#logits • Tags • Transformer System

2 posts · Transformer System

Tag: #logits

M3.8 — Output Head and Decoding Policy

🗓 2026-07-13 • Transformer System • ⏱ 59 min read

A derivation- and implementation-oriented reconstruction of vocabulary projection, logits and loss gradients, weight tying, score processors, temperature, top-k/top-p, beam search, and local token-selection policy.

#numinens #transformer #output-head #lm-head #logits #log-softmax

Read →

M3.10 — Single-Request Inference Execution

🗓 2026-06-22 • Transformer System • ⏱ 71 min read

A derivation- and implementation-oriented study of one autoregressive request: prompt state, prefill/decode, last-row logits, token selection, KV append timing, positions, stopping, beam reorder, streaming, and cache equivalence.

#numinens #transformer #inference #autoregressive #decoding #logits

Read →

Su	Mo	Tu	We	Th	Fr	Sa
27	28	29	30	1	2	3
4	5	6	7	8	9	10
11	12	13	14	15	16	17
18	19	20	21	22	23	24
25	26	27	28	29	30	31

| #logits

Tag: #logits

M3.8 — Output Head and Decoding Policy

M3.10 — Single-Request Inference Execution