like history hasn’t proven how easy LLMs are hackable.
LLMS
-
5 ChatGPT Prompts for Students to Boost Studying
By
–
I'm shocked why most students don't use ChatGPT for studying. Here are 5 ChatGPT Prompts for students to know about:
-
Synthetic data from large models boost small models to top performance
By
–
Really useful! For a specific task, a synthetic dataset generated from a bigger model can be all you need to bring a smaller model all the way up to its bigger sibling's performance!
-
Fast LLaMA 3 Fine-tuning with ORPO on 8xH100 GPUs
By
–
all info available in the model page: https://
huggingface.co/abhishek/autot
rain-llama3-orpo
… 🙂 finetuning took ~30 mins on 8xH100 -
Mixing API and Local Models for Terminal Usage
By
–
If you want to mix API model and local model usage together, or use API models from the terminal generally
-
AutoTrain Fine-tuned Llama3 Outperforms Base Instruct Model
By
–
AutoTrain finetuned llama3 beats the base instruct model on all but one benchmark on the open llm leaderboard This model was finetuned using ORPO and is also publicly available. No code was written and all the setup took less than 10mins. `pip install autotrain-advanced`
-
LLM Technology Plateau: Disruption and Adoption Implications
By
–
I should also add that even if LLM technology plateaus at GPT-4 level (which I still think is unlikely), we are likely to see:
1) Continued improvement as new approaches developed
2) 5-10 years of continued disruption as we learn what GPT-4 level can do and start to adopt widely -
GPT-5 Predictions: Multi-modal generation from any input with precise refinement
By
–
Alright, GPT-5 Predictions: Generate text, images, video, 3D assets or music FROM text, images, video, or music (basically it can take anything as input, and produce any medium for output depending on the prompt direction) with much more precise refinement process to massage an
-
GPT-5 Race: OpenAI’s Lead in Advanced AI Model Development
By
–
The current state-of-play in the key question of how good AI gets: So much depends on GPT-5. OpenAI had a year+ lead in creating a GPT-4 class model. Now there are four GPT-4 class models. If exponential growth is still possible, OpenAI should be the first to show us. Or not.
-
MMLU Performance Shows Exponential Growth Over Time
By
–
I had the same thought when I was listening to this — I think it’s exponential if u plot MMLU (model performance) vs time (year achieved)