Stephen Blum
stephenblum@mastodon.social
<p>CTO @ PubNub | $134M raised | Series E | +1 Billion devices | 68 Patent claims | 25 Years</p>
Posts
-
Post #2768560
Built Spoken and an OBS interface for fast speech to text. #speech #javascript #streaming
-
Post #2768558
We built and trained our own embedding model from scratch today. #ai #embeddings #machinelearning
-
Post #2768557
Maybe we are gel packs inside a star powered simulation. #simulation #philosophy #science
-
Post #2768556
I thought she coded it, but it was her engineer instead. #ai #coding #milestones
-
Post #2768555
AI learning backward shows huge negative number around 1e21 today. #ai #math #debugging
-
Post #2768554
Today we learn PyTorch autograd and write a simple loss function. #pytorch #autograd #machinelearning
-
Post #2768553
Free twice, then they pay you. Wild pricing page. #pricing #startup #marketing
-
Post #2768552
Loss dropped slightly after one epoch, so yes it worked. #machinelearning #training #deeplearning
-
Post #2768551
Kernel changes shift output sizes and reveal a tutorial bug. #deeplearning #cnn #debugging
-
Post #2768549
Great UX means users do almost nothing and still succeed. #ux #ui #ai
-
Post #2768548
Start now, learn along the way, and keep moving forward. #motivation #growth #mindset
-
Post #2768547
Today we test multiple optimizers together to see what happens. #ai #deeplearning #training
-
Post #2768546
We are okay, just ignore the marketing hype for now. #marketing #hype #perspective
-
Post #2768545
I see the grass outside, but I still never touch it. #nature #mindset #life
-
Post #2768544
Benchmark Burn versus PyTorch on one million parameters and compare speed. #rust #pytorch #benchmark
-
Post #2768543
Clicked a button, ran code remotely, and watched it execute live. #coding #automation #software
-
Post #2768542
Train an AI today, despite bugs, and level up fast. #ai #coding #streaming
-
Post #2768541
Choose a license so others can use code fairly and safely. #opensource #licensing #software
-
Post #2768540
LoRA lets you save tiny updates, not huge model copies. #ai #lora #llm
-
Post #2768539
Ship an MVP fast, no matter what tools you use. #startup #mvp #product