INDEX

Explanations

conversational fragments

np_max-act · gemini-2.0-flash

words containing the “ho” letter sequence, especially at the start of a word (with strongest responses to hoof-related terms).

oai_token-act-pair · gpt-5 Triggered by @liuliujiujiu05

New Auto-Interp

Configuration

mwhanna/qwen3-4b-transcoders/layer_11.safetensors

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

163,840

Data Type

float32

Hook Name

blocks.11.mlp.hook_in

Architecture

transcoder

Context Size

8,192

Dataset

monology/pile-uncopyrighted

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 clone

-0.32

Clone

-0.30

 oversh

-0.28

éĢĴ

-0.28

-ste

-0.28

_clone

-0.27

èī

-0.27

 partial

-0.27

ivative

-0.27

æľīä»»ä½ķ

-0.26

POSITIVE LOGITS

won

0.27

æ¶¯

0.27

 Aren

0.25

æ¶¦æ»ĳ

0.25

 Mata

0.25

ACTION

0.25

 Flynn

0.24

ç®¡çĲĨåĪ¶åº¦

0.23

ä¸ĸ

0.23

oming

0.23

Activations Density 0.035%

conversational fragments

words containing the “ho” letter sequence, especially at the start of a word (with strongest responses to hoof-related terms).

No Comments

No Known Activations