Upload folder using huggingface_hub

README.md

@@ -0,0 +1,202 @@
+---
+library_name: peft
+base_model: Qwen/Qwen-VL-Chat
+---
+
+# 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 0.10.0

adapter_config.json

@@ -0,0 +1,380 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen-VL-Chat",
+  "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": 16,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "transformer.h.25.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.25.attn.out_proj",
+    "transformer.h.12.mlp.w1",
+    "transformer.visual.transformer.resblocks.45.mlp.c_fc",
+    "transformer.visual.transformer.resblocks.46.mlp.c_fc",
+    "transformer.h.27.attn.c_proj",
+    "transformer.h.4.attn.c_attn",
+    "transformer.h.21.attn.c_proj",
+    "transformer.visual.transformer.resblocks.8.attn.out_proj",
+    "transformer.visual.transformer.resblocks.27.attn.out_proj",
+    "transformer.visual.transformer.resblocks.22.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.36.mlp.c_fc",
+    "transformer.h.3.mlp.w1",
+    "transformer.visual.transformer.resblocks.8.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.4.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.16.mlp.c_fc",
+    "transformer.h.20.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.18.attn.in_proj",
+    "transformer.visual.transformer.resblocks.10.attn.in_proj",
+    "transformer.visual.transformer.resblocks.13.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.32.attn.in_proj",
+    "transformer.h.4.attn.c_proj",
+    "transformer.visual.transformer.resblocks.19.attn.in_proj",
+    "transformer.h.5.attn.c_attn",
+    "transformer.visual.transformer.resblocks.14.attn.in_proj",
+    "transformer.visual.transformer.resblocks.32.mlp.c_fc",
+    "transformer.h.9.mlp.w2",
+    "transformer.h.7.attn.c_attn",
+    "transformer.visual.transformer.resblocks.34.attn.in_proj",
+    "transformer.h.22.mlp.c_proj",
+    "transformer.visual.conv1",
+    "transformer.visual.transformer.resblocks.30.attn.out_proj",
+    "transformer.visual.transformer.resblocks.34.mlp.c_fc",
+    "transformer.h.31.mlp.w2",
+    "transformer.h.28.mlp.w1",
+    "transformer.h.12.mlp.w2",
+    "transformer.h.8.attn.c_attn",
+    "transformer.h.6.mlp.w2",
+    "transformer.visual.transformer.resblocks.9.mlp.c_fc",
+    "transformer.h.14.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.26.mlp.c_proj",
+    "transformer.h.24.attn.c_attn",
+    "transformer.h.15.mlp.c_proj",
+    "transformer.h.19.attn.c_proj",
+    "transformer.visual.transformer.resblocks.35.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.35.attn.out_proj",
+    "transformer.h.6.attn.c_proj",
+    "transformer.h.31.mlp.w1",
+    "transformer.visual.transformer.resblocks.7.attn.out_proj",
+    "transformer.h.16.attn.c_attn",
+    "transformer.h.25.mlp.w2",
+    "transformer.visual.transformer.resblocks.16.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.23.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.12.mlp.c_proj",
+    "transformer.h.16.mlp.w2",
+    "transformer.visual.transformer.resblocks.19.attn.out_proj",
+    "transformer.visual.transformer.resblocks.32.attn.out_proj",
+    "transformer.visual.transformer.resblocks.19.mlp.c_fc",
+    "transformer.h.15.attn.c_proj",
+    "transformer.h.30.attn.c_attn",
+    "transformer.h.22.mlp.w2",
+    "transformer.visual.transformer.resblocks.38.mlp.c_fc",
+    "transformer.h.2.mlp.c_proj",
+    "transformer.h.13.mlp.w1",
+    "transformer.h.5.attn.c_proj",
+    "transformer.visual.transformer.resblocks.40.attn.out_proj",
+    "transformer.visual.transformer.resblocks.17.attn.out_proj",
+    "transformer.visual.transformer.resblocks.28.mlp.c_proj",
+    "transformer.h.4.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.9.attn.in_proj",
+    "transformer.visual.transformer.resblocks.1.attn.in_proj",
+    "transformer.h.17.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.33.attn.in_proj",
+    "transformer.visual.transformer.resblocks.10.attn.out_proj",
+    "transformer.visual.transformer.resblocks.42.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.47.attn.in_proj",
+    "transformer.visual.transformer.resblocks.20.mlp.c_fc",
+    "transformer.h.19.mlp.w2",
+    "transformer.visual.transformer.resblocks.38.attn.in_proj",
+    "transformer.visual.transformer.resblocks.25.attn.in_proj",
+    "transformer.visual.transformer.resblocks.20.attn.in_proj",
+    "transformer.visual.transformer.resblocks.39.attn.in_proj",
