INDEX

Explanations

words related to preference or support

oai_token-act-pair · gpt-3.5-turbo

instances of the word "favor" and its variations, indicating preference or support

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

sis

-0.88

Ä±

-0.83

pt

-0.80

thur

-0.79

mberg

-0.77

RT

-0.76

hid

-0.76

att

-0.75

gren

-0.75

raz

-0.72

POSITIVE LOGITS

 favored

1.16

itism

1.07

 favoring

0.99

 favors

0.93

 favoured

0.93

nesday

0.86

 favorites

0.86

 whipping

0.80

 hitters

0.74

 favorable

0.73

Activations Density 0.008%

words related to preference or support

instances of the word "favor" and its variations, indicating preference or support

No Comments

No Known Activations