Update fine-tuning-considerations.md

Author: ssalgadodev

articles/ai-services/openai/concepts/fine-tuning-considerations.md

@@ -39,19 +39,19 @@ Fine-tuning smaller models can achieve performance levels comparable to larger,
 
 Fine-tuning is suited for times when you have a small amount of data and want to improve the performance of your model. Fine-tuning can be for different kinds of use cases - but they often fall into broader categories. 
 
-**Reducing prompt engineering overhead**: Many users begin with few-shot learning, appending examples of desired outputs to their system message. Over time, this can lead to increasingly long prompts, driving up token counts and latency. Fine-tuning lets you embed these examples into the model by training on the expected outputs. This is particularly valuable in scenarios with numerous edge cases.
+* **Reducing prompt engineering overhead**: Many users begin with few-shot learning, appending examples of desired outputs to their system message. Over time, this can lead to increasingly long prompts, driving up token counts and latency. Fine-tuning lets you embed these examples into the model by training on the expected outputs. This is particularly valuable in scenarios with numerous edge cases.
 
-**Modifying style and tone**: Fine-tuning helps align model outputs with a desired style or tone, ensuring consistency in applications like customer service chatbots and brand-specific communication.
+* **Modifying style and tone**: Fine-tuning helps align model outputs with a desired style or tone, ensuring consistency in applications like customer service chatbots and brand-specific communication.
 
-**Generating outputs in specific formats or schemas**: Models can be fine-tuned to produce outputs in specific formats or schemas, making them ideal for structured data generation, reports, or formatted responses.
+* **Generating outputs in specific formats or schemas**: Models can be fine-tuned to produce outputs in specific formats or schemas, making them ideal for structured data generation, reports, or formatted responses.
 
-**Enhancing tool usage**: While the chat completions API supports tool calling, listing many tools increases token usage and may lead to hallucinations. Fine-tuning with tool examples enhances accuracy and consistency, even without full tool definitions.
+* **Enhancing tool usage**: While the chat completions API supports tool calling, listing many tools increases token usage and may lead to hallucinations. Fine-tuning with tool examples enhances accuracy and consistency, even without full tool definitions.
 
-**Enhancing retrieval-based performance**: Combining fine-tuning with retrieval methods improves a model’s ability to integrate external knowledge, perform complex tasks, and provide more accurate, context-aware responses. Fine-tuning trains the model to effectively use retrieved data while filtering out irrelevant information.
+* **Enhancing retrieval-based performance**: Combining fine-tuning with retrieval methods improves a model’s ability to integrate external knowledge, perform complex tasks, and provide more accurate, context-aware responses. Fine-tuning trains the model to effectively use retrieved data while filtering out irrelevant information.
 
-**Optimizing for efficiency**: Fine-tuning can also be used to transfer knowledge from a larger model to a smaller one, allowing the smaller model to achieve similar task performance with lower cost and latency. For example, production data from a high-performing model can be used to fine-tune a smaller, more efficient model. This approach helps scale AI solutions while maintaining quality and reducing computational overhead.
+* **Optimizing for efficiency**: Fine-tuning can also be used to transfer knowledge from a larger model to a smaller one, allowing the smaller model to achieve similar task performance with lower cost and latency. For example, production data from a high-performing model can be used to fine-tune a smaller, more efficient model. This approach helps scale AI solutions while maintaining quality and reducing computational overhead.
 
-**Distillation**: Model Distillation uses a large model's outputs to fine-tune a smaller model, allowing it to perform similarly on a specific task, for example collecting production traffic from an o1 deployment and using that as training data to fine tune 4o-mini. This can cut cost and latency since smaller models are usually more efficient. 
+* **Distillation**: Model Distillation uses a large model's outputs to fine-tune a smaller model, allowing it to perform similarly on a specific task, for example collecting production traffic from an o1 deployment and using that as training data to fine tune 4o-mini. This can cut cost and latency since smaller models are usually more efficient. 
 
 ## Types of fine-tuning