nateraw/openchat_3.5-awq 🔢📝 → 📝
About
OpenChat: Advancing Open-source Language Models with Mixed-Quality Data

Example Output
"Write a poem about cheese. It should conclude that parmigiano reggiano is the GOAT. "
Output
In the realm of dairy delights, where the milk turns to magic,
There's a world of flavor and taste, that's far more than just static.
From the creamy to the crumbly, the soft to the hard,
There's a cheese for every palate, that's been known to make us all card.
From the land of the mushroom, where the cows graze on grass,
Comes a cheese that's a champion, that's made to be class.
It's rich in flavor and history, with a rind that's quite rough,
But beneath its hard exterior, lies a world so soft and tough.
Oh, parmigiano reggiano, you're the king of the cheese throne,
The master of flavor, the one that we all adore.
From the hills of Italy, you've been crafted with care,
A legacy of tradition, that's been passed down through the years.
Your rind may be hard, but your heart is softer still,
A symphony of flavors, that's known to make us all thrill.
Your nutty, savory taste, is a gift from the gods above,
A sprinkle on pasta, or a slice on a knife, you're the one we all love.
For in this cheese-filled world, there are many to choose,
But when it comes to the GOAT, there's only one we can muse.
So raise a glass to parmigiano, and let's give him his due,
For he's the king of the cheeses, and we all know it's true.
Performance Metrics
All Input Parameters
{ "top_k": 50, "top_p": 0.95, "prompt": "Write a poem about cheese. It should conclude that parmigiano reggiano is the GOAT. ", "temperature": 0.8, "max_new_tokens": 512, "prompt_template": "GPT4 User: {prompt}<|end_of_turn|>GPT4 Assistant:", "presence_penalty": 0, "frequency_penalty": 0 }
Input Parameters
- top_k
- The number of highest probability tokens to consider for generating the output. If > 0, only keep the top k tokens with highest probability (top-k filtering).
- top_p
- A probability threshold for generating the output. If < 1.0, only keep the top tokens with cumulative probability >= top_p (nucleus filtering). Nucleus filtering is described in Holtzman et al. (http://arxiv.org/abs/1904.09751).
- prompt (required)
- temperature
- The value used to modulate the next token probabilities.
- max_new_tokens
- The maximum number of tokens the model should generate as output.
- prompt_template
- The template used to format the prompt. The input prompt is inserted into the template using the `{prompt}` placeholder.
- presence_penalty
- Presence penalty
- frequency_penalty
- Frequency penalty
Output Schema
Output
Example Execution Logs
Generated 372 tokens in 4.583674192428589 seconds.
Version Details
- Version ID
ded16ea9889fe7c536c105b0b5f5142db79e4e38f92db2703e0ff59da1c10999
- Version Created
- November 9, 2023