INDEX

Explanations

superlative terms indicating extreme positivity or greatness

oai_token-act-pair · gpt-3.5-turbo

the emphasis on "greatest" in various contexts or comparisons

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

scrib

-0.92

ttes

-0.85

skirts

-0.81

kel

-0.80

amber

-0.79

chini

-0.78

enda

-0.77

wat

-0.76

ourke

-0.75

lishes

-0.74

POSITIVE LOGITS

 possible

0.91

 accomplishment

0.90

 achievement

0.89

 conceivable

0.87

 exponent

0.86

 obstacle

0.84

 gift

0.81

 unsolved

0.80

 boon

0.80

 challenge

0.79

Activations Density 0.055%

superlative terms indicating extreme positivity or greatness

the emphasis on "greatest" in various contexts or comparisons

No Comments

No Known Activations