss": 0.7946,
+      "step": 95
+    },
+    {
+      "epoch": 48.0,
+      "grad_norm": 5.841665744781494,
+      "learning_rate": 9.795206961605466e-07,
+      "loss": 0.8136,
+      "step": 96
+    },
+    {
+      "epoch": 48.5,
+      "grad_norm": 3.1043038368225098,
+      "learning_rate": 9.818939136727774e-07,
+      "loss": 0.7518,
+      "step": 97
+    },
+    {
+      "epoch": 49.0,
+      "grad_norm": 5.523963928222656,
+      "learning_rate": 9.842414742769674e-07,
+      "loss": 0.8265,
+      "step": 98
+    },
+    {
+      "epoch": 49.5,
+      "grad_norm": 2.8792777061462402,
+      "learning_rate": 9.865639267998492e-07,
+      "loss": 0.7474,
+      "step": 99
+    },
+    {
+      "epoch": 50.0,
+      "grad_norm": 2.840219497680664,
+      "learning_rate": 9.888618026444236e-07,
+      "loss": 0.784,
+      "step": 100
+    },
+    {
+      "epoch": 50.5,
+      "grad_norm": 2.887335777282715,
+      "learning_rate": 9.91135616519784e-07,
+      "loss": 0.7575,
+      "step": 101
+    },
+    {
+      "epoch": 51.0,
+      "grad_norm": 2.592064619064331,
+      "learning_rate": 9.933858671331222e-07,
+      "loss": 0.7871,
+      "step": 102
+    },
+    {
+      "epoch": 51.5,
+      "grad_norm": 2.881988048553467,
+      "learning_rate": 9.956130378462473e-07,
+      "loss": 0.8135,
+      "step": 103
+    },
+    {
+      "epoch": 52.0,
+      "grad_norm": 3.1426312923431396,
+      "learning_rate": 9.978175972987748e-07,
+      "loss": 0.6766,
+      "step": 104
+    },
+    {
+      "epoch": 52.5,
+      "grad_norm": 2.951082229614258,
+      "learning_rate": 9.999999999999997e-07,
+      "loss": 0.6936,
+      "step": 105
+    },
+    {
+      "epoch": 53.0,
+      "grad_norm": 5.04337215423584,
+      "learning_rate": 1e-06,
+      "loss": 0.7773,
+      "step": 106
+    },
+    {
+      "epoch": 53.5,
+      "grad_norm": 3.2338459491729736,
+      "learning_rate": 1e-06,
+      "loss": 0.833,
+      "step": 107
+    },
+    {
+      "epoch": 54.0,
+      "grad_norm": 2.7215511798858643,
+      "learning_rate": 1e-06,
+      "loss": 0.6974,
+      "step": 108
+    },
+    {
+      "epoch": 54.5,
+      "grad_norm": 2.6445066928863525,
+      "learning_rate": 1e-06,
+      "loss": 0.7189,
+      "step": 109
+    },
+    {
+      "epoch": 55.0,
+      "grad_norm": 3.2087759971618652,
+      "learning_rate": 1e-06,
+      "loss": 0.6723,
+      "step": 110
+    },
+    {
+      "epoch": 55.5,
+      "grad_norm": 3.1238620281219482,
+      "learning_rate": 1e-06,
+      "loss": 0.6299,
+      "step": 111
+    },
+    {
+      "epoch": 56.0,
+      "grad_norm": 5.280316352844238,
+      "learning_rate": 1e-06,
+      "loss": 0.7757,
+      "step": 112
+    },
+    {
+      "epoch": 56.5,
+      "grad_norm": 4.072208881378174,
+      "learning_rate": 1e-06,
+      "loss": 0.6786,
+      "step": 113
+    },
+    {
+      "epoch": 57.0,
+      "grad_norm": 3.1515462398529053,
+      "learning_rate": 1e-06,
+      "loss": 0.5982,
+      "step": 114
+    },
+    {
+      "epoch": 57.5,
+      "grad_norm": 2.891411781311035,
+      "learning_rate": 1e-06,
+      "loss": 0.575,
+      "step": 115
+    },
+    {
+      "epoch": 58.0,
+      "grad_norm": 5.040134429931641,
+      "learning_rate": 1e-06,
+      "loss": 0.6497,
+      "step": 116
+    },
+    {
+      "epoch": 58.5,
+      "grad_norm": 2.7814080715179443,
+      "learning_rate": 1e-06,
+      "loss": 0.5745,
+      "step": 117
+    },
+    {
+      "epoch": 59.0,
+      "grad_norm": 3.400952100753784,
+      "learning_rate": 1e-06,
+      "loss": 0.6237,
+      "step": 118
+    },
+    {
+      "epoch": 59.5,
+      "grad_norm": 2.8887929916381836,
+      "learning_rate": 1e-06,
+      "loss": 0.5832,
+      "step": 119
+    },
+    {
+      "epoch": 60.0,
+      "grad_norm": 5.328385353088379,
+      "learning_rate": 1e-06,
+      "loss": 0.5677,
+      "step": 120
+    },
+    {
+      "epoch": 60.5,
+      "grad_norm": 3.790076494216919,
+      "learning_rate": 1e-06,
+      "loss": 0.5726,
+      "step": 121
+    },
+    {
+      "epoch": 61.0,
+      "grad_norm": 3.321824550628662,
+      "learning_rate": 1e-06,
+      "loss": 0.5161,
+      "step": 122
+    },
+    {
+      "epoch": 61.5,
+      "grad_norm": 4.064387798309326,
+      "learning_rate": 1e-06,
+      "loss": 0.6423,
+      "step": 123
+    },
+    {
+      "epoch": 62.0,
+      "grad_norm": 3.380586624145508,
+      "learning_rate": 1e-06,
+      "loss": 0.5178,
+      "step": 124
+    },
+    {
+      "epoch": 62.5,
+      "grad_norm": 3.088805675506592,
+      "learning_rate": 1e-06,
+      "loss": 0.5651,
+      "step": 125
+    },
+    {
+      "epoch": 63.0,
+      "grad_norm": 3.5499460697174072,
+      "learning_rate": 1e-06,
+      "loss": 0.5635,
+      "step": 126
+    },
+    {
+      "epoch": 63.5,
+      "grad_norm": 2.888054847717285,
+      "learning_rate": 1e-06,
+      "loss": 0.5304,
+      "step": 127
+    },
+    {
+      "epoch": 64.0,
+      "grad_norm": 3.3510334491729736,
+      "learning_rate": 1e-06,
+      "loss": 0.5247,
+      "step": 128
+    },
+    {
+      "epoch": 64.5,
+      "grad_norm": 5.040398597717285,
+      "learning_rate": 1e-06,
+      "loss": 0.4425,
+      "step": 129
+    },
+    {
+      "epoch": 65.0,
+      "grad_norm": 3.9134650230407715,
+      "learning_rate": 1e-06,
+      "loss": 0.4855,
+      "step": 130
+    },
+    {
+      "epoch": 65.5,
+      "grad_norm": 3.4794704914093018,
+      "learning_rate": 1e-06,
+      "loss": 0.523,
+      "step": 131
+    },
+    {
+      "epoch": 66.0,
+      "grad_norm": 2.619807243347168,
+      "learning_rate": 1e-06,
+      "loss": 0.4555,
+      "step": 132
+    },
+    {
+      "epoch": 66.5,
+      "grad_norm": 3.3289549350738525,
+      "learning_rate": 1e-06,
+      "loss": 0.4624,
+      "step": 133
+    },
+    {
+      "epoch": 67.0,
+      "grad_norm": 2.907986640930176,
+      "learning_rate": 1e-06,
+      "loss": 0.449,
+      "step": 134
+    },
+    {
+      "epoch": 67.5,
+      "grad_norm": 3.5478975772857666,
+      "learning_rate": 1e-06,
+      "loss": 0.4897,
+      "step": 135
+    },
+    {
+      "epoch": 68.0,
+      "grad_norm": 3.1988377571105957,
+      "learning_rate": 1e-06,
+      "loss": 0.4456,
+      "step": 136
+    },
+    {
+      "epoch": 68.5,
+      "grad_norm": 4.699775218963623,
+      "learning_rate": 1e-06,
+      "loss": 0.5154,
+      "step": 137
+    },
+    {
+      "epoch": 69.0,
+      "grad_norm": 3.398183584213257,
+      "learning_rate": 1e-06,
+      "loss": 0.4472,
+      "step": 138
+    },
+    {
+      "epoch": 69.5,
+      "grad_norm": 3.498777389526367,
+      "learning_rate": 1e-06,
+      "loss": 0.3895,
+      "step": 139
+    },
+    {
+      "epoch": 70.0,
+      "grad_norm": 3.4113645553588867,
+      "learning_rate": 1e-06,
+      "loss": 0.435,
+      "step": 140
+    },
+    {
+      "epoch": 70.5,
+      "grad_norm": 5.803523063659668,
+      "learning_rate": 1e-06,
+      "loss": 0.3916,
+      "step": 141
+    },
+    {
+      "epoch": 71.0,
+      "grad_norm": 3.308669328689575,
+      "learning_rate": 1e-06,
+      "loss": 0.3971,
+      "step": 142
+    },
+    {
+      "epoch": 71.5,
+      "grad_norm": 3.3992626667022705,
+      "learning_rate": 1e-06,
+      "loss": 0.399,
+      "step": 143
+    },
+    {
+      "epoch": 72.0,
+      "grad_norm": 4.154728889465332,
+      "learning_rate": 1e-06,
+      "loss": 0.3975,
+      "step": 144
+    },
+    {
+      "epoch": 72.5,
+      "grad_norm": 4.499520301818848,
+      "learning_rate": 1e-06,
+      "loss": 0.4032,
+      "step": 145
+    },
+    {
+      "epoch": 73.0,
+      "grad_norm": 3.4898831844329834,
+      "learning_rate": 1e-06,
+      "loss": 0.3774,
+      "step": 146
+    },
+    {
+      "epoch": 73.5,
+      "grad_norm": 3.3620753288269043,
+      "learning_rate": 1e-06,
+      "loss": 0.3527,
+      "step": 147
+    },
+    {
+      "epoch": 74.0,
+      "grad_norm": 3.7331016063690186,
+      "learning_rate": 1e-06,
+      "loss": 0.3784,
+      "step": 148
+    },
+    {
+      "epoch": 74.5,
+      "grad_norm": 3.6364359855651855,
+      "learning_rate": 1e-06,
+      "loss": 0.3857,
+      "step": 149
+    },
+    {
+      "epoch": 75.0,
+      "grad_norm": 3.118908405303955,
+      "learning_rate": 1e-06,
+      "loss": 0.3055,
+      "step": 150
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5000,
+  "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": 2.098224817063854e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-150/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b36bec9488b445350964041059369935c653f243588cf528b415499fc00dcca2
+size 6968

