INDEX

Explanations

references to previous states or values in a programming context

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

old or prior

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

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

httphttps

-0.50

iettivo

-0.48

bootstrapcdn

-0.47

 الحره

-0.46

 ब्रेकडाउन

-0.46

 EconPapers

-0.46

ConstraintMaker

-0.45

matchCondition

-0.44

pró

-0.44

modelBuilder

-0.44

POSITIVE LOGITS

Old

0.59



0.57

Old

0.56

old

0.52

 ancien

0.50

OLD

0.50

prevState

0.48

cũ

0.47

旧

0.47

Older

0.46

Activations Density 0.035%

references to previous states or values in a programming context

old or prior

No Comments

No Known Activations