Amey9766
/

llama32B-hospitality-review-triage

@@ -1,56 +1,80 @@
-🏨 LLaMA-3.2 Hotel Review Triage Model
-🔹 Model Overview
-Model Name: llama32-hotel-review-triage
-Base Model: meta-llama/Llama-3.2-1B-Instruct
-Domain: Hospitality / Hotel Operations
-Task: Hotel review triage → structured JSON output
-Fine-tuning Method: Supervised Fine-Tuning (SFT) with LoRA, merged into base
-Language: English
-This model converts unstructured hotel guest reviews into structured, machine-readable JSON, enabling automated complaint routing, severity detection, and service analytics.
-🎯 Intended Use
-✅ Primary Use Cases
-Hotel guest review analysis
-Complaint triage and categorization
-Department routing (housekeeping, engineering, front desk)
-Severity and priority detection
-Input preprocessing for dashboards and ticketing systems
-🧠 Typical Applications
-Review ingestion pipelines (Google Reviews, TripAdvisor, surveys)
-Hospitality analytics platforms
-AI-powered hotel service agents
-Internal customer experience tools
-🧾 Input & Output Format
-Input
-Plain-text hotel guest review
-Output
-Strict JSON only
-No explanations, no natural language
-Example Output
 {
   "issues": [
     {
@@ -69,238 +93,96 @@ Example Output
   "overall_sentiment": "negative",
   "priority": "high"
 }
-🏗 Model Architecture
-Architecture: Decoder-only causal language model
-Parameters: ~1B
-Backbone: LLaMA 3.2
-Tokenizer: LLaMA 3.2 tokenizer with preserved chat template
-Precision: BF16 / FP16
-Fine-Tuning Details
-LoRA adapters applied to:
-Attention projections
-MLP layers
-Adapters merged post-training for standalone deployment
-📊 Training Details
-Dataset
-Source: Public hotel review dataset
-Domain: Real-world hotel guest feedback
-Language: English
-Total Training Examples: 120,000
-Validation Set: Held-out split used during training
-Preprocessing
-Removed hotel names and dates to reduce memorization
-Converted reviews into instruction-style chat format
-Supervised training toward structured JSON outputs
-⚙️ Training Configuration
-Epochs: 3
-Max Sequence Length: 512
-Optimizer: AdamW
-Hardware: NVIDIA A100 GPU
-Training Time: ~10 hours
-Fine-Tuning Strategy: QLoRA-style training, merged after completion
-📈 Evaluation
-Quantitative Metrics
-Validation Perplexity: 3.02
-A perplexity of 3.02 indicates strong domain adaptation and fluent generation within hospitality review data.
-Qualitative Evaluation
-The model was evaluated on:
-Multi-issue complaints
-Mixed positive and negative reviews
-Policy-related feedback
-Hygiene and safety-critical cases
-Observed Strengths
-Consistent JSON formatting
-Accurate department routing
-Appropriate severity assignment
-Robust handling of noisy real-world text
-⚠️ Limitations
-Trained only on English hotel reviews
-Not suitable for legal, medical, or safety-critical decisions
-JSON schema is not formally enforced (prompt-dependent)
-May struggle with:
-Very short or sarcastic reviews
-Highly ambiguous feedback
-Non-hotel hospitality domains (e.g., airlines, cruises)
-This model should be used as a decision-support system, not as a final authority.
-���️ Ethical Considerations
-May reflect biases present in user-generated reviews
-Should not be used for profiling individuals
-No personal or sensitive data should be passed into the model
-🚀 How to Use
-from transformers import AutoTokenizer, AutoModelForCausalLM
-model_id = "Amey9766/llama32-hotel-review-triage"
-tokenizer = AutoTokenizer.from_pretrained(model_id)
-model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto")
-prompt = tokenizer.apply_chat_template(
-    [
-        {"role": "system", "content": "You are a hospitality review triage assistant. Output ONLY valid JSON."},
-        {"role": "user", "content": "The room was dirty and the AC didn’t work."}
-    ],
-    tokenize=False
-)
-inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
-output = model.generate(**inputs, max_new_tokens=256)
-print(tokenizer.decode(output[0], skip_special_tokens=True))
-📚 Citation
-If you use this model in research or applied systems, please cite:
-@misc{tillu2026llama32hoteltriage,
-  title        = {LLaMA-3.2 Hotel Review Triage Model},
-  author       = {Amey Tillu},
-  year         = {2026},
-  howpublished = {\url{https://huggingface.co/Amey9766/llama32-hotel-review-triage}},
-  note         = {Fine-tuned on hospitality reviews for structured JSON triage}
-}
-📜 License
-This model inherits the license of its base model:
-LLaMA 3.2 License (Meta)
-Please ensure compliance with the base model’s license when using this model.
-🙏 Acknowledgements
-Meta AI for the LLaMA 3.2 base model
-Hugging Face ecosystem (Transformers, PEFT, TRL)
-Public hotel review datasets used for training
-If you want next-level polish, I can:
-optimize tags + metadata for Hugging Face search
-add visual badges (perplexity, domain, task)
-write a short launch post for LinkedIn / X
-Just tell me 👍
-it is still looking weird
-You’re right — what you have works, but Hugging Face renders Markdown very literally.
