INDEX

Explanations

code-related commands and conditional statements

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

status and states

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

IntoConstraints

-0.78



-0.60

RTSC

-0.58

########.

-0.57

RTEE

-0.57

RTGC

-0.52

 GenerationType

-0.52

findpost

-0.50

EReference

-0.50

featureID

-0.50

POSITIVE LOGITS

status

0.49

 status

0.49

Status

0.41

 mouth

0.40

setStatus

0.36

 statuses

0.35

Rohy

0.35

getStatus

0.35

Appearance

0.34

 Status

0.34

Activations Density 0.008%

code-related commands and conditional statements

status and states

No Comments

No Known Activations