INDEX

Explanations

negative descriptors or connotations related to characters or actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 36-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_36/width_16k/average_l0_120

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.36.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

WriteBarrier

-0.59

featureID

-0.50

Concern

-0.44

HasBeenSet

-0.42

OGND

-0.42

Personendaten

-0.42

prepareStatement

-0.42

مراجع

-0.41

KEYCODE

-0.41

IZON

-0.41

POSITIVE LOGITS

bad

0.84

Bad

0.82

Bad

0.81

bad

0.74

BAD

0.68

 luck

0.67

BAD

0.67

 news

0.65

gering

0.59

gered

0.59

Activations Density 0.172%

negative descriptors or connotations related to characters or actions

No Comments

No Known Activations