checkpoint-150/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)

checkpoint-200/README.md

@@ -0,0 +1,202 @@
+---
+base_model: openbmb/MiniCPM-V-2_6
+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 0.12.0

checkpoint-200/adapter_config.json

@@ -0,0 +1,33 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": {
+    "base_model_class": "MiniCPMV",
+    "parent_library": "transformers_modules.openbmb.MiniCPM-V-2_6.4e4be000cd81feda8b96d14b53f1791b4010b038.modeling_minicpmv"
+  },
+  "base_model_name_or_path": "openbmb/MiniCPM-V-2_6",
+  "bias": "none",
+  "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": 64,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": [
+    "embed_tokens",
+    "resampler",
+    "vpm"
+  ],
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": "llm\\..*layers\\.\\d+\\.self_attn\\.(q_proj|k_proj|v_proj|o_proj)",
+  "task_type": null,
+  "use_dora": false,
+  "use_rslora": false
+}

checkpoint-200/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:68bdf3edc72b018569a9f37258843e755e0e21d233ceafb66cc9e5504a9b7588
+size 2140796552

checkpoint-200/added_tokens.json

@@ -0,0 +1,25 @@
+{
+  "</box>": 151651,
+  "</image>": 151647,
+  "</image_id>": 151659,
+  "</point>": 151655,
+  "</quad>": 151653,
+  "</ref>": 151649,
+  "</slice>": 151657,
+  "<box>": 151650,
+  "<image>": 151646,
+  "<image_id>": 151658,
+  "<point>": 151654,
+  "<quad>": 151652,
+  "<ref>": 151648,
+  "<slice>": 151656,
+  "<|endoftext|>": 151643,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|reserved_special_token_0|>": 151660,
+  "<|reserved_special_token_1|>": 151661,
+  "<|reserved_special_token_2|>": 151662,
+  "<|reserved_special_token_3|>": 151663,
+  "<|reserved_special_token_4|>": 151664,
+  "<|reserved_special_token_5|>": 151665
+}

checkpoint-200/latest

@@ -0,0 +1 @@
+global_step200

checkpoint-200/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:14578ae87fba73684b3aaa347b784bf6ad87559a3346e00e346d8fd8c900a36e
+size 15024

checkpoint-200/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:acca0c430eb36d141333c665b26b567344cc638617fa76be7cbfa1da0733ce5d
+size 15024

checkpoint-200/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:783e9d405ba161da7cba4bb5e3f1dab4e2c323db0e03ad76789abf974f5aacca
+size 15024

checkpoint-200/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bc5b28179f49dbc021af9c51f84c3558f342539cbe23490b4a34198851baa403
+size 15024

