INDEX

Explanations

superlatives denoting the highest or greatest degree

oai_token-act-pair · gpt-3.5-turbo

instances of the word "largest" and variations of the term to indicate comparisons or superlatives

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

jad

-0.82

haw

-0.76

bara

-0.74

ulton

-0.73

earth

-0.73

bal

-0.71

Mub

-0.71

lah

-0.70

amber

-0.70

bel

-0.70

POSITIVE LOGITS

gest

0.89

 intestine

0.82

(>

0.80

 extant

0.79

 scale

0.79

 largest

0.77

 proport

0.76

 profitable

0.75

 shareholder

0.74

 creditor

0.74

Activations Density 0.026%

superlatives denoting the highest or greatest degree

instances of the word "largest" and variations of the term to indicate comparisons or superlatives

No Comments

No Known Activations