+    "transformer.visual.transformer.resblocks.5.mlp.c_fc",
+    "transformer.visual.transformer.resblocks.12.mlp.c_fc",
+    "transformer.h.27.attn.c_attn",
+    "transformer.visual.transformer.resblocks.34.attn.out_proj",
+    "transformer.visual.transformer.resblocks.2.attn.out_proj",
+    "transformer.h.16.mlp.c_proj",
+    "transformer.h.30.mlp.w1",
+    "transformer.visual.transformer.resblocks.0.attn.out_proj",
+    "transformer.visual.transformer.resblocks.22.attn.in_proj",
+    "transformer.h.23.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.2.attn.in_proj",
+    "transformer.h.4.mlp.w2",
+    "transformer.h.7.attn.c_proj",
+    "transformer.h.6.attn.c_attn",
+    "transformer.visual.transformer.resblocks.7.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.0.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.40.mlp.c_fc",
+    "transformer.h.28.mlp.w2",
+    "transformer.h.1.mlp.c_proj",
+    "transformer.h.16.mlp.w1",
+    "transformer.h.0.attn.c_attn",
+    "transformer.h.4.mlp.w1",
+    "transformer.h.11.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.6.mlp.c_proj",
+    "transformer.h.15.mlp.w1",
+    "transformer.h.28.attn.c_proj",
+    "transformer.visual.transformer.resblocks.43.attn.in_proj",
+    "transformer.h.18.mlp.c_proj",
+    "transformer.h.17.attn.c_proj",
+    "transformer.h.8.attn.c_proj",
+    "transformer.h.31.mlp.c_proj",
+    "transformer.h.9.mlp.w1",
+    "transformer.visual.transformer.resblocks.28.attn.out_proj",
+    "transformer.visual.transformer.resblocks.42.mlp.c_fc",
+    "transformer.visual.transformer.resblocks.1.mlp.c_proj",
+    "transformer.h.7.mlp.w2",
+    "transformer.visual.transformer.resblocks.29.attn.out_proj",
+    "transformer.h.23.mlp.w2",
+    "transformer.visual.transformer.resblocks.0.mlp.c_fc",
+    "transformer.h.11.attn.c_attn",
+    "transformer.h.30.mlp.w2",
+    "transformer.visual.transformer.resblocks.29.attn.in_proj",
+    "transformer.visual.transformer.resblocks.15.attn.in_proj",
+    "transformer.h.26.attn.c_attn",
+    "transformer.h.12.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.25.mlp.c_fc",
+    "transformer.h.5.mlp.w2",
+    "transformer.visual.transformer.resblocks.12.attn.in_proj",
+    "transformer.visual.transformer.resblocks.31.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.22.attn.out_proj",
+    "transformer.h.17.mlp.w1",
+    "transformer.visual.transformer.resblocks.45.attn.out_proj",
+    "transformer.visual.transformer.resblocks.17.mlp.c_fc",
+    "transformer.visual.transformer.resblocks.47.mlp.c_proj",
+    "transformer.h.2.attn.c_attn",
+    "transformer.visual.transformer.resblocks.23.mlp.c_fc",
+    "transformer.h.22.mlp.w1",
+    "transformer.h.10.mlp.w2",
+    "transformer.visual.transformer.resblocks.23.attn.in_proj",
+    "transformer.visual.transformer.resblocks.9.mlp.c_proj",
+    "transformer.h.30.attn.c_proj",
+    "transformer.h.24.mlp.c_proj",
+    "transformer.h.10.attn.c_attn",
+    "transformer.visual.transformer.resblocks.2.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.5.attn.out_proj",
+    "transformer.h.12.attn.c_attn",
+    "transformer.visual.transformer.resblocks.42.attn.out_proj",
+    "transformer.visual.transformer.resblocks.6.mlp.c_fc",
+    "transformer.visual.transformer.resblocks.26.attn.out_proj",
+    "transformer.h.17.attn.c_attn",
+    "transformer.visual.transformer.resblocks.5.attn.in_proj",
+    "transformer.visual.transformer.resblocks.37.attn.out_proj",
+    "transformer.visual.transformer.resblocks.47.attn.out_proj",
+    "transformer.h.21.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.41.attn.out_proj",
+    "transformer.h.9.attn.c_attn",
+    "transformer.visual.transformer.resblocks.41.mlp.c_fc",
+    "transformer.visual.transformer.resblocks.13.attn.in_proj",
+    "transformer.h.18.attn.c_proj",
+    "transformer.h.18.mlp.w2",
+    "transformer.h.20.mlp.w1",
+    "transformer.visual.transformer.resblocks.31.mlp.c_fc",
+    "transformer.visual.transformer.resblocks.38.attn.out_proj",
+    "transformer.visual.transformer.resblocks.13.attn.out_proj",
+    "transformer.h.8.mlp.c_proj",
+    "transformer.h.29.mlp.w2",
+    "transformer.h.25.attn.c_proj",
+    "transformer.visual.transformer.resblocks.15.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.37.mlp.c_proj",
+    "transformer.h.21.mlp.w1",
+    "transformer.visual.transformer.resblocks.39.mlp.c_fc",
+    "transformer.visual.transformer.resblocks.46.attn.in_proj",