checkpoint-200/special_tokens_map.json

@@ -0,0 +1,52 @@
+{
+  "additional_special_tokens": [
+    "<image>",
+    "</image>",
+    "<ref>",
+    "</ref>",
+    "<box>",
+    "</box>",
+    "<quad>",
+    "</quad>",
+    "<point>",
+    "</point>",
+    "<slice>",
+    "</slice>",
+    "<image_id>",
+    "</image_id>",
+    "<|reserved_special_token_0|>",
+    "<|reserved_special_token_1|>",
+    "<|reserved_special_token_2|>",
+    "<|reserved_special_token_3|>",
+    "<|reserved_special_token_4|>",
+    "<|reserved_special_token_5|>"
+  ],
+  "bos_token": {
+    "content": "<|im_start|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

checkpoint-200/tokenizer_config.json

@@ -0,0 +1,235 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "128244": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "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": "<image>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "</image>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<ref>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "</ref>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<box>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "</box>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<quad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "</quad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<point>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "</point>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<slice>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "</slice>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151658": {
+      "content": "<image_id>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151659": {
+      "content": "</image_id>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151660": {
+      "content": "<|reserved_special_token_0|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151661": {
+      "content": "<|reserved_special_token_1|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151662": {
+      "content": "<|reserved_special_token_2|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151663": {
+      "content": "<|reserved_special_token_3|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151664": {
+      "content": "<|reserved_special_token_4|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151665": {
+      "content": "<|reserved_special_token_5|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<image>",
+    "</image>",
+    "<ref>",
+    "</ref>",
+    "<box>",
+    "</box>",
+    "<quad>",
+    "</quad>",
+    "<point>",
+    "</point>",
+    "<slice>",
+    "</slice>",
+    "<image_id>",
+    "</image_id>",
+    "<|reserved_special_token_0|>",
+    "<|reserved_special_token_1|>",
+    "<|reserved_special_token_2|>",
+    "<|reserved_special_token_3|>",
+    "<|reserved_special_token_4|>",
+    "<|reserved_special_token_5|>"
+  ],
+  "auto_map": {
+    "AutoTokenizer": [
+      "openbmb/MiniCPM-V-2_6--tokenization_minicpmv_fast.MiniCPMVTokenizerFast",
+      null
+    ]
+  },
+  "bos_token": "<|im_start|>",
+  "chat_template": "{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n' }}{% endif %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "MiniCPMVTokenizer",
+  "unk_token": "<unk>"
+}

