microsoft/phi-3-mini-128k-instruct 🔢📝 → 📝
About
Phi-3-Mini-128K-Instruct is a 3.8 billion-parameter, lightweight, state-of-the-art open model trained using the Phi-3 datasets

Example Output
"How would you explain Internet to a medieval knight?"
Output
Good morrow, noble Sir Knight! I shall endeavor to elucidate the concept of this 'Internet' in terms thou might comprehend from thy own experiences and knowledge.
Imagine, if thou wilt, an immense library vast as all the lands under our kingdoms combined. This grand repository is not bound by walls or stone; it stretches across every corner of the realm, accessible at any time when daylight graces us with its presence. It contains countless scrolls, tomes, and manuscripts on diverse subjects - tales of valorous deeds, wisdom of scholars, maps of distant territories, songs of minstrels, and even images that capture moments frozen in time.
Now, imagine these scrolls being able to communicate instantly with one another without needing messengers or couriers. They can share their contents freely amongst themselves, allowing anyone who seeks knowledge to access them directly. These scrolls could be written by common folk, nobles alike, merchants, bards, and sages. Each scroll holds information about different aspects of life, just like how each person has unique skills and knowledge.
This magical library we speak of is what ye call the "Internet". It exists within mystic devices known as computers, which act as portals for accessing this wealth of information. To use such a device, one must learn certain incantations (or rather, commands) to navigate through this expansive collection of scrolls. Yet remember, while this magic offers great power, it should always be used wisely and responsibly, much like your chivalric code guides thee in thine actions.<|end|>
Performance Metrics
All Input Parameters
{ "top_k": 1, "top_p": 1, "prompt": "How would you explain Internet to a medieval knight?", "max_length": 512, "temperature": 0.1, "system_prompt": "You are a helpful AI assistant.", "repetition_penalty": 1.1 }
Input Parameters
- seed
- The seed for the random number generator
- top_k
- When decoding text, samples from the top k most likely tokens; lower to ignore less likely tokens.
- top_p
- When decoding text, samples from the top p percentage of most likely tokens; lower to ignore less likely tokens.
- prompt (required)
- Text prompt to send to the model.
- max_length
- Maximum number of tokens to generate. A word is generally 2-3 tokens.
- temperature
- Adjusts randomness of outputs, greater than 1 is random and 0 is deterministic.
- system_prompt
- System prompt.
- repetition_penalty
- Penalty for repeated words in generated text; 1 is no penalty, values greater than 1 discourage repetition, less than 1 encourage it.
Output Schema
Output
Version Details
- Version ID
45ba1bd0a3cf3d5254becd00d937c4ba0c01b13fa1830818f483a76aa844205e
- Version Created
- April 25, 2024