@zetalyrae Let me introduce you to an improved version: now the fly can detect the smell of the vibe code with its neurons and will fly straight to the sources of your B2B AI SAAS startup - https://t.co/7YFd4h8tuu
@waterloo_intern >Within a block, we do q(kᵀv). This is score first, the normal attention order with masking.
I think there is a typo in formula:
```
attn=(q_c@k_c.transpose(-1,-2)).tril()
o_curr=attn@v_c
```
is a (qkᵀ)v
Tired of getting rejects and reading reviewer comments that clearly missed the point of your paper?
I built an peer review skill for Claude Code that actually gives you useful feedback before you submit.
👇
I finally finished my little pet project: a 854K parameter transformer that fits on an Artix-7 FPGA. Runs at 95 MHz with Fmax around 98.8, generating around 1100 tokens/sec
https://t.co/TtI9nCDmHT