Update README.md

README.md

@@ -23,43 +23,6 @@ This model is a fine-tuned version of LLaMA 3 8B using Alpaca-style instruction-
 - **Paper:** https://arxiv.org/abs/2302.13971 (LLaMA 3)
 - **Demo:** [Example usage with llama-cli]
 
-## Uses
-
-### Direct Use
-
-- Chatbots for general conversation
-- Instruction-following NLP tasks
-- Generating structured or unstructured text
-- Educational assistants or question answering
-
-### Downstream Use
-
-- Fine-tuning for task-specific NLP tasks
-- Integration into larger applications such as AI assistants or virtual agents
-
-### Out-of-Scope Use
-
-- Should not be used for high-stakes decision-making (medical, legal, or financial advice)
-- May generate biased, unsafe, or incorrect outputs if unchecked
-
-## Bias, Risks, and Limitations
-
-- Model reflects biases present in its training dataset.
-- May generate unsafe or offensive outputs if prompted maliciously.
-- Performance may vary with non-English inputs unless pre-translated.
-
-### Recommendations
-
-- Always monitor outputs in deployment.
-- Use prompt filtering or moderation for sensitive domains.
-- Combine with human-in-the-loop evaluation for critical tasks.
-
-## How to Get Started with the Model
-
-```bash
-llama-cli --model meta-llama-3.1-8b-alpaca.Q4_K_M.gguf -p "Hello Eliza"
-```
-
 ## Training Details
 
 ### Training Data
@@ -136,23 +99,4 @@ llama-cli --model meta-llama-3.1-8b-alpaca.Q4_K_M.gguf -p "Hello Eliza"
   year={2025},
   howpublished={\url{https://huggingface.co/unsloth/meta-llama-3.1-8b-alpaca}}
 }
-```
-
-**APA:**
-Anezatra. (2025). *LLaMA 3 8B Alpaca Fine-Tuned Model.* HuggingFace. https://huggingface.co/unsloth/meta-llama-3.1-8b-alpaca
-
-## Glossary
-
-- **LoRA:** Low-Rank Adaptation for efficient fine-tuning
-- **GGUF:** Quantized model format compatible with llama.cpp
-- **bf16:** Brain Floating Point 16-bit precision
-- **Q4_K_M:** Quantization method for reduced memory footprint
-
-## Model Card Authors
-
-- Anezatra
-
-## Model Card Contact
-
-- HuggingFace: https://huggingface.co/unsloth
-- Email: [email protected]
+```