Language Models as Knowledge Bases: On Entity Representations, Storage Capacity, and Paraphrased Queries in 2021 these folks measure the exact number of 'statements' models can memorize (a) capacity scales linearly with parameters
(b) more training samples -> less memorization
LLMS
-
Language Models Knowledge Storage Capacity and Memorization Scaling
By
–
-
Knowledge Storage Capacity in Language Model Parameters
By
–
How Much Knowledge Can You Pack Into the Parameters of a Language Model? enter, transformers- in 2020 these folks had a different idea: once "pretraining" is a thing, you can measure the amount of 'facts' that the model 'knows', without training it at all
-
Capacity Measurements for Language Models Research
By
–
in prep for our new research dropping on ArXiv tomorrow (i think), here is a thread about…. CAPACITY MEASUREMENTS FOR LANGUAGE MODELS
-
Average of Five Top AI Models You Talk With Most
By
–
You are the average of the 5 people you talk to the most: Opus
Sonnet
o3
Gemini Pro
DeepSeek R1 -
Claude 3.5 Availability: A Game-Changer for AI Adoption
By
–
Till 3.5 comes, it basically doesn’t exist for me.
-
Gemini API improvements and model quality assessment
By
–
I love Gemini, for the record 🙂 They only recently fixed their caching situation, and their API is still too complicated, but the models are great.
-
LLM Coding Capabilities: From GPT-2 to GPT-4o Progress
By
–
> 2020: gpt2 can't write code
> 2021: gpt3 can't reliably write python
> 2022: instructgpt can't write blocks of code without syntax errors
> 2023: chatgpt can't do leetcode
> 2024: gpt4 can't debug CUDA
> 2025: gpt4o can't implement entire PR these LLMs are not to be trusted… -
Modify Script to Use LM Studio or Ollama Provider
By
–
Yes, so you could take my script and legit modify one line of code to use @lmstudio or @ollama as a provider and use your favorite model with it.
-
Year of the Agent: Critical Analysis of AI Advancement Limits
By
–
“2025 is the Year of the Agent” — everyone. But once you sweep away the hype, what actually changed? Here’s the condensed field report—from someone working in the field much before any "LLM" came to exist: 1. We (potentially) hit the prediction ceiling. Pre-training gains
-
Pine AI promoted as an execution-focused agent
By
–
AI is evolving: → Claude thinks
→ ChatGPT writes
→ Gemini searches
→ Pine acts This isn’t chat. This is execution. Try Pine AI now → https://
cutt.ly/Frvwful0 Real-world problems need real-world AI.