INDEX

Explanations

phrases indicating directionality or preference

oai_token-act-pair · gpt-3.5-turbo

phrases referring to directions or methods, often using the word "way."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

uster

-0.89

usters

-0.84

ĸļ

-0.76

erville

-0.64

ifer

-0.64

Ore

-0.64

 Explos

-0.59

iasco

-0.58

teness

-0.57

 subsequ

-0.56

POSITIVE LOGITS

fare

1.28

finding

1.20

ward

1.19

WARD

0.97

forward

0.93

point

0.88

points

0.87

finder

0.84

lay

0.80

bill

0.80

Activations Density 0.063%

phrases indicating directionality or preference

phrases referring to directions or methods, often using the word "way."

No Comments

No Known Activations