INDEX

Explanations

comparative words indicating varying degrees

oai_token-act-pair · gpt-3.5-turbo

instances of the word "way" in various contexts

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

tein

-0.76

uster

-0.74

usters

-0.72

 prosec

-0.70

encer

-0.68

lict

-0.67

iners

-0.67

 livest

-0.67

encers

-0.63

querque

-0.63

POSITIVE LOGITS

fare

1.08

finding

1.00

ward

0.99

point

0.98

points

0.93

forward

0.88

bos

0.83

WARD

0.79

bill

0.77

YY

0.76

Activations Density 0.035%

comparative words indicating varying degrees

instances of the word "way" in various contexts

No Comments

No Known Activations