I don’t like how it punishes the user for exploring / not knowing the rules though. When ChatGPT refuses you just edit the prompt and do something else. Any refusal tar pit addressed by forced restarting is downstream of the missing edit button IMO.
LLMS
-
Grok’s Real-Time Event Summarization Capabilities Improve Rapidly
By
–
I've been pretty critical of Grok (and it keeps not working quite right), but its ability to do real-time summarizing of changing events is improving rapidly, and is differentiated from other LLMs. No idea of the actual accuracy though.
-
LLM Agent Solutions and Training Data Contamination Concerns
By
–
my biggest concern is that an LLM Agent can solve it in 1-2 years but by then related forks and code and discussions seep into its training data in an undefined way, making the result unsatisfying.
-
JSON Mode and Client-Side Retries for Timeout Issues
By
–
its totally fair. i actually had the instructor js docs open this morning. i ultimately just wanted to 1) try json mode over function calling, and 2) for this weird usecase i wanted retries to be done clientside as i run into timeout issues otherwise
-
LLM Agents Challenge: Porting PyTorch GPT-2 Training to C
By
–
Btw writing the llm.c training code would imo be a very interesting, impressive, self-contained and very meta challenge for LLM agents. The prompt is: Take the PyTorch code train_gpt2.py
And write, compile and unit test a single .c file that reproduces the training: train_gpt2.c -
PyTorch Layers Migration to C: Tutorial and Implementation Guide
By
–
I added a quick crappy tutorial on how PyTorch layers are moved to C, with a few possibly helpful pointers: https://
github.com/karpathy/llm.c
/blob/master/doc/layernorm/layernorm.md
… -
LLaMA 3 Release Expected to Surpass Mistral Models
By
–
LLaMA 3's will start to drop next week. Assuming there's a 7B version, I'm expecting it to far surpass the current Mistral model.
-
Llama2.c translations and broader training design possibilities
By
–
😀 😀 I really do love how llama2.c was translated to a zillion languages I've never heard about. Looking forward to that with the training too! The design space is a lot wider too because you can train stuff, not just inference a fixed, given thing.
-
Using Long Context Models to Manage Multi-File Projects
By
–
With long context models like Claude you can send an entire project – spanning multiple files – in a single prompt and then ask questions about it or ask it to rewrite sections of the code for you
-
Using LLM Claude Opus to Complete Code with Prompts
By
–
My favourite prompt from this post is still the first one. cat files_to_prompt/cli.py | llm -m opus –system 'finish this code for me' Love that we have tools that can do this now!