INDEX

Explanations

words related to formats that indicate accessibility or usability

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

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

 TÉCN

-0.47

ups

-0.45

 match

-0.45

rungsseite

-0.45

ing

-0.45

 strøm

-0.44

 look

-0.44

 Inscrivez

-0.43

 context

-0.43

 first

-0.43

POSITIVE LOGITS

illable

0.97

vable

0.90

ctable

0.86

ggable

0.85

entable

0.84

izable

0.84

codable

0.84

ortable

0.83

chable

0.82

tainable

0.82

Activations Density 0.222%

words related to formats that indicate accessibility or usability

No Comments

No Known Activations