+    "transformer.visual.transformer.resblocks.31.attn.out_proj",
+    "transformer.h.6.mlp.w1",
+    "transformer.visual.transformer.resblocks.17.mlp.c_proj",
+    "transformer.h.11.attn.c_proj",
+    "transformer.h.1.attn.c_proj",
+    "transformer.h.24.attn.c_proj",
+    "transformer.h.15.attn.c_attn",
+    "transformer.h.5.mlp.w1",
+    "transformer.h.27.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.4.attn.out_proj",
+    "transformer.visual.transformer.resblocks.7.mlp.c_fc",
+    "transformer.visual.transformer.resblocks.24.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.15.attn.out_proj",
+    "transformer.visual.transformer.resblocks.21.attn.in_proj",
+    "transformer.h.27.mlp.w2",
+    "transformer.visual.transformer.resblocks.43.mlp.c_fc",
+    "transformer.visual.transformer.resblocks.39.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.11.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.43.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.2.mlp.c_fc",
+    "transformer.visual.transformer.resblocks.13.mlp.c_fc",
+    "transformer.visual.transformer.resblocks.19.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.30.mlp.c_fc",
+    "transformer.h.14.attn.c_proj",
+    "transformer.visual.transformer.resblocks.21.mlp.c_fc",
+    "transformer.visual.transformer.resblocks.27.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.40.attn.in_proj",
+    "transformer.visual.transformer.resblocks.6.attn.out_proj",
+    "transformer.h.20.attn.c_proj",
+    "transformer.visual.transformer.resblocks.1.mlp.c_fc",
+    "transformer.h.29.mlp.w1",
+    "transformer.h.0.mlp.w1",
+    "transformer.visual.transformer.resblocks.3.mlp.c_fc",
+    "transformer.visual.transformer.resblocks.16.attn.in_proj",
+    "transformer.visual.transformer.resblocks.25.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.3.mlp.c_proj",
+    "transformer.h.2.attn.c_proj",
+    "transformer.visual.transformer.resblocks.44.attn.out_proj",
+    "transformer.h.1.mlp.w1",
+    "transformer.visual.transformer.resblocks.11.attn.in_proj",
+    "transformer.visual.transformer.resblocks.21.mlp.c_proj",
+    "transformer.h.13.attn.c_proj",
+    "transformer.h.19.mlp.w1",
+    "transformer.h.3.attn.c_proj",
+    "transformer.visual.transformer.resblocks.18.mlp.c_fc",
+    "transformer.h.3.mlp.w2",
+    "transformer.visual.transformer.resblocks.36.attn.in_proj",
+    "transformer.visual.transformer.resblocks.10.mlp.c_fc",
+    "transformer.h.14.attn.c_attn",
+    "transformer.visual.transformer.resblocks.35.attn.in_proj",
+    "transformer.h.13.attn.c_attn",
+    "transformer.visual.transformer.resblocks.34.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.15.mlp.c_fc",
+    "transformer.visual.transformer.resblocks.24.mlp.c_fc",
+    "transformer.h.7.mlp.w1",
+    "transformer.h.8.mlp.w2",
+    "transformer.h.26.mlp.w2",
+    "transformer.h.0.attn.c_proj",
+    "transformer.visual.transformer.resblocks.6.attn.in_proj",
+    "transformer.h.20.mlp.w2",
+    "transformer.visual.transformer.resblocks.11.mlp.c_fc",
+    "transformer.visual.transformer.resblocks.29.mlp.c_proj",
+    "transformer.h.0.mlp.w2",
+    "transformer.h.22.attn.c_attn",
+    "transformer.h.23.mlp.w1",
+    "transformer.visual.transformer.resblocks.35.mlp.c_fc",
+    "transformer.h.0.mlp.c_proj",
+    "transformer.h.29.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.12.attn.out_proj",
+    "transformer.visual.transformer.resblocks.24.attn.in_proj",
+    "transformer.visual.transformer.resblocks.4.mlp.c_fc",
+    "transformer.visual.transformer.resblocks.45.attn.in_proj",
+    "transformer.h.15.mlp.w2",
+    "transformer.visual.transformer.resblocks.27.attn.in_proj",
+    "transformer.visual.transformer.resblocks.17.attn.in_proj",
+    "transformer.h.18.attn.c_attn",
+    "transformer.h.11.mlp.w1",
+    "transformer.visual.transformer.resblocks.1.attn.out_proj",
+    "transformer.visual.transformer.resblocks.33.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.41.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.24.attn.out_proj",
+    "transformer.visual.transformer.resblocks.18.mlp.c_proj",
+    "transformer.h.17.mlp.w2",
+    "transformer.h.22.attn.c_proj",
+    "transformer.visual.transformer.resblocks.45.mlp.c_proj",
+    "transformer.h.31.attn.c_attn",
+    "transformer.h.24.mlp.w2",
+    "transformer.h.2.mlp.w1",
+    "transformer.h.12.attn.c_proj",
+    "transformer.visual.transformer.resblocks.26.attn.in_proj",