checkpoint-200/trainer_state.json

@@ -0,0 +1,1433 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 100.0,
+  "eval_steps": 1000,
+  "global_step": 200,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.0,
+      "learning_rate": 0,
+      "loss": 1.4695,
+      "step": 1
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.0,
+      "learning_rate": 0,
+      "loss": 1.4475,
+      "step": 2
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.0,
+      "learning_rate": 0,
+      "loss": 1.5187,
+      "step": 3
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.0,
+      "learning_rate": 0,
+      "loss": 1.3848,
+      "step": 4
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.0,
+      "learning_rate": 0,
+      "loss": 1.3918,
+      "step": 5
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 5.2773966789245605,
+      "learning_rate": 0.0,
+      "loss": 1.4477,
+      "step": 6
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 6.704706192016602,
+      "learning_rate": 1.5051499783199055e-07,
+      "loss": 1.5294,
+      "step": 7
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 18.195552825927734,
+      "learning_rate": 2.385606273598312e-07,
+      "loss": 1.4874,
+      "step": 8
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 4.960075378417969,
+      "learning_rate": 3.010299956639811e-07,
+      "loss": 1.2974,
+      "step": 9
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 7.742049217224121,
+      "learning_rate": 3.494850021680093e-07,
+      "loss": 1.5687,
+      "step": 10
+    },
+    {
+      "epoch": 5.5,
+      "grad_norm": 8.554511070251465,
+      "learning_rate": 3.8907562519182173e-07,
+      "loss": 1.3764,
+      "step": 11
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 6.75316047668457,
+      "learning_rate": 4.2254902000712834e-07,
+      "loss": 1.509,
+      "step": 12
+    },
+    {
+      "epoch": 6.5,
+      "grad_norm": 7.616246700286865,
+      "learning_rate": 4.5154499349597166e-07,
+      "loss": 1.4578,
+      "step": 13
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 5.040093421936035,
+      "learning_rate": 4.771212547196623e-07,
+      "loss": 1.4536,
+      "step": 14
+    },
+    {
+      "epoch": 7.5,
+      "grad_norm": 7.984886646270752,
+      "learning_rate": 4.999999999999999e-07,
+      "loss": 1.4045,
+      "step": 15
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 7.743239879608154,
+      "learning_rate": 5.206963425791124e-07,
+      "loss": 1.4374,
+      "step": 16
+    },
+    {
+      "epoch": 8.5,
+      "grad_norm": 8.207304000854492,
+      "learning_rate": 5.395906230238123e-07,
+      "loss": 1.5724,
+      "step": 17
+    },
+    {
+      "epoch": 9.0,
+      "grad_norm": 6.423882961273193,
+      "learning_rate": 5.569716761534182e-07,
+      "loss": 1.4628,
+      "step": 18
+    },
+    {
+      "epoch": 9.5,
+      "grad_norm": 5.485311985015869,
+      "learning_rate": 5.730640178391189e-07,
+      "loss": 1.4199,
+      "step": 19
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 5.854134559631348,
+      "learning_rate": 5.880456295278405e-07,
+      "loss": 1.3805,
+      "step": 20
+    },
+    {
+      "epoch": 10.5,
+      "grad_norm": 6.40692663192749,
+      "learning_rate": 6.020599913279622e-07,
+      "loss": 1.5921,
+      "step": 21
+    },
+    {
+      "epoch": 11.0,
+      "grad_norm": 4.778651237487793,
+      "learning_rate": 6.15224460689137e-07,
+      "loss": 1.4138,
+      "step": 22
+    },
+    {
+      "epoch": 11.5,
+      "grad_norm": 6.353730201721191,
+      "learning_rate": 6.276362525516529e-07,
+      "loss": 1.421,
+      "step": 23
+    },
+    {
+      "epoch": 12.0,
+      "grad_norm": 3.764188051223755,
+      "learning_rate": 6.393768004764143e-07,
+      "loss": 1.272,
+      "step": 24
+    },
+    {
+      "epoch": 12.5,
+      "grad_norm": 4.659372806549072,
+      "learning_rate": 6.505149978319905e-07,
+      "loss": 1.5793,
+      "step": 25
+    },
+    {
+      "epoch": 13.0,
+      "grad_norm": 3.8088481426239014,
+      "learning_rate": 6.611096473669595e-07,
+      "loss": 1.2148,
+      "step": 26
+    },
+    {
+      "epoch": 13.5,
+      "grad_norm": 4.605009078979492,
+      "learning_rate": 6.712113404111031e-07,
+      "loss": 1.4131,
+      "step": 27
+    },
+    {
+      "epoch": 14.0,
+      "grad_norm": 4.647908687591553,
+      "learning_rate": 6.808639180087963e-07,
+      "loss": 1.2516,
+      "step": 28
+    },
+    {
+      "epoch": 14.5,
+      "grad_norm": 4.514801502227783,
+      "learning_rate": 6.901056208558029e-07,
+      "loss": 1.3228,
+      "step": 29
+    },
+    {
+      "epoch": 15.0,
+      "grad_norm": 4.330278396606445,
+      "learning_rate": 6.989700043360186e-07,
+      "loss": 1.3317,
+      "step": 30
+    },
+    {
+      "epoch": 15.5,
+      "grad_norm": 5.9373321533203125,
+      "learning_rate": 7.074866739854088e-07,
+      "loss": 1.4139,
+      "step": 31
+    },
+    {
+      "epoch": 16.0,
+      "grad_norm": 4.761633396148682,
+      "learning_rate": 7.156818820794935e-07,
+      "loss": 1.3568,
+      "step": 32
+    },
+    {
+      "epoch": 16.5,
+      "grad_norm": 4.071909427642822,
+      "learning_rate": 7.235790156711094e-07,
+      "loss": 1.314,
+      "step": 33
+    },
+    {
+      "epoch": 17.0,
+      "grad_norm": 4.649173259735107,
+      "learning_rate": 7.311989989494779e-07,
+      "loss": 1.4079,
+      "step": 34
+    },
+    {
+      "epoch": 17.5,
+      "grad_norm": 3.983167886734009,
+      "learning_rate": 7.38560627359831e-07,
+      "loss": 1.2689,
+      "step": 35
+    },
+    {
+      "epoch": 18.0,
+      "grad_norm": 4.708627700805664,
+      "learning_rate": 7.456808469171361e-07,
+      "loss": 1.2591,
+      "step": 36
+    },
+    {
+      "epoch": 18.5,
+      "grad_norm": 4.3896260261535645,
+      "learning_rate": 7.525749891599529e-07,
+      "loss": 1.2727,
+      "step": 37
+    },
+    {
+      "epoch": 19.0,
+      "grad_norm": 3.192983627319336,
+      "learning_rate": 7.592569699389436e-07,
+      "loss": 1.2811,
+      "step": 38
+    },
+    {
+      "epoch": 19.5,
+      "grad_norm": 4.576615810394287,
+      "learning_rate": 7.657394585211274e-07,
+      "loss": 1.2692,
+      "step": 39
+    },
+    {
+      "epoch": 20.0,
