Update README.md

make it look nice

README.md

@@ -1,54 +1,75 @@
 ---
 license: mit
+library_name: peft
+base_model: NousResearch/Llama-2-7b-hf
+datasets:
+  - Bton/vine-reviews
+tags:
+  - peft
+  - lora
+  - text-generation
+  - personalized
+  - fine-tuned
+  - amazon-reviews
+  - jsonl
+pipeline_tag: text-generation
 ---
 
-# ⚠️ Warning: Occasionally starts speaking German for no reason
-
-This model was trained on Amazon reviews, not Berlin travel blogs. If it suddenly says *"Wundervoll, aber zu teuer!"*, just roll with it. We're not sure why it happens, but it *really* likes European flashlights.
+# ⚠️ Warning: Occasionally starts speaking German for no reason. dunno.  
+Trained on a Google Colab T4 GPU — not the prettiest, but it gets the job done.
 
 ---
 
-# Fine-Tuned LLaMA 2 (7B) with PEFT
+# 🧠 Fine-Tuned LLaMA 2 (7B) with PEFT
+
+## Model Summary
+
+This model is a parameter-efficient fine-tuned version of [NousResearch/Llama-2-7b-hf](https://huggingface.co/NousResearch/Llama-2-7b-hf), built using the `peft` library with LoRA.  
+It was trained to replicate the tone, language, and reviewing habits of my dad, a long-time Amazon Vine reviewer.
+
+Training was done on a custom dataset derived from years of Amazon reviews, scraped and structured into instruction-tuned format for use in conversational modeling.  
+Example format:
+
+```json
+{"text": "<s>[INST] Does not include rechargeable batteries [/INST] I thought that these included rechargeable batteries, but after re-reading the description... </s>"}
+The data was split into:
+
+train.jsonl
+
+valid.jsonl
+
+test.jsonl
+
+Each entry follows the <s>[INST] instruction [/INST] response </s> structure to support compatibility with LLaMA-style dialogue tuning.
+
 
-## 🧠 Model Summary
+✅ Intended Use
+Direct Use
+Regenerate product reviews in the style of a prolific Amazon Vine reviewer
 
-This model is a parameter-efficient fine-tuned version of [NousResearch/Llama-2-7b-hf](https://huggingface.co/NousResearch/Llama-2-7b-hf), utilizing the PEFT library. It is designed to replicate the tone, style, and expression of a specific individual's writing style—specifically, the author's father—based on his Amazon product reviews over the years.
+Emulate personal tone in ecommerce content, chatbots, or stylized summaries
 
-## ✅ Intended Use
+Out-of-Scope Use
+Not for high-stakes domains (legal, medical, financial)
 
-### Direct Use
-- Text generation in the voice of a real Amazon reviewer (the author's dad)
-- Use in writing prompts, product review emulation, or humorous content generation
+Not intended for impersonation, misinformation, or deceptive representations
 
-### Out-of-Scope Use
-- Not for high-stakes domains (legal, medical, financial)
-- Not intended for impersonation, misinformation, or deceptive use
+⚠️ Risks and Limitations
+May reflect strong personal opinions — especially about polyester and glove insulation
 
-## ⚠️ Risks and Limitations
+Not guaranteed to be factually accurate or hallucination-free
 
-- May reflect biases or strong opinions from the training data (a single individual)
-- Not guaranteed to be factually correct or neutral
-- Can randomly switch to German mid-sentence—cause unknown
-- Reflects a personal and informal tone; not suited for formal applications
+Prone to occasional repetition
 
-## 🏋️ Training Details
+Can randomly switch to German mid-sentence (don’t ask, we don’t know either)
 
-- **PEFT Method:** LoRA
-- **Precision:** bf16
-- **Training Data:** Scraped and cleaned Amazon reviews written by the author's father over many years, curated to replicate tone and expression
-- **Hardware:** [Insert your training hardware here, e.g., 1x A100, M3 MacBook, etc.]
+🏋️ Training Details
+PEFT Method: LoRA (Low-Rank Adaptation)
 
-## 💻 Example Usage
+Precision: bf16
 
-```python
-from peft import PeftModel, PeftConfig
-from transformers import AutoModelForCausalLM, AutoTokenizer
+Training Data: Bton/vine-reviews — scraped, cleaned, and formatted Amazon Vine reviews written by better reviewer than myself.
 
-config = PeftConfig.from_pretrained("your-model-path")
-base_model = AutoModelForCausalLM.from_pretrained(config.base_model_name_or_path)
-model = PeftModel.from_pretrained(base_model, "your-model-path")
-tokenizer = AutoTokenizer.from_pretrained(config.base_model_name_or_path)
+Data Format: JSONL with instruction-style <s>[INST] ... [/INST] ... </s> prompts
 
-inputs = tokenizer("Write a review of a flashlight", return_tensors="pt")
-outputs = model.generate(**inputs)
-print(tokenizer.decode(outputs[0], skip_special_tokens=True))
+Hardware: Google Colab 1 x T4 GPU