INDEX

Explanations

comparative relationships expressed through words relating to quantity or intensity

oai_token-act-pair · gpt-3.5-turbo

comparative statements related to degree or intensity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

«ĺ

-0.84

Ń·

-0.78

è£

-0.78

ãĤ´ãĥ³

-0.75

ï¸

-0.73

WAR

-0.73

ELS

-0.70

galitarian

-0.70

imilar

-0.70

lain

-0.69

POSITIVE LOGITS

than

0.84

 than

0.74

 rede

0.73

 absorbed

0.68

 renew

0.66

 churn

0.66

 absorb

0.65

 induct

0.64

 absorbs

0.63

 dwell

0.63

Activations Density 0.104%

comparative relationships expressed through words relating to quantity or intensity

comparative statements related to degree or intensity

No Comments

No Known Activations