+      "grad_norm": 6.580528736114502,
+      "learning_rate": 7.720340221751376e-07,
+      "loss": 1.2739,
+      "step": 40
+    },
+    {
+      "epoch": 20.5,
+      "grad_norm": 3.075108528137207,
+      "learning_rate": 7.781512503836435e-07,
+      "loss": 1.1164,
+      "step": 41
+    },
+    {
+      "epoch": 21.0,
+      "grad_norm": 4.4012041091918945,
+      "learning_rate": 7.841008620334974e-07,
+      "loss": 1.3188,
+      "step": 42
+    },
+    {
+      "epoch": 21.5,
+      "grad_norm": 4.855476379394531,
+      "learning_rate": 7.89891798308405e-07,
+      "loss": 1.113,
+      "step": 43
+    },
+    {
+      "epoch": 22.0,
+      "grad_norm": 5.26778507232666,
+      "learning_rate": 7.955323035132494e-07,
+      "loss": 1.3752,
+      "step": 44
+    },
+    {
+      "epoch": 22.5,
+      "grad_norm": 3.732664108276367,
+      "learning_rate": 8.01029995663981e-07,
+      "loss": 1.251,
+      "step": 45
+    },
+    {
+      "epoch": 23.0,
+      "grad_norm": 3.8093373775482178,
+      "learning_rate": 8.063919283598676e-07,
+      "loss": 1.1708,
+      "step": 46
+    },
+    {
+      "epoch": 23.5,
+      "grad_norm": 5.319915771484375,
+      "learning_rate": 8.116246451989502e-07,
+      "loss": 1.292,
+      "step": 47
+    },
+    {
+      "epoch": 24.0,
+      "grad_norm": 3.3320486545562744,
+      "learning_rate": 8.16734227789793e-07,
+      "loss": 1.1096,
+      "step": 48
+    },
+    {
+      "epoch": 24.5,
+      "grad_norm": 4.5503411293029785,
+      "learning_rate": 8.217263382430935e-07,
+      "loss": 1.0945,
+      "step": 49
+    },
+    {
+      "epoch": 25.0,
+      "grad_norm": 4.538294792175293,
+      "learning_rate": 8.266062568876716e-07,
+      "loss": 1.18,
+      "step": 50
+    },
+    {
+      "epoch": 25.5,
+      "grad_norm": 4.297428131103516,
+      "learning_rate": 8.313789158407869e-07,
+      "loss": 1.217,
+      "step": 51
+    },
+    {
+      "epoch": 26.0,
+      "grad_norm": 5.326357364654541,
+      "learning_rate": 8.360489289678585e-07,
+      "loss": 1.1089,
+      "step": 52
+    },
+    {
+      "epoch": 26.5,
+      "grad_norm": 22.330251693725586,
+      "learning_rate": 8.406206186877934e-07,
+      "loss": 1.2477,
+      "step": 53
+    },
+    {
+      "epoch": 27.0,
+      "grad_norm": 3.5450849533081055,
+      "learning_rate": 8.450980400142567e-07,
+      "loss": 1.0081,
+      "step": 54
+    },
+    {
+      "epoch": 27.5,
+      "grad_norm": 3.588421106338501,
+      "learning_rate": 8.494850021680092e-07,
+      "loss": 1.1557,
+      "step": 55
+    },
+    {
+      "epoch": 28.0,
+      "grad_norm": 5.4207987785339355,
+      "learning_rate": 8.53785088048968e-07,
+      "loss": 1.2518,
+      "step": 56
+    },
+    {
+      "epoch": 28.5,
+      "grad_norm": 9.896635055541992,
+      "learning_rate": 8.580016718173995e-07,
+      "loss": 1.0695,
+      "step": 57
+    },
+    {
+      "epoch": 29.0,
+      "grad_norm": 6.4676666259765625,
+      "learning_rate": 8.621379348003944e-07,
+      "loss": 1.2276,
+      "step": 58
+    },
+    {
+      "epoch": 29.5,
+      "grad_norm": 4.066528797149658,
+      "learning_rate": 8.661968799114842e-07,
+      "loss": 1.0339,
+      "step": 59
+    },
+    {
+      "epoch": 30.0,
+      "grad_norm": 4.463730812072754,
+      "learning_rate": 8.701813447471218e-07,
+      "loss": 1.0844,
+      "step": 60
+    },
+    {
+      "epoch": 30.5,
+      "grad_norm": 3.523402690887451,
+      "learning_rate": 8.740940135031001e-07,
+      "loss": 1.0907,
+      "step": 61
+    },
+    {
+      "epoch": 31.0,
+      "grad_norm": 4.35223913192749,
+      "learning_rate": 8.779374278362456e-07,
+      "loss": 1.1098,
+      "step": 62
+    },
+    {
+      "epoch": 31.5,
+      "grad_norm": 9.153762817382812,
+      "learning_rate": 8.817139967814684e-07,
+      "loss": 1.1932,
+      "step": 63
+    },
+    {
+      "epoch": 32.0,
+      "grad_norm": 3.6358563899993896,
+      "learning_rate": 8.854260058210719e-07,
+      "loss": 0.984,
+      "step": 64
+    },
+    {
+      "epoch": 32.5,
+      "grad_norm": 3.1282856464385986,
+      "learning_rate": 8.890756251918216e-07,
+      "loss": 1.004,
+      "step": 65
+    },
+    {
+      "epoch": 33.0,
+      "grad_norm": 3.8792643547058105,
+      "learning_rate": 8.926649175053833e-07,
+      "loss": 1.0181,
+      "step": 66
+    },
+    {
+      "epoch": 33.5,
+      "grad_norm": 3.481888771057129,
+      "learning_rate": 8.961958447491268e-07,
+      "loss": 1.0099,
+      "step": 67
+    },
+    {
+      "epoch": 34.0,
+      "grad_norm": 4.767885684967041,
+      "learning_rate": 8.996702747267907e-07,
+      "loss": 0.99,
+      "step": 68
+    },
+    {
+      "epoch": 34.5,
+      "grad_norm": 6.9532952308654785,
+      "learning_rate": 9.030899869919433e-07,
+      "loss": 0.9372,
+      "step": 69
+    },
+    {
+      "epoch": 35.0,
+      "grad_norm": 5.427297592163086,
+      "learning_rate": 9.064566783214276e-07,
+      "loss": 1.023,
+      "step": 70
+    },
+    {
+      "epoch": 35.5,
+      "grad_norm": 3.7851903438568115,
+      "learning_rate": 9.097719677709341e-07,
+      "loss": 0.9984,
+      "step": 71
+    },
+    {
+      "epoch": 36.0,
+      "grad_norm": 3.750429630279541,
+      "learning_rate": 9.13037401350413e-07,
+      "loss": 0.9991,
+      "step": 72
+    },
+    {
+      "epoch": 36.5,
+      "grad_norm": 3.222670793533325,
+      "learning_rate": 9.162544563531181e-07,
+      "loss": 0.8962,
+      "step": 73
+    },
+    {
+      "epoch": 37.0,
+      "grad_norm": 2.7355740070343018,
+      "learning_rate": 9.194245453686276e-07,
+      "loss": 1.0381,
+      "step": 74
+    },
+    {
+      "epoch": 37.5,
+      "grad_norm": 6.130687236785889,
+      "learning_rate": 9.225490200071283e-07,
+      "loss": 0.943,
+      "step": 75
+    },
+    {
+      "epoch": 38.0,
+      "grad_norm": 4.569997310638428,
+      "learning_rate": 9.256291743595375e-07,
+      "loss": 1.0464,
+      "step": 76
+    },
+    {
+      "epoch": 38.5,
+      "grad_norm": 4.291537761688232,
+      "learning_rate": 9.28666248215634e-07,
+      "loss": 0.965,
+      "step": 77
+    },
+    {
+      "epoch": 39.0,
+      "grad_norm": 3.6804943084716797,
+      "learning_rate": 9.316614300602277e-07,