-Right now your card looks like a wall of text because it’s missing clear hierarchy, spacing, bullets, and visual anchors.
-Below is a clean, well-structured, Hugging Face–optimized model card that you can copy-paste directly into README.md.
-It will render nicely with bold headings, bullets, code blocks, and emojis (used sparingly).
-🚀 LLaMA-3.2 Hotel Review Triage Model
-Model ID: Amey9766/llama32-hotel-review-triage
-Base Model: meta-llama/Llama-3.2-1B-Instruct
-Author: Amey Tillu
-Task: Hospitality Review Triage → Structured JSON Output
-Perplexity (validation): 3.02
-🧠 What This Model Does
-This model is fine-tuned to analyze hotel guest reviews and convert unstructured feedback into clean, machine-readable JSON suitable for:
-Complaint routing
-Guest experience analytics
-Operational alerting
-AI hotel agents
-It is designed to behave as a deterministic triage system, not a conversational chatbot.
-📌 Core Capabilities
-Identifies complaint category (e.g. cleanliness, staff, facilities)
-Estimates severity level (low / medium / high)
-Infers department ownership (housekeeping, front desk, maintenance)
-Produces strict, valid JSON output
-Handles both negative complaints and positive feedback
-🏨 Example Use Case
-Input
-The room was dirty and the AC didn’t work.
-Output
-{
-  "category": "room_maintenance",
-  "severity": "high",
-  "department": "engineering",
-  "sentiment": "negative",
-  "summary": "Guest reports cleanliness issues and malfunctioning air conditioning."
-}
-🧪 How to Use
 from transformers import AutoTokenizer, AutoModelForCausalLM
 model_id = "Amey9766/llama32-hotel-review-triage"
@@ -313,14 +195,8 @@ model = AutoModelForCausalLM.from_pretrained(
 prompt = tokenizer.apply_chat_template(
     [
-        {
-            "role": "system",
-            "content": "You are a hospitality review triage assistant. Output ONLY valid JSON."
-        },
-        {
-            "role": "user",
-            "content": "The room was dirty and the AC didn’t work."
-        }
     ],
     tokenize=False
 )
@@ -329,57 +205,13 @@ inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
 output = model.generate(**inputs, max_new_tokens=256)
 print(tokenizer.decode(output[0], skip_special_tokens=True))
-📊 Training Details
-Dataset: Public hotel review datasets (Hugging Face)
-Total examples used: ~120,000
-Train / Validation split: 90% / 10%
-Sequence length: 512 tokens
-Fine-tuning method: LoRA (PEFT)
-Training hardware: NVIDIA A100
-Training time: ~10 hours
-Epochs: 3
-📈 Evaluation
-Validation Perplexity: 3.02
-Stable loss convergence
-No observed mode collapse
-Minor variance expected due to diverse review lengths and tone
-⚠️ Limitations
-Not trained on live hotel policies or property-specific rules
-Does not replace human judgment for legal or compensation decisions
-Assumes English-language reviews
-Output schema is opinionated and may require adaptation
-🔐 License
-This model inherits the license of its base model:
-LLaMA 3.2 License (Meta)
-Please ensure compliance with the base model’s license when using or redistributing this model.
-📚 Citation
-If you use this model in research or production systems, please cite:
 @misc{tillu2026llama32hoteltriage,
   title        = {LLaMA-3.2 Hotel Review Triage Model},
   author       = {Amey Tillu},
@@ -387,23 +219,31 @@ If you use this model in research or production systems, please cite:
   howpublished = {\url{https://huggingface.co/Amey9766/llama32-hotel-review-triage}},
   note         = {Fine-tuned on hospitality reviews for structured JSON triage}
 }
-🙏 Acknowledgements
-Meta AI for the LLaMA-3.2 base model
-Hugging Face ecosystem (Transformers, PEFT, TRL)
-Public hospitality review datasets used for training
-⭐ When to Use This Model
-Use this model when you need:
-Reliable structured outputs
-Fast review triage
-Integration into hotel operations pipelines
-A foundation for AI hospitality agents

+---
+license: mit
+language:
+- en
+base_model:
+- meta-llama/Llama-3.2-3B-Instruct
+---
+---
+license: llama3
+tags:
+  - hotel-review-triage
+  - hospitality
+  - structured-output
+  - json-generation
+  - complaint-routing
+  - llama
+  - sft
+  - lora
+  - transformers
+  - bitsandbytes
+model_name: llama32-hotel-review-triage
+base_model: meta-llama/Llama-3.2-1B-Instruct
+datasets:
+  - public-hotel-reviews
+language: en
+pipeline_tag: text-generation
+inference: true
+---
+# 🏨 LLaMA‑3.2 Hotel Review Triage Model
+### **Structured JSON Extraction for Hospitality Operations**
+---
+## 🔹 Model Overview
+**Model Name:** `llama32-hotel-review-triage`
+**Base Model:** `meta-llama/Llama-3.2-1B-Instruct`
+**Domain:** Hospitality / Hotel Operations
+**Task:** Hotel review triage → structured JSON output
+**Fine‑tuning Method:** SFT + LoRA (merged)
+**Language:** English
+**Validation Perplexity:** **3.02**
+This model converts unstructured hotel guest reviews into **clean, machine‑readable JSON**, enabling automated complaint routing, severity detection, and service analytics.
