aucson
/

llama3-code-math-avg-merge

@@ -3,197 +3,119 @@ library_name: transformers
 tags: []
 ---
-# 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. -->
-This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
-- **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]

 tags: []
 ---
+See details in https://github.com/AuCson/RegMean-LLama3-8B.
+## Fast and Numerically Stable RegMean for Merging LLama3-8B
+This repo is a fast and numerically stable re-implementation of RegMean model merging algorithm for LLama3-8B.
+We merge the following two models.
+- [Code Model] [rombodawg/rombos_Replete-Coder-Llama3-8B](https://huggingface.co/rombodawg/rombos_Replete-Coder-Llama3-8B) (Re-implementation of Replete-Coder)
+- [Math Model][ TIGER-Lab/MAmmoTH2-8B](https://huggingface.co/TIGER-Lab/MAmmoTH2-8B)
+## Results
+| Method/Benchmark | GSM8k (Math) | Mathqa (Math) | HumanEval-Instruct (Code) | MBPP (Code) |
+|  ---- | ---- | ---- | ---- | ---- |
+|   |  5-shot EM  | 0-shot Acc-norm | 0-shot Pass@1 | 3-shot Pass@1 |
+|  [Math Model](https://huggingface.co/TIGER-Lab/MAmmoTH2-8B) |  70.40* | 43.85 | 36.59 | 40.04 |
+|  [Code Model](https://huggingface.co/rombodawg/rombos_Replete-Coder-Llama3-8B) | 57.92 | 37.35 | 42.07 | 49.20 |
+|  [Average](https://huggingface.co/aucson/llama3-code-math-avg-merge) | 65.27 | 44.05 | 43.29  | 47.80 |
+|  [RegMean ($\alpha$=0.1)](https://huggingface.co/aucson/llama3-code-math-regmean-merge/tree/main) | 68.31 | 44.99 | 44.51 | 45.20 |
+\* Official result
+\* We found the zero-shot results are sensitive to chat templates and reported best achievable result for HumanInstruct for all models: we modified `lm-evaluation-harness/lm_eval/tasks/humaneval/humaneval.yaml` so that "\`\`\`" can be considered as end of responses.
+The merged models, along with the activation inner product matrices, are avaiable on the huggingface hub.
+## What's new?
+RegMean solves a least square regression problem at each linear layer of the transformer. This is now implemented with built-in PyTorch linalg.lstsq function.
+```python
+# old
+# sum_gram_inv = torch.inverse(sum_gram)
+# wt = torch.matmul(sum_gram_inv, sum_gram_m_ws)
+# new
+wt = torch.linalg.lstsq(sum_gram, sum_gram_m_ws).solution
+```
+According to PyTorch's official doumentation,
+```
+This function computes X = A.pinverse() @ B in a faster and more numerically stable way than performing the computations separately.
+```
+## Computational efficiency
+- **Computing gram matrices**: We compute inner product matrics for code and math models on 10k training examples. Each of them take 3-hour on one Quadro RTX A6000 GPU (which can probably accelerated with more efficient LLM inference framework). But we have uploaded them under the [merged model repo](https://huggingface.co/aucson/llama3-code-math-regmean-merge/tree/main) so that you do not need to re-compute.
+- **Merging Models**: ~2 minutes on the same GPU for this re-implementation. Please note loading two 8B models and (almost) equally sized inner product matrices at once can take >150GB memory.
+## Reproducing the results
+1. Create a python environment and install the modified lm-eval-harness library for evaluating merged models.
+```
+cd lm-eval-harness
+pip install -e .
+```
+The only modification is `lm_eval/tasks/humaneval/humaneval.yaml`.
+2. Preparing activation inner product matrices.
+You can download them from the [merged model repo](https://huggingface.co/aucson/llama3-code-math-regmean-merge/tree/main) and place them under `runs/merges/math-llama3/gram.pkl` and `runs/merges/code-llama3/gram.pkl`. Alternatively, you can compute them yourself with,
+```
+python compute_gram.py code
+python compute_gram.py math
+```
+3. Merging models
+```
+python merge_model.py avg
+python merge_model.py regmean
+```
+4. Evaluation with `lm-eval-harness`. Please follow the safety guidelines of humaneval and mbpp regarding execution of LLM generated code.
+```
+merge_exp=regmean_0.1
+# merge_exp=avg
+HF_ALLOW_CODE_EVAL=1 lm_eval --model vllm --model_args pretrained=runs/merges/${merge_exp},tokenizer=meta-llama/Meta-Llama-3-8B,tensor_parallel_size=1,dtype=bfloat16 --tasks mathqa,gsm8k,humaneval_instruct,mbpp --output_path runs/merges/${merge_exp}/lm_eval_results_preds --log_samples --trust_remote_code --confirm_run_unsafe_code
+```
+## Caveats
+Overall, simple averaging works well for LLMs and the benefits of merging algorithms diminishes for merging algorithms [1]
+## Citations
+For the RegMean algorithm.
+```
+@inproceedings{
+    jin2023dataless,
+    title={Dataless Knowledge Fusion by Merging Weights of Language Models},
+    author={Xisen Jin and Xiang Ren and Daniel Preotiuc-Pietro and Pengxiang Cheng},
+    booktitle={The Eleventh International Conference on Learning Representations },
+    year={2023},
+    url={https://openreview.net/forum?id=FCnohuR6AnM}
+}
+```
+Here are other useful references that greatly inspired this re-implementation.
+[1] Yadav et al. 2024, [What Matters for Model Merging at Scale?](https://arxiv.org/abs/2410.03617)
+[2] Tam et al. 2024, [Merging by Matching Models in Task Parameter Subspaces](https://openreview.net/forum?id=qNGo6ghWFB)