INDEX

Explanations

expressions of gratitude

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 38-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_38/width_16k/average_l0_128

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.38.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

wijl

-0.48

 怎样

-0.47

ThroughAttribute

-0.47

mobileqq

-0.45

 peggio

-0.45

⟬

-0.42

themselves

-0.42

 addirittura

-0.41

 kanssa

-0.40

льності

-0.40

POSITIVE LOGITS

 very

0.84

 again

0.68

 kindly

0.65

 everyone

0.64

 ever

0.59

 guys

0.59

 heaps

0.58

 bunches

0.56

very

0.55

 VERY

0.53

Activations Density 0.069%

expressions of gratitude

No Comments

No Known Activations