+      "loss": 0.9621,
+      "step": 78
+    },
+    {
+      "epoch": 39.5,
+      "grad_norm": 3.408571720123291,
+      "learning_rate": 9.346158598654879e-07,
+      "loss": 0.9144,
+      "step": 79
+    },
+    {
+      "epoch": 40.0,
+      "grad_norm": 3.565375566482544,
+      "learning_rate": 9.375306316958498e-07,
+      "loss": 0.8959,
+      "step": 80
+    },
+    {
+      "epoch": 40.5,
+      "grad_norm": 5.186458110809326,
+      "learning_rate": 9.404067961403955e-07,
+      "loss": 0.9522,
+      "step": 81
+    },
+    {
+      "epoch": 41.0,
+      "grad_norm": 3.754727602005005,
+      "learning_rate": 9.432453625862408e-07,
+      "loss": 0.8659,
+      "step": 82
+    },
+    {
+      "epoch": 41.5,
+      "grad_norm": 7.740540504455566,
+      "learning_rate": 9.4604730134524e-07,
+      "loss": 0.8567,
+      "step": 83
+    },
+    {
+      "epoch": 42.0,
+      "grad_norm": 5.675983905792236,
+      "learning_rate": 9.488135456452205e-07,
+      "loss": 0.9199,
+      "step": 84
+    },
+    {
+      "epoch": 42.5,
+      "grad_norm": 4.611757755279541,
+      "learning_rate": 9.515449934959715e-07,
+      "loss": 0.9635,
+      "step": 85
+    },
+    {
+      "epoch": 43.0,
+      "grad_norm": 2.9257593154907227,
+      "learning_rate": 9.542425094393247e-07,
+      "loss": 0.9204,
+      "step": 86
+    },
+    {
+      "epoch": 43.5,
+      "grad_norm": 3.5699961185455322,
+      "learning_rate": 9.569069261918583e-07,
+      "loss": 0.9794,
+      "step": 87
+    },
+    {
+      "epoch": 44.0,
+      "grad_norm": 4.43387508392334,
+      "learning_rate": 9.59539046188037e-07,
+      "loss": 0.8352,
+      "step": 88
+    },
+    {
+      "epoch": 44.5,
+      "grad_norm": 3.140352487564087,
+      "learning_rate": 9.621396430309406e-07,
+      "loss": 0.8887,
+      "step": 89
+    },
+    {
+      "epoch": 45.0,
+      "grad_norm": 5.941178798675537,
+      "learning_rate": 9.647094628571462e-07,
+      "loss": 0.8571,
+      "step": 90
+    },
+    {
+      "epoch": 45.5,
+      "grad_norm": 2.8487863540649414,
+      "learning_rate": 9.672492256217836e-07,
+      "loss": 0.8364,
+      "step": 91
+    },
+    {
+      "epoch": 46.0,
+      "grad_norm": 2.790127754211426,
+      "learning_rate": 9.69759626309309e-07,
+      "loss": 0.8898,
+      "step": 92
+    },
+    {
+      "epoch": 46.5,
+      "grad_norm": 2.601271629333496,
+      "learning_rate": 9.722413360750842e-07,
+      "loss": 0.8499,
+      "step": 93
+    },
+    {
+      "epoch": 47.0,
+      "grad_norm": 6.0029296875,
+      "learning_rate": 9.74695003322456e-07,
+      "loss": 0.7976,
+      "step": 94
+    },
+    {
+      "epoch": 47.5,
+      "grad_norm": 3.055148124694824,
+      "learning_rate": 9.771212547196622e-07,
+      "loss": 0.7946,
+      "step": 95
+    },
+    {
+      "epoch": 48.0,
+      "grad_norm": 5.841665744781494,
+      "learning_rate": 9.795206961605466e-07,
+      "loss": 0.8136,
+      "step": 96
+    },
+    {
+      "epoch": 48.5,
+      "grad_norm": 3.1043038368225098,
+      "learning_rate": 9.818939136727774e-07,
+      "loss": 0.7518,
+      "step": 97
+    },
+    {
+      "epoch": 49.0,
+      "grad_norm": 5.523963928222656,
+      "learning_rate": 9.842414742769674e-07,
+      "loss": 0.8265,
+      "step": 98
+    },
+    {
+      "epoch": 49.5,
+      "grad_norm": 2.8792777061462402,
+      "learning_rate": 9.865639267998492e-07,
+      "loss": 0.7474,
+      "step": 99
+    },
+    {
+      "epoch": 50.0,
+      "grad_norm": 2.840219497680664,
+      "learning_rate": 9.888618026444236e-07,
+      "loss": 0.784,
+      "step": 100
+    },
+    {
+      "epoch": 50.5,
+      "grad_norm": 2.887335777282715,
+      "learning_rate": 9.91135616519784e-07,
+      "loss": 0.7575,
+      "step": 101
+    },
+    {
+      "epoch": 51.0,
+      "grad_norm": 2.592064619064331,
+      "learning_rate": 9.933858671331222e-07,
+      "loss": 0.7871,
+      "step": 102
+    },
+    {
+      "epoch": 51.5,
+      "grad_norm": 2.881988048553467,
+      "learning_rate": 9.956130378462473e-07,
+      "loss": 0.8135,
+      "step": 103
+    },
+    {
+      "epoch": 52.0,
+      "grad_norm": 3.1426312923431396,
+      "learning_rate": 9.978175972987748e-07,
+      "loss": 0.6766,
+      "step": 104
+    },
+    {
+      "epoch": 52.5,
+      "grad_norm": 2.951082229614258,
+      "learning_rate": 9.999999999999997e-07,
+      "loss": 0.6936,
+      "step": 105
+    },
+    {
+      "epoch": 53.0,
+      "grad_norm": 5.04337215423584,
+      "learning_rate": 1e-06,
+      "loss": 0.7773,
+      "step": 106
+    },
+    {
+      "epoch": 53.5,
+      "grad_norm": 3.2338459491729736,
+      "learning_rate": 1e-06,
+      "loss": 0.833,
+      "step": 107
+    },
+    {
+      "epoch": 54.0,
+      "grad_norm": 2.7215511798858643,
+      "learning_rate": 1e-06,
+      "loss": 0.6974,
+      "step": 108
+    },
+    {
+      "epoch": 54.5,
+      "grad_norm": 2.6445066928863525,
+      "learning_rate": 1e-06,
+      "loss": 0.7189,
+      "step": 109
+    },
+    {
+      "epoch": 55.0,
+      "grad_norm": 3.2087759971618652,
+      "learning_rate": 1e-06,
+      "loss": 0.6723,
+      "step": 110
+    },
+    {
+      "epoch": 55.5,
+      "grad_norm": 3.1238620281219482,
+      "learning_rate": 1e-06,
+      "loss": 0.6299,
+      "step": 111
+    },
+    {
+      "epoch": 56.0,
+      "grad_norm": 5.280316352844238,
+      "learning_rate": 1e-06,
+      "loss": 0.7757,
+      "step": 112
+    },
+    {
+      "epoch": 56.5,
+      "grad_norm": 4.072208881378174,
+      "learning_rate": 1e-06,
+      "loss": 0.6786,
+      "step": 113
+    },
+    {
+      "epoch": 57.0,
+      "grad_norm": 3.1515462398529053,
+      "learning_rate": 1e-06,
+      "loss": 0.5982,
+      "step": 114
+    },
+    {
+      "epoch": 57.5,
+      "grad_norm": 2.891411781311035,
+      "learning_rate": 1e-06,
+      "loss": 0.575,
+      "step": 115
+    },
+    {
+      "epoch": 58.0,
+      "grad_norm": 5.040134429931641,
+      "learning_rate": 1e-06,
+      "loss": 0.6497,
+      "step": 116
+    },
+    {
+      "epoch": 58.5,
+      "grad_norm": 2.7814080715179443,
+      "learning_rate": 1e-06,
+      "loss": 0.5745,
+      "step": 117
+    },
+    {
+      "epoch": 59.0,
+      "grad_norm": 3.400952100753784,
