INDEX

Explanations

phrases expressing strong recommendations or endorsements

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_20/width_16k/average_l0_71

Prompts (Dashboard)

36,864 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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

RenderAtEndOf

-0.65

 caufe

-0.63

Autoritní

-0.63

 pleaſure

-0.63

 fubject

-0.63

 Majefty

-0.62

 שוליים

-0.60

 cauſe

-0.60

 ſmall

-0.59

elry

-0.57

POSITIVE LOGITS

 recommend

1.34

 recommends

1.05

 recommended

1.01

 highly

1.00

recommend

0.98

 Recommend

0.98

 strongly

0.94

Recommend

0.94

 RECOMMEND

0.92

 recommending

0.91

Activations Density 0.078%

phrases expressing strong recommendations or endorsements

No Comments

No Known Activations