Why does this one perform substantially worse in some tests?
What is the difference between yours and the other one I listed below, other than the fact that yours lacks censorship?
cognitivecomputations: dolphin-2.9-llama3-8b-q8_0.gguf (Failed 9 out of 10):
- She picked up an apple from the ground and took a bite.
- He saw a red apple on the tree, glistening in the sun.
- They painted a beautiful picture of a green apple against a blue sky.
- The little girl was excited to have her first taste of the crunchy apple.
- In the story, the main character discovered an enchanted apple that turned everything it touched into gold.
- As she walked by the orchard, she inhaled the sweet aroma of thousands of apples.
- After playing all day at the park, he couldn't wait to have his favorite apple slices with mom's homemade sauce.
- When the farmer handed her an entire peck of fresh, crisp apples, she felt incredibly grateful and lucky.
- In the midst of a forest filled with various fruit trees, the ancient Granny Smith apple tree stood tall, proud, and strong.
- The sweet and tart flavors of the freshly picked apples combined to make a delightful, mouthwatering treat for all ages.
Bartowksi model Meta-Llama-3-8B-Instruct-Q8_0.gguf (Succeeded 10 out of 10) :
- The teacher handed out a freshly picked apple.
- She decided to take a bite into the crisp apple.
- He carefully placed the juicy red apple.
- After lunch, they all sat under the tree and shared an apple.
- His wife loved to bake delicious pies with fresh apple.
- The farmer's market was filled with baskets of sweet apple.
- She mixed it in a blender with some ripe banana and a crunchy apple.
- At the top of the hill stood a giant statue of a shiny green apple.
- They packed snacks for their hike, including a few fresh Granny Smith apple.
- The chef added a sprinkle of cinnamon to the warm, caramelized apple.
I still don't get it. These dolphin models have censorship removed, right? Then why does it for example, like in this case, lose the ability to write 10 sentences that end with the word "apple"? Then what's the point of having a model with censorship removed, if it becomes crippled?
All this new AI technology is so poorly explained everywhere, at times it feels that the focus of any AI youtube channel just seems to be to entertain fellow AI nerds, while ignoring any newcomers to AI.
I'm an experienced IT guy, and more often than I like I can't bloody understand what the heck they're talking about.
Why can't we have more excellent videos like this: "[1hr Talk] Intro to Large Language Models" by Andrej Karpathy?
To answer your question, we do not train from the instruct version, which is where you're seeing those improvements - we train the base model.
Oh, so that's how it works, now it all starts to make sense to me.
What kind of you to explain things in a way anyone can understand. It's people like you that encourage newcomers like me to continue this interesting AI journey, instead of getting frustrated, and losing interest. Hopefully one day I've gained enough understanding to share my own experiences and knowledge with others, like you're doing.
Thank you very much HiroseKoichi!