+    "transformer.visual.transformer.resblocks.9.attn.out_proj",
+    "transformer.h.21.attn.c_attn",
+    "transformer.h.29.attn.c_attn",
+    "transformer.h.3.attn.c_attn",
+    "transformer.h.30.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.5.mlp.c_proj",
+    "transformer.h.24.mlp.w1",
+    "transformer.visual.transformer.resblocks.10.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.41.attn.in_proj",
+    "transformer.visual.transformer.resblocks.44.mlp.c_fc",
+    "transformer.h.6.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.46.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.30.mlp.c_proj",
+    "transformer.h.28.attn.c_attn",
+    "transformer.h.13.mlp.c_proj",
+    "transformer.h.7.mlp.c_proj",
+    "transformer.h.14.mlp.w2",
+    "transformer.visual.transformer.resblocks.7.attn.in_proj",
+    "transformer.visual.transformer.resblocks.44.mlp.c_proj",
+    "transformer.h.9.attn.c_proj",
+    "transformer.visual.transformer.resblocks.44.attn.in_proj",
+    "transformer.visual.transformer.resblocks.28.mlp.c_fc",
+    "transformer.visual.transformer.resblocks.0.attn.in_proj",
+    "transformer.h.21.mlp.w2",
+    "transformer.h.8.mlp.w1",
+    "transformer.h.10.attn.c_proj",
+    "transformer.h.25.attn.c_attn",
+    "transformer.h.1.mlp.w2",
+    "transformer.visual.transformer.resblocks.46.attn.out_proj",
+    "transformer.visual.transformer.resblocks.18.attn.out_proj",
+    "transformer.visual.transformer.resblocks.28.attn.in_proj",
+    "transformer.h.31.attn.c_proj",
+    "transformer.h.26.attn.c_proj",
+    "transformer.visual.transformer.resblocks.16.attn.out_proj",
+    "transformer.visual.transformer.resblocks.3.attn.out_proj",
+    "transformer.h.16.attn.c_proj",
+    "transformer.visual.transformer.resblocks.42.attn.in_proj",
+    "transformer.visual.transformer.resblocks.38.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.43.attn.out_proj",
+    "transformer.visual.transformer.resblocks.14.mlp.c_fc",
+    "transformer.visual.transformer.resblocks.32.mlp.c_proj",
+    "transformer.h.11.mlp.w2",
+    "transformer.h.10.mlp.w1",
+    "transformer.visual.transformer.resblocks.29.mlp.c_fc",
+    "transformer.h.2.mlp.w2",
+    "transformer.visual.transformer.resblocks.36.mlp.c_proj",
+    "transformer.h.10.mlp.c_proj",
+    "transformer.h.14.mlp.w1",
+    "transformer.visual.transformer.resblocks.14.attn.out_proj",
+    "transformer.visual.transformer.resblocks.4.attn.in_proj",
+    "transformer.visual.transformer.resblocks.23.attn.out_proj",
+    "transformer.visual.transformer.resblocks.11.attn.out_proj",
+    "transformer.h.3.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.14.mlp.c_proj",
+    "transformer.h.25.mlp.w1",
+    "transformer.visual.transformer.resblocks.22.mlp.c_fc",
+    "transformer.visual.transformer.resblocks.27.mlp.c_fc",
+    "transformer.visual.transformer.resblocks.40.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.20.attn.out_proj",
+    "transformer.visual.transformer.resblocks.21.attn.out_proj",
+    "transformer.visual.transformer.resblocks.8.attn.in_proj",
+    "transformer.h.5.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.37.mlp.c_fc",
+    "transformer.h.19.attn.c_attn",
+    "transformer.visual.transformer.resblocks.39.attn.out_proj",
+    "transformer.visual.transformer.resblocks.36.attn.out_proj",
+    "transformer.visual.transformer.resblocks.8.mlp.c_fc",
+    "transformer.h.13.mlp.w2",
+    "transformer.h.28.mlp.c_proj",
+    "transformer.h.29.attn.c_proj",
+    "transformer.h.18.mlp.w1",
+    "transformer.visual.transformer.resblocks.47.mlp.c_fc",
+    "transformer.h.20.attn.c_attn",
+    "transformer.visual.transformer.resblocks.31.attn.in_proj",
+    "transformer.visual.transformer.resblocks.30.attn.in_proj",
+    "transformer.h.27.mlp.w1",
+    "transformer.visual.transformer.resblocks.37.attn.in_proj",
+    "transformer.visual.transformer.resblocks.26.mlp.c_fc",
+    "transformer.visual.transformer.resblocks.33.mlp.c_fc",
+    "transformer.h.23.attn.c_proj",
+    "transformer.h.19.mlp.c_proj",
+    "transformer.h.23.attn.c_attn",
+    "transformer.visual.transformer.resblocks.20.mlp.c_proj",
+    "transformer.h.1.attn.c_attn",
+    "transformer.visual.transformer.resblocks.3.attn.in_proj",
+    "transformer.h.9.mlp.c_proj",
+    "transformer.h.26.mlp.c_proj",
+    "transformer.visual.transformer.resblocks.33.attn.out_proj",
+    "transformer.h.26.mlp.w1"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}

adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a4d62d92c4b35ef29608ae3d385e2dbbd3bd2e44e1e9c274952fed6b383e48e
+size 469105640

latest

@@ -0,0 +1 @@
+global_step800

rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a343dd93cd21bdc90d289f3ca48ab49de24b9f748799acb23184c62f5d2b505a
+size 15920

rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e78f906506770f43e59c54fef023c80264ba4db0c95909db5aa497d4875f1e32
+size 15920

rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1801fc92aac20f4b2cd6c241493cc948c1ce8800b14797fdefee2b1f494d7b9f
+size 15920

rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb7617cd6b75e491a348879fed069c07f2a2f52647a39a51812a3039227e011e
+size 15920

rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f2aa9e524787be3fd2130cbb1a33ce0d917090fdf18cf026905505c6c1f67c64
+size 15920

rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:048ab222ccc631300416028b25a3132d82f849b7a32356b338d26e9eef8ae3fb
+size 15920

rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a9670c00e2e4b001bb5f458d57d181a0ae7bf4587cc05947eef4b84e438e4178
+size 15920

rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:da5ed04d69158bc88c3dc621620dae175703ddfab9924471e44fc939b4c4386c
+size 15920

scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:901dc2645bb26444439097220bce3343e3d0a315e276f271fbd122fb8170ad53
+size 1064

special_tokens_map.json

@@ -0,0 +1,3 @@
+{
+  "pad_token": "<|endoftext|>"
+}

tokenizer_config.json

@@ -0,0 +1,14 @@
+{
+  "added_tokens_decoder": {},
+  "auto_map": {
+    "AutoTokenizer": [
+      "Qwen/Qwen-VL-Chat--tokenization_qwen.QWenTokenizer",
+      null
+    ]
+  },
+  "clean_up_tokenization_spaces": true,
+  "model_max_length": 1280,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "right",
+  "tokenizer_class": "QWenTokenizer"
+}

