INDEX

Explanations

specific terms related to biological or chemical processes and interactions

oai_token-act-pair · gpt-4o-mini Triggered by @bot

complex scientific/technical notation and abbreviations.

oai_token-act-pair · claude-3-5-haiku-20241022 Triggered by @neilrathi

The neuron detects punctuation characters (forward slashes, hyphens, etc.) commonly used in scientific and technical notation.

oai_token-act-pair · claude-3-7-sonnet-20250219 Triggered by @neilrathi

paired terms or abbreviations linked with slashes or en dashes, especially in technical or scientific nomenclature.

oai_token-act-pair · gpt-5 Triggered by @jyhe0408

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_16k/average_l0_68

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 ſol

-0.46

InputTagHelper

-0.41

 paſſ

-0.41

 Inscrivez

-0.39

ſp

-0.38

 Inſ

-0.38

 ſeveral

-0.37

 Sünde

-0.37

ériale

-0.36

fällig

-0.36

POSITIVE LOGITS

ImageContext

0.60

帖最后由

0.58

Rujuakan

0.55

AutoScaleMode

0.55

 Numerade

0.54

Personensuche

0.54

complexContent

0.50

SequentialGroup

0.49

 nahilalakip

0.46

 gynhyrchwyd