+      "learning_rate": 1e-06,
+      "loss": 0.6237,
+      "step": 118
+    },
+    {
+      "epoch": 59.5,
+      "grad_norm": 2.8887929916381836,
+      "learning_rate": 1e-06,
+      "loss": 0.5832,
+      "step": 119
+    },
+    {
+      "epoch": 60.0,
+      "grad_norm": 5.328385353088379,
+      "learning_rate": 1e-06,
+      "loss": 0.5677,
+      "step": 120
+    },
+    {
+      "epoch": 60.5,
+      "grad_norm": 3.790076494216919,
+      "learning_rate": 1e-06,
+      "loss": 0.5726,
+      "step": 121
+    },
+    {
+      "epoch": 61.0,
+      "grad_norm": 3.321824550628662,
+      "learning_rate": 1e-06,
+      "loss": 0.5161,
+      "step": 122
+    },
+    {
+      "epoch": 61.5,
+      "grad_norm": 4.064387798309326,
+      "learning_rate": 1e-06,
+      "loss": 0.6423,
+      "step": 123
+    },
+    {
+      "epoch": 62.0,
+      "grad_norm": 3.380586624145508,
+      "learning_rate": 1e-06,
+      "loss": 0.5178,
+      "step": 124
+    },
+    {
+      "epoch": 62.5,
+      "grad_norm": 3.088805675506592,
+      "learning_rate": 1e-06,
+      "loss": 0.5651,
+      "step": 125
+    },
+    {
+      "epoch": 63.0,
+      "grad_norm": 3.5499460697174072,
+      "learning_rate": 1e-06,
+      "loss": 0.5635,
+      "step": 126
+    },
+    {
+      "epoch": 63.5,
+      "grad_norm": 2.888054847717285,
+      "learning_rate": 1e-06,
+      "loss": 0.5304,
+      "step": 127
+    },
+    {
+      "epoch": 64.0,
+      "grad_norm": 3.3510334491729736,
+      "learning_rate": 1e-06,
+      "loss": 0.5247,
+      "step": 128
+    },
+    {
+      "epoch": 64.5,
+      "grad_norm": 5.040398597717285,
+      "learning_rate": 1e-06,
+      "loss": 0.4425,
+      "step": 129
+    },
+    {
+      "epoch": 65.0,
+      "grad_norm": 3.9134650230407715,
+      "learning_rate": 1e-06,
+      "loss": 0.4855,
+      "step": 130
+    },
+    {
+      "epoch": 65.5,
+      "grad_norm": 3.4794704914093018,
+      "learning_rate": 1e-06,
+      "loss": 0.523,
+      "step": 131
+    },
+    {
+      "epoch": 66.0,
+      "grad_norm": 2.619807243347168,
+      "learning_rate": 1e-06,
+      "loss": 0.4555,
+      "step": 132
+    },
+    {
+      "epoch": 66.5,
+      "grad_norm": 3.3289549350738525,
+      "learning_rate": 1e-06,
+      "loss": 0.4624,
+      "step": 133
+    },
+    {
+      "epoch": 67.0,
+      "grad_norm": 2.907986640930176,
+      "learning_rate": 1e-06,
+      "loss": 0.449,
+      "step": 134
+    },
+    {
+      "epoch": 67.5,
+      "grad_norm": 3.5478975772857666,
+      "learning_rate": 1e-06,
+      "loss": 0.4897,
+      "step": 135
+    },
+    {
+      "epoch": 68.0,
+      "grad_norm": 3.1988377571105957,
+      "learning_rate": 1e-06,
+      "loss": 0.4456,
+      "step": 136
+    },
+    {
+      "epoch": 68.5,
+      "grad_norm": 4.699775218963623,
+      "learning_rate": 1e-06,
+      "loss": 0.5154,
+      "step": 137
+    },
+    {
+      "epoch": 69.0,
+      "grad_norm": 3.398183584213257,
+      "learning_rate": 1e-06,
+      "loss": 0.4472,
+      "step": 138
+    },
+    {
+      "epoch": 69.5,
+      "grad_norm": 3.498777389526367,
+      "learning_rate": 1e-06,
+      "loss": 0.3895,
+      "step": 139
+    },
+    {
+      "epoch": 70.0,
+      "grad_norm": 3.4113645553588867,
+      "learning_rate": 1e-06,
+      "loss": 0.435,
+      "step": 140
+    },
+    {
+      "epoch": 70.5,
+      "grad_norm": 5.803523063659668,
+      "learning_rate": 1e-06,
+      "loss": 0.3916,
+      "step": 141
+    },
+    {
+      "epoch": 71.0,
+      "grad_norm": 3.308669328689575,
+      "learning_rate": 1e-06,
+      "loss": 0.3971,
+      "step": 142
+    },
+    {
+      "epoch": 71.5,
+      "grad_norm": 3.3992626667022705,
+      "learning_rate": 1e-06,
+      "loss": 0.399,
+      "step": 143
+    },
+    {
+      "epoch": 72.0,
+      "grad_norm": 4.154728889465332,
+      "learning_rate": 1e-06,
+      "loss": 0.3975,
+      "step": 144
+    },
+    {
+      "epoch": 72.5,
+      "grad_norm": 4.499520301818848,
+      "learning_rate": 1e-06,
+      "loss": 0.4032,
+      "step": 145
+    },
+    {
+      "epoch": 73.0,
+      "grad_norm": 3.4898831844329834,
+      "learning_rate": 1e-06,
+      "loss": 0.3774,
+      "step": 146
+    },
+    {
+      "epoch": 73.5,
+      "grad_norm": 3.3620753288269043,
+      "learning_rate": 1e-06,
+      "loss": 0.3527,
+      "step": 147
+    },
+    {
+      "epoch": 74.0,
+      "grad_norm": 3.7331016063690186,
+      "learning_rate": 1e-06,
+      "loss": 0.3784,
+      "step": 148
+    },
+    {
+      "epoch": 74.5,
+      "grad_norm": 3.6364359855651855,
+      "learning_rate": 1e-06,
+      "loss": 0.3857,
+      "step": 149
+    },
+    {
+      "epoch": 75.0,
+      "grad_norm": 3.118908405303955,
+      "learning_rate": 1e-06,
+      "loss": 0.3055,
+      "step": 150
+    },
+    {
+      "epoch": 75.5,
+      "grad_norm": 4.4062981605529785,
+      "learning_rate": 1e-06,
+      "loss": 0.3657,
+      "step": 151
+    },
+    {
+      "epoch": 76.0,
+      "grad_norm": 3.2687151432037354,
+      "learning_rate": 1e-06,
+      "loss": 0.3153,
+      "step": 152
+    },
+    {
+      "epoch": 76.5,
+      "grad_norm": 3.214362144470215,
+      "learning_rate": 1e-06,
+      "loss": 0.3302,
+      "step": 153
+    },
+    {
+      "epoch": 77.0,
+      "grad_norm": 4.000756740570068,
+      "learning_rate": 1e-06,
+      "loss": 0.3273,
+      "step": 154
+    },
+    {
+      "epoch": 77.5,
+      "grad_norm": 3.965745449066162,
+      "learning_rate": 1e-06,
+      "loss": 0.2978,
+      "step": 155
+    },
+    {
+      "epoch": 78.0,
+      "grad_norm": 6.183568477630615,
+      "learning_rate": 1e-06,
+      "loss": 0.2978,
+      "step": 156
+    },
+    {
+      "epoch": 78.5,
+      "grad_norm": 4.420961856842041,
+      "learning_rate": 1e-06,
+      "loss": 0.2647,
+      "step": 157
+    },
+    {
+      "epoch": 79.0,
+      "grad_norm": 4.480051517486572,
+      "learning_rate": 1e-06,
+      "loss": 0.3208,
+      "step": 158
+    },
+    {
+      "epoch": 79.5,
+      "grad_norm": 4.489965915679932,
+      "learning_rate": 1e-06,
+      "loss": 0.2445,
+      "step": 159
+    },
+    {
+      "epoch": 80.0,
+      "grad_norm": 3.858898878097534,
+      "learning_rate": 1e-06,
+      "loss": 0.2722,