trainer_state.json

@@ -0,0 +1,593 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.12325706802249442,
+  "eval_steps": 500,
+  "global_step": 800,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0015407133502811801,
+      "grad_norm": 48.20575104273089,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 1.3971,
+      "step": 10
+    },
+    {
+      "epoch": 0.0030814267005623602,
+      "grad_norm": 33.11975747134839,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 1.4053,
+      "step": 20
+    },
+    {
+      "epoch": 0.004622140050843541,
+      "grad_norm": 39.826606299507084,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 1.3472,
+      "step": 30
+    },
+    {
+      "epoch": 0.0061628534011247205,
+      "grad_norm": 6.773742739301677,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 1.1311,
+      "step": 40
+    },
+    {
+      "epoch": 0.007703566751405901,
+      "grad_norm": 12.278712041952108,
+      "learning_rate": 2.307692307692308e-05,
+      "loss": 0.9715,
+      "step": 50
+    },
+    {
+      "epoch": 0.009244280101687081,
+      "grad_norm": 3.776870082439811,
+      "learning_rate": 2.7692307692307694e-05,
+      "loss": 0.8937,
+      "step": 60
+    },
+    {
+      "epoch": 0.010784993451968261,
+      "grad_norm": 6.812660145408396,
+      "learning_rate": 2.9999955171465948e-05,
+      "loss": 0.8472,
+      "step": 70
+    },
+    {
+      "epoch": 0.012325706802249441,
+      "grad_norm": 14.1920325479403,
+      "learning_rate": 2.9999596544801216e-05,
+      "loss": 0.8418,
+      "step": 80
+    },
+    {
+      "epoch": 0.013866420152530621,
+      "grad_norm": 4.760185564415112,
+      "learning_rate": 2.999887930004599e-05,
+      "loss": 0.8275,
+      "step": 90
+    },
+    {
+      "epoch": 0.015407133502811803,
+      "grad_norm": 5.64069685533569,
+      "learning_rate": 2.9997803454348518e-05,
+      "loss": 0.8085,
+      "step": 100
+    },
+    {
+      "epoch": 0.01694784685309298,
+      "grad_norm": 8.910755610595423,
+      "learning_rate": 2.9996369033430674e-05,
+      "loss": 0.8105,
+      "step": 110
+    },
+    {
+      "epoch": 0.018488560203374162,
+      "grad_norm": 7.128375140746789,
+      "learning_rate": 2.9994576071587345e-05,
+      "loss": 0.7647,
+      "step": 120
+    },
+    {
+      "epoch": 0.020029273553655344,
+      "grad_norm": 4.883749519467239,
+      "learning_rate": 2.9992424611685575e-05,
+      "loss": 0.7472,
+      "step": 130
+    },
+    {
+      "epoch": 0.021569986903936522,
+      "grad_norm": 6.669698248447361,
+      "learning_rate": 2.9989914705163582e-05,
+      "loss": 0.7644,
+      "step": 140
+    },
+    {
+      "epoch": 0.023110700254217704,
+      "grad_norm": 8.814324372744018,
+      "learning_rate": 2.9987046412029506e-05,
+      "loss": 0.7642,
+      "step": 150
+    },
+    {
+      "epoch": 0.024651413604498882,
+      "grad_norm": 5.408352128962619,
+      "learning_rate": 2.9983819800859976e-05,
+      "loss": 0.7506,
+      "step": 160
+    },
+    {
+      "epoch": 0.026192126954780064,
+      "grad_norm": 4.916006313310609,
+      "learning_rate": 2.998023494879848e-05,
+      "loss": 0.7086,
+      "step": 170
+    },
+    {
+      "epoch": 0.027732840305061242,
+      "grad_norm": 6.4968532868316515,
+      "learning_rate": 2.9976291941553508e-05,
+      "loss": 0.7302,
+      "step": 180
+    },
+    {
+      "epoch": 0.029273553655342423,
+      "grad_norm": 5.414282856209666,
+      "learning_rate": 2.9971990873396512e-05,
+      "loss": 0.7389,
+      "step": 190
+    },
+    {
+      "epoch": 0.030814267005623605,
+      "grad_norm": 35.78019922037886,
+      "learning_rate": 2.996733184715964e-05,
+      "loss": 0.7247,
+      "step": 200
+    },
+    {
+      "epoch": 0.03235498035590478,
+      "grad_norm": 4.344484335761467,
+      "learning_rate": 2.9962314974233306e-05,
+      "loss": 0.7239,
+      "step": 210
+    },
+    {
+      "epoch": 0.03389569370618596,
+      "grad_norm": 4.761393885682967,
+      "learning_rate": 2.995694037456349e-05,
+      "loss": 0.7219,
+      "step": 220
+    },
+    {
+      "epoch": 0.03543640705646715,
+      "grad_norm": 6.082748106236669,
+      "learning_rate": 2.995120817664889e-05,
+      "loss": 0.7036,
+      "step": 230
+    },
+    {
+      "epoch": 0.036977120406748325,
+      "grad_norm": 7.925002223835823,
+      "learning_rate": 2.9945118517537857e-05,
+      "loss": 0.6795,
+      "step": 240
+    },
+    {
+      "epoch": 0.0385178337570295,
+      "grad_norm": 4.33748938479485,
+      "learning_rate": 2.9938671542825102e-05,
+      "loss": 0.6894,
+      "step": 250
+    },
+    {
+      "epoch": 0.04005854710731069,
+      "grad_norm": 3.419861222736367,
+      "learning_rate": 2.993186740664821e-05,
+      "loss": 0.674,
+      "step": 260
+    },
+    {
+      "epoch": 0.041599260457591866,
+      "grad_norm": 4.418012569894865,
+      "learning_rate": 2.9924706271683993e-05,
+      "loss": 0.7091,
+      "step": 270
+    },
+    {
+      "epoch": 0.043139973807873044,
+      "grad_norm": 5.5535932232942065,
+      "learning_rate": 2.9917188309144548e-05,
+      "loss": 0.7114,
+      "step": 280
+    },
+    {
+      "epoch": 0.04468068715815422,
+      "grad_norm": 5.346051172390458,
+      "learning_rate": 2.990931369877321e-05,
