Upload README.md with huggingface_hub

README.md

@@ -1,63 +1,59 @@
 ---
-base_model: unsloth/meta-llama-3.1-8b-instruct-bnb-4bit
-library_name: peft
-model_name: ces_phase3b_lora
+license: mit
+base_model: meta-llama/Meta-Llama-3.1-8B-Instruct
 tags:
-- base_model:adapter:unsloth/meta-llama-3.1-8b-instruct-bnb-4bit
+- llama
 - lora
-- sft
-- transformers
-- trl
+- political-science
+- survey-replication
+- canadian-election-study
+- peft
 - unsloth
-licence: license
+datasets:
+- custom
+language:
+- en
 pipeline_tag: text-generation
 ---
 
-# Model Card for ces_phase3b_lora
+# CES Phase 3B LoRA: With Party ID
 
-This model is a fine-tuned version of [unsloth/meta-llama-3.1-8b-instruct-bnb-4bit](https://huggingface.co/unsloth/meta-llama-3.1-8b-instruct-bnb-4bit).
-It has been trained using [TRL](https://github.com/huggingface/trl).
+LoRA adapter that includes party identification as an input variable. **For most use cases, prefer [Phase 3A](https://huggingface.co/baglecake/ces-phase3a-lora) instead.**
 
-## Quick start
+## Performance
 
-```python
-from transformers import pipeline
-
-question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
-generator = pipeline("text-generation", model="None", device="cuda")
-output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
-print(output["generated_text"])
-```
-
-## Training procedure
-
- 
+| Model | Variables | r |
+|-------|-----------|---|
+| Phase 3A | Demographics + Leader Ratings + Wedge Issues | 0.560 |
+| **Phase 3B (this model)** | Same + Party ID | **0.574** |
 
+**Partisan Delta = 0.014** (essentially zero)
 
-This model was trained with SFT.
+## Why Phase 3A is Preferred
 
-### Framework versions
+Adding party ID only improves correlation by 1.4%. This proves party identity is **redundant** — it's already encoded in leader affect and policy positions.
 
-- PEFT 0.18.0
-- TRL: 0.24.0
-- Transformers: 4.57.2
-- Pytorch: 2.9.1
-- Datasets: 4.3.0
-- Tokenizers: 0.22.1
+Phase 3B exists for reproducibility and to demonstrate this null result.
 
-## Citations
+## Training Details
 
+- **Base model**: meta-llama/Meta-Llama-3.1-8B-Instruct (4-bit quantized via Unsloth)
+- **Training data**: ~14,455 examples from CES 2021
+- **LoRA rank**: 32
+- **LoRA alpha**: 64
+- **Epochs**: 3
 
+## Citation
 
-Cite TRL as:
-    
 ```bibtex
-@misc{vonwerra2022trl,
-	title        = {{TRL: Transformer Reinforcement Learning}},
-	author       = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
-	year         = 2020,
-	journal      = {GitHub repository},
-	publisher    = {GitHub},
-	howpublished = {\url{https://github.com/huggingface/trl}}
+@software{ces-phase3-lora,
+  title = {CES Phase 3 LoRA: Leader Affect and Policy Prediction},
+  author = {Coburn, Del},
+  year = {2025},
+  url = {https://huggingface.co/baglecake/ces-phase3a-lora}
 }
-```
+```
+
+## Part of emile-GCE
+
+This model is part of the [emile-GCE](https://github.com/delcoburn/emile-gce) project for Generative Computational Ethnography.