INDEX

Explanations

question answering

np_max-act · gemini-2.0-flash

section headers and meta-format markers that label parts of structured prompts or dialogues, especially label-plus-colon formatting like “X: …”

oai_token-act-pair · gpt-5 Triggered by @vetterc0

New Auto-Interp

Configuration

andyrdt/saes-qwen2.5-7b-instruct/resid_post_layer_15/trainer_1

Dataset (Dashboard)

Various

Features

131,072

Data Type

float32

Hook Name

blocks.15.hook_resid_post

Architecture

standard

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

สโม

-0.08

有所帮助

-0.07

邿

-0.07

 hely

-0.06

','-

-0.06

 grep

-0.06

裛

-0.06

打卡

-0.06

.scrollHeight

-0.06

 ylim

-0.06

POSITIVE LOGITS

ID

0.08

QC

0.08

CHE

0.07

筛

0.07

hide

0.07

GRID

0.06

ricane

0.06

 Thousands

0.06

 becomes

0.06

 sound

0.06

Activations Density 0.079%

question answering

section headers and meta-format markers that label parts of structured prompts or dialogues, especially label-plus-colon formatting like “X: …”

No Comments

No Known Activations