+      "loss": 0.7092,
+      "step": 290
+    },
+    {
+      "epoch": 0.04622140050843541,
+      "grad_norm": 5.04585392782648,
+      "learning_rate": 2.9901082628840216e-05,
+      "loss": 0.7079,
+      "step": 300
+    },
+    {
+      "epoch": 0.047762113858716586,
+      "grad_norm": 5.3108067226217095,
+      "learning_rate": 2.989249529613823e-05,
+      "loss": 0.7044,
+      "step": 310
+    },
+    {
+      "epoch": 0.049302827208997764,
+      "grad_norm": 12.936374781230974,
+      "learning_rate": 2.9883551905977647e-05,
+      "loss": 0.6795,
+      "step": 320
+    },
+    {
+      "epoch": 0.05084354055927895,
+      "grad_norm": 4.969368758151501,
+      "learning_rate": 2.987425267218164e-05,
+      "loss": 0.7365,
+      "step": 330
+    },
+    {
+      "epoch": 0.05238425390956013,
+      "grad_norm": 17.59963354249395,
+      "learning_rate": 2.9864597817081083e-05,
+      "loss": 0.6459,
+      "step": 340
+    },
+    {
+      "epoch": 0.053924967259841305,
+      "grad_norm": 4.287709062556245,
+      "learning_rate": 2.985458757150924e-05,
+      "loss": 0.7151,
+      "step": 350
+    },
+    {
+      "epoch": 0.055465680610122484,
+      "grad_norm": 2.353889201419343,
+      "learning_rate": 2.9844222174796224e-05,
+      "loss": 0.6982,
+      "step": 360
+    },
+    {
+      "epoch": 0.05700639396040367,
+      "grad_norm": 6.487203178892158,
+      "learning_rate": 2.983350187476328e-05,
+      "loss": 0.6946,
+      "step": 370
+    },
+    {
+      "epoch": 0.05854710731068485,
+      "grad_norm": 8.604226568596971,
+      "learning_rate": 2.982242692771688e-05,
+      "loss": 0.7024,
+      "step": 380
+    },
+    {
+      "epoch": 0.060087820660966025,
+      "grad_norm": 3.7381224404208107,
+      "learning_rate": 2.9810997598442558e-05,
+      "loss": 0.6813,
+      "step": 390
+    },
+    {
+      "epoch": 0.06162853401124721,
+      "grad_norm": 7.21302363512955,
+      "learning_rate": 2.9799214160198622e-05,
+      "loss": 0.6572,
+      "step": 400
+    },
+    {
+      "epoch": 0.06316924736152839,
+      "grad_norm": 7.446657585655329,
+      "learning_rate": 2.9787076894709592e-05,
+      "loss": 0.6612,
+      "step": 410
+    },
+    {
+      "epoch": 0.06470996071180957,
+      "grad_norm": 16.149479699980372,
+      "learning_rate": 2.977458609215946e-05,
+      "loss": 0.6823,
+      "step": 420
+    },
+    {
+      "epoch": 0.06625067406209074,
+      "grad_norm": 3.654646693112352,
+      "learning_rate": 2.9761742051184786e-05,
+      "loss": 0.6941,
+      "step": 430
+    },
+    {
+      "epoch": 0.06779138741237192,
+      "grad_norm": 7.468345726914931,
+      "learning_rate": 2.9748545078867524e-05,
+      "loss": 0.6254,
+      "step": 440
+    },
+    {
+      "epoch": 0.06933210076265311,
+      "grad_norm": 6.559292545318311,
+      "learning_rate": 2.9734995490727696e-05,
+      "loss": 0.6977,
+      "step": 450
+    },
+    {
+      "epoch": 0.0708728141129343,
+      "grad_norm": 8.749994768413615,
+      "learning_rate": 2.9721093610715844e-05,
+      "loss": 0.6742,
+      "step": 460
+    },
+    {
+      "epoch": 0.07241352746321547,
+      "grad_norm": 14.746822726186753,
+      "learning_rate": 2.9706839771205282e-05,
+      "loss": 0.6986,
+      "step": 470
+    },
+    {
+      "epoch": 0.07395424081349665,
+      "grad_norm": 2.1125271451210494,
+      "learning_rate": 2.9692234312984156e-05,
+      "loss": 0.6708,
+      "step": 480
+    },
+    {
+      "epoch": 0.07549495416377783,
+      "grad_norm": 4.191787346884265,
+      "learning_rate": 2.9677277585247296e-05,
+      "loss": 0.6839,
+      "step": 490
+    },
+    {
+      "epoch": 0.077035667514059,
+      "grad_norm": 6.099946769106883,
+      "learning_rate": 2.9661969945587867e-05,
+      "loss": 0.7253,
+      "step": 500
+    },
+    {
+      "epoch": 0.07857638086434018,
+      "grad_norm": 8.272898482647978,
+      "learning_rate": 2.9646311759988804e-05,
+      "loss": 0.6972,
+      "step": 510
+    },
+    {
+      "epoch": 0.08011709421462138,
+      "grad_norm": 12.38761120543235,
+      "learning_rate": 2.9630303402814095e-05,
+      "loss": 0.7174,
+      "step": 520
+    },
+    {
+      "epoch": 0.08165780756490255,
+      "grad_norm": 7.826170900143438,
+      "learning_rate": 2.961394525679979e-05,
+      "loss": 0.7227,
+      "step": 530
+    },
+    {
+      "epoch": 0.08319852091518373,
+      "grad_norm": 3.926770198625081,
+      "learning_rate": 2.9597237713044888e-05,
+      "loss": 0.6682,
+      "step": 540
+    },
+    {
+      "epoch": 0.08473923426546491,
+      "grad_norm": 3.3408474311965692,
+      "learning_rate": 2.9580181171001962e-05,
+      "loss": 0.6634,
+      "step": 550
+    },
+    {
+      "epoch": 0.08627994761574609,
+      "grad_norm": 15.954087186336258,
+      "learning_rate": 2.956277603846761e-05,
+      "loss": 0.7005,
+      "step": 560
+    },
+    {
+      "epoch": 0.08782066096602727,
+      "grad_norm": 3.3366594091813475,
+      "learning_rate": 2.9545022731572723e-05,
+      "loss": 0.6752,
+      "step": 570
+    },
+    {
+      "epoch": 0.08936137431630845,
+      "grad_norm": 4.359220549879328,
+      "learning_rate": 2.9526921674772522e-05,
+      "loss": 0.6985,
+      "step": 580
+    },
+    {
+      "epoch": 0.09090208766658964,