+      "step": 160
+    },
+    {
+      "epoch": 80.5,
+      "grad_norm": 3.5918991565704346,
+      "learning_rate": 1e-06,
+      "loss": 0.2409,
+      "step": 161
+    },
+    {
+      "epoch": 81.0,
+      "grad_norm": 3.5857317447662354,
+      "learning_rate": 1e-06,
+      "loss": 0.2819,
+      "step": 162
+    },
+    {
+      "epoch": 81.5,
+      "grad_norm": 3.174011468887329,
+      "learning_rate": 1e-06,
+      "loss": 0.2632,
+      "step": 163
+    },
+    {
+      "epoch": 82.0,
+      "grad_norm": 3.2102479934692383,
+      "learning_rate": 1e-06,
+      "loss": 0.2651,
+      "step": 164
+    },
+    {
+      "epoch": 82.5,
+      "grad_norm": 3.1816534996032715,
+      "learning_rate": 1e-06,
+      "loss": 0.2294,
+      "step": 165
+    },
+    {
+      "epoch": 83.0,
+      "grad_norm": 4.917560577392578,
+      "learning_rate": 1e-06,
+      "loss": 0.2352,
+      "step": 166
+    },
+    {
+      "epoch": 83.5,
+      "grad_norm": 3.8099558353424072,
+      "learning_rate": 1e-06,
+      "loss": 0.2136,
+      "step": 167
+    },
+    {
+      "epoch": 84.0,
+      "grad_norm": 4.569125652313232,
+      "learning_rate": 1e-06,
+      "loss": 0.2114,
+      "step": 168
+    },
+    {
+      "epoch": 84.5,
+      "grad_norm": 3.7342379093170166,
+      "learning_rate": 1e-06,
+      "loss": 0.2433,
+      "step": 169
+    },
+    {
+      "epoch": 85.0,
+      "grad_norm": 4.781056880950928,
+      "learning_rate": 1e-06,
+      "loss": 0.1591,
+      "step": 170
+    },
+    {
+      "epoch": 85.5,
+      "grad_norm": 3.466123342514038,
+      "learning_rate": 1e-06,
+      "loss": 0.1865,
+      "step": 171
+    },
+    {
+      "epoch": 86.0,
+      "grad_norm": 3.0353050231933594,
+      "learning_rate": 1e-06,
+      "loss": 0.1781,
+      "step": 172
+    },
+    {
+      "epoch": 86.5,
+      "grad_norm": 6.922516822814941,
+      "learning_rate": 1e-06,
+      "loss": 0.2116,
+      "step": 173
+    },
+    {
+      "epoch": 87.0,
+      "grad_norm": 2.7158310413360596,
+      "learning_rate": 1e-06,
+      "loss": 0.1633,
+      "step": 174
+    },
+    {
+      "epoch": 87.5,
+      "grad_norm": 2.4507789611816406,
+      "learning_rate": 1e-06,
+      "loss": 0.1576,
+      "step": 175
+    },
+    {
+      "epoch": 88.0,
+      "grad_norm": 3.600067615509033,
+      "learning_rate": 1e-06,
+      "loss": 0.1651,
+      "step": 176
+    },
+    {
+      "epoch": 88.5,
+      "grad_norm": 8.590092658996582,
+      "learning_rate": 1e-06,
+      "loss": 0.1463,
+      "step": 177
+    },
+    {
+      "epoch": 89.0,
+      "grad_norm": 2.5606284141540527,
+      "learning_rate": 1e-06,
+      "loss": 0.1501,
+      "step": 178
+    },
+    {
+      "epoch": 89.5,
+      "grad_norm": 3.4214353561401367,
+      "learning_rate": 1e-06,
+      "loss": 0.1603,
+      "step": 179
+    },
+    {
+      "epoch": 90.0,
+      "grad_norm": 3.066343069076538,
+      "learning_rate": 1e-06,
+      "loss": 0.1598,
+      "step": 180
+    },
+    {
+      "epoch": 90.5,
+      "grad_norm": 2.48077654838562,
+      "learning_rate": 1e-06,
+      "loss": 0.1411,
+      "step": 181
+    },
+    {
+      "epoch": 91.0,
+      "grad_norm": 2.7177133560180664,
+      "learning_rate": 1e-06,
+      "loss": 0.1223,
+      "step": 182
+    },
+    {
+      "epoch": 91.5,
+      "grad_norm": 2.207670211791992,
+      "learning_rate": 1e-06,
+      "loss": 0.106,
+      "step": 183
+    },
+    {
+      "epoch": 92.0,
+      "grad_norm": 2.798039674758911,
+      "learning_rate": 1e-06,
+      "loss": 0.122,
+      "step": 184
+    },
+    {
+      "epoch": 92.5,
+      "grad_norm": 2.9118175506591797,
+      "learning_rate": 1e-06,
+      "loss": 0.1114,
+      "step": 185
+    },
+    {
+      "epoch": 93.0,
+      "grad_norm": 3.4291181564331055,
+      "learning_rate": 1e-06,
+      "loss": 0.0851,
+      "step": 186
+    },
+    {
+      "epoch": 93.5,
+      "grad_norm": 2.244668960571289,
+      "learning_rate": 1e-06,
+      "loss": 0.0966,
+      "step": 187
+    },
+    {
+      "epoch": 94.0,
+      "grad_norm": 2.280467987060547,
+      "learning_rate": 1e-06,
+      "loss": 0.0949,
+      "step": 188
+    },
+    {
+      "epoch": 94.5,
+      "grad_norm": 2.1926937103271484,
+      "learning_rate": 1e-06,
+      "loss": 0.1037,
+      "step": 189
+    },
+    {
+      "epoch": 95.0,
+      "grad_norm": 2.3507280349731445,
+      "learning_rate": 1e-06,
+      "loss": 0.0843,
+      "step": 190
+    },
+    {
+      "epoch": 95.5,
+      "grad_norm": 1.7736583948135376,
+      "learning_rate": 1e-06,
+      "loss": 0.067,
+      "step": 191
+    },
+    {
+      "epoch": 96.0,
+      "grad_norm": 1.8927035331726074,
+      "learning_rate": 1e-06,
+      "loss": 0.1017,
+      "step": 192
+    },
+    {
+      "epoch": 96.5,
+      "grad_norm": 5.801133155822754,
+      "learning_rate": 1e-06,
+      "loss": 0.0744,
+      "step": 193
+    },
+    {
+      "epoch": 97.0,
+      "grad_norm": 4.504685878753662,
+      "learning_rate": 1e-06,
+      "loss": 0.0777,
+      "step": 194
+    },
+    {
+      "epoch": 97.5,
+      "grad_norm": 1.8973733186721802,
+      "learning_rate": 1e-06,
+      "loss": 0.0852,
+      "step": 195
+    },
+    {
+      "epoch": 98.0,
+      "grad_norm": 1.8072805404663086,
+      "learning_rate": 1e-06,
+      "loss": 0.0669,
+      "step": 196
+    },
+    {
+      "epoch": 98.5,
+      "grad_norm": 2.0617663860321045,
+      "learning_rate": 1e-06,
+      "loss": 0.0748,
+      "step": 197
+    },
+    {
+      "epoch": 99.0,
+      "grad_norm": 1.3142532110214233,
+      "learning_rate": 1e-06,
+      "loss": 0.0572,
+      "step": 198
+    },
+    {
+      "epoch": 99.5,
+      "grad_norm": 2.0814568996429443,
+      "learning_rate": 1e-06,
+      "loss": 0.06,
+      "step": 199
+    },
+    {
+      "epoch": 100.0,
+      "grad_norm": 2.759458065032959,
+      "learning_rate": 1e-06,
+      "loss": 0.0719,
+      "step": 200
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5000,
+  "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": 2.798657154800681e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-200/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b36bec9488b445350964041059369935c653f243588cf528b415499fc00dcca2
+size 6968