+---
+## 🎯 Intended Use
+### **Primary Use Cases**
+- Hotel guest review analysis
+- Complaint triage & categorization
+- Department routing (housekeeping, engineering, front desk)
+- Severity & priority detection
+- Input preprocessing for dashboards & ticketing systems
+### **Typical Applications**
+- Review ingestion pipelines (Google Reviews, TripAdvisor, surveys)
+- Hospitality analytics platforms
+- AI‑powered hotel service agents
+- Internal customer experience tools
+---
+## 🧾 Input & Output Format
+### **Input**
+Plain‑text hotel guest review.
+### **Output**
+- Strict JSON
+- No explanations
+- No natural language outside JSON
+### **Example Output**
+```json
 {
   "issues": [
     {
   "overall_sentiment": "negative",
   "priority": "high"
 }
+```
+---
+## 🏗 Model Architecture
+- **Architecture:** Decoder‑only causal language model
+- **Parameters:** ~1B
+- **Backbone:** LLaMA 3.2
+- **Tokenizer:** LLaMA 3.2 tokenizer (chat template preserved)
+- **Precision:** BF16 / FP16
+### **LoRA Fine‑Tuning Details**
+Adapters applied to:
+- Attention projections
+- MLP layers
+Adapters merged post‑training for standalone deployment.
+---
+## 📊 Training Details
+### **Dataset**
+- Source: Public hotel review datasets
+- Domain: Real‑world guest feedback
+- Language: English
+- Training Examples: **120,000**
+- Validation: Held‑out split
+### **Preprocessing**
+- Removed hotel names & dates
+- Converted reviews into instruction‑style chat format
+- Supervised training toward structured JSON outputs
+### **Training Configuration**
+- Epochs: **3**
+- Max Sequence Length: **512**
+- Optimizer: **AdamW**
+- Hardware: **NVIDIA A100 GPU**
+- Training Time: ~10 hours
+- Strategy: QLoRA‑style training, merged after completion
+---
+## 📈 Evaluation
+### **Quantitative**
+- **Validation Perplexity:** 3.02
+### **Qualitative**
+Evaluated on:
+- Multi‑issue complaints
+- Mixed sentiment reviews
+- Policy‑related feedback
+- Hygiene & safety‑critical cases
+**Observed Strengths**
+- Consistent JSON formatting
+- Accurate department routing
+- Appropriate severity assignment
+- Robust handling of noisy real‑world text
+---
+## ⚠️ Limitations
+- Trained only on English hotel reviews
+- Not suitable for legal, medical, or safety‑critical decisions
+- JSON schema is prompt‑dependent
+- May struggle with:
+  - Very short or sarcastic reviews
+  - Highly ambiguous feedback
+  - Non‑hotel domains (airlines, cruises, etc.)
+Use as a **decision‑support tool**, not a final authority.
+---
+## ⚖️ Ethical Considerations
+- May reflect biases present in user‑generated reviews
+- Should not be used for profiling individuals
+- Avoid passing personal or sensitive data into the model
+---
+## 🚀 How to Use
+```python
 from transformers import AutoTokenizer, AutoModelForCausalLM
 model_id = "Amey9766/llama32-hotel-review-triage"
 prompt = tokenizer.apply_chat_template(
     [
+        {"role": "system", "content": "You are a hospitality review triage assistant. Output ONLY valid JSON."},
+        {"role": "user", "content": "The room was dirty and the AC didn’t work."}
     ],
     tokenize=False
 )
 output = model.generate(**inputs, max_new_tokens=256)
 print(tokenizer.decode(output[0], skip_special_tokens=True))
+```
+---
+## 📚 Citation
+```
 @misc{tillu2026llama32hoteltriage,
   title        = {LLaMA-3.2 Hotel Review Triage Model},
   author       = {Amey Tillu},
   howpublished = {\url{https://huggingface.co/Amey9766/llama32-hotel-review-triage}},
   note         = {Fine-tuned on hospitality reviews for structured JSON triage}
 }
+```
+---
+## 📜 License
+This model inherits the license of its base model:
+**LLaMA 3.2 License (Meta)**
+Please ensure compliance when using or redistributing this model.
+---
+## 🙏 Acknowledgements
+- Meta AI for the LLaMA‑3.2 base model
+- Hugging Face ecosystem (Transformers, PEFT, TRL)
+- Public hospitality review datasets used for training
+---
+## ⭐ When to Use This Model
+Use this model when you need:
+- Reliable structured outputs
+- Fast review triage
+- Integration into hotel operations pipelines
+- A foundation for AI hospitality agents