+      "grad_norm": 5.585924353950877,
+      "learning_rate": 2.95084733008364e-05,
+      "loss": 0.6729,
+      "step": 590
+    },
+    {
+      "epoch": 0.09244280101687082,
+      "grad_norm": 2.5002732127626075,
+      "learning_rate": 2.94896780508376e-05,
+      "loss": 0.6881,
+      "step": 600
+    },
+    {
+      "epoch": 0.093983514367152,
+      "grad_norm": 8.22666868773466,
+      "learning_rate": 2.9470536374142656e-05,
+      "loss": 0.6918,
+      "step": 610
+    },
+    {
+      "epoch": 0.09552422771743317,
+      "grad_norm": 14.050897991765508,
+      "learning_rate": 2.9451048728400644e-05,
+      "loss": 0.6584,
+      "step": 620
+    },
+    {
+      "epoch": 0.09706494106771435,
+      "grad_norm": 3.117445397285045,
+      "learning_rate": 2.9431215579532253e-05,
+      "loss": 0.6495,
+      "step": 630
+    },
+    {
+      "epoch": 0.09860565441799553,
+      "grad_norm": 7.000286096448409,
+      "learning_rate": 2.9411037401718628e-05,
+      "loss": 0.6568,
+      "step": 640
+    },
+    {
+      "epoch": 0.1001463677682767,
+      "grad_norm": 3.09698455955492,
+      "learning_rate": 2.939051467739006e-05,
+      "loss": 0.7095,
+      "step": 650
+    },
+    {
+      "epoch": 0.1016870811185579,
+      "grad_norm": 4.163429426840949,
+      "learning_rate": 2.936964789721442e-05,
+      "loss": 0.6573,
+      "step": 660
+    },
+    {
+      "epoch": 0.10322779446883908,
+      "grad_norm": 3.282543980811516,
+      "learning_rate": 2.934843756008546e-05,
+      "loss": 0.6901,
+      "step": 670
+    },
+    {
+      "epoch": 0.10476850781912025,
+      "grad_norm": 7.4918954929572115,
+      "learning_rate": 2.932688417311085e-05,
+      "loss": 0.6826,
+      "step": 680
+    },
+    {
+      "epoch": 0.10630922116940143,
+      "grad_norm": 49.53066344289919,
+      "learning_rate": 2.9304988251600084e-05,
+      "loss": 0.6668,
+      "step": 690
+    },
+    {
+      "epoch": 0.10784993451968261,
+      "grad_norm": 23.218366206451197,
+      "learning_rate": 2.9282750319052154e-05,
+      "loss": 0.6643,
+      "step": 700
+    },
+    {
+      "epoch": 0.10939064786996379,
+      "grad_norm": 2.59944116748961,
+      "learning_rate": 2.9260170907143012e-05,
+      "loss": 0.6709,
+      "step": 710
+    },
+    {
+      "epoch": 0.11093136122024497,
+      "grad_norm": 4.59006482180365,
+      "learning_rate": 2.9237250555712887e-05,
+      "loss": 0.6878,
+      "step": 720
+    },
+    {
+      "epoch": 0.11247207457052616,
+      "grad_norm": 3.1508976338573507,
+      "learning_rate": 2.9213989812753366e-05,
+      "loss": 0.6512,
+      "step": 730
+    },
+    {
+      "epoch": 0.11401278792080734,
+      "grad_norm": 4.5342197552006835,
+      "learning_rate": 2.9190389234394285e-05,
+      "loss": 0.6687,
+      "step": 740
+    },
+    {
+      "epoch": 0.11555350127108852,
+      "grad_norm": 6.514898099534209,
+      "learning_rate": 2.9166449384890446e-05,
+      "loss": 0.6868,
+      "step": 750
+    },
+    {
+      "epoch": 0.1170942146213697,
+      "grad_norm": 5.1251296980659005,
+      "learning_rate": 2.9142170836608115e-05,
+      "loss": 0.6869,
+      "step": 760
+    },
+    {
+      "epoch": 0.11863492797165087,
+      "grad_norm": 5.896478552497953,
+      "learning_rate": 2.9117554170011352e-05,
+      "loss": 0.667,
+      "step": 770
+    },
+    {
+      "epoch": 0.12017564132193205,
+      "grad_norm": 3.4672489526983483,
+      "learning_rate": 2.909259997364811e-05,
+      "loss": 0.6674,
+      "step": 780
+    },
+    {
+      "epoch": 0.12171635467221323,
+      "grad_norm": 5.413662982418376,
+      "learning_rate": 2.9067308844136193e-05,
+      "loss": 0.6891,
+      "step": 790
+    },
+    {
+      "epoch": 0.12325706802249442,
+      "grad_norm": 5.807164850831801,
+      "learning_rate": 2.9041681386148966e-05,
+      "loss": 0.6447,
+      "step": 800
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 6490,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 400,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 7.289521573986304e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ca3eb9606621dd3b3f8b506f95a7ce6bf2218992e9565b54ee61e2cecb502260
+size 6392

zero_to_fp32.py

@@ -0,0 +1,587 @@
+#!/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):
+    """
+    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)
+    elif zero_stage == 3:
+        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states)
+
+
+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 _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()
+            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):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    _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):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    _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):
+    """
+    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``
+
+    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)
+
+
+def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None):
+    """
+    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``
+    """
+
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+    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("-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)