INDEX

Explanations

references to specific groups of people or demographic categories

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

aneously

-0.76

stals

-0.71

ctors

-0.66

izens

-0.64

unts

-0.64

ints

-0.63

 ..........

-0.63

stein

-0.63

neys

-0.62

 Galile

-0.60

POSITIVE LOGITS

heet

1.37

hip

1.21

hare

1.16

mith

1.16

cape

1.07

cale

1.06

ilver

1.04

pring

1.04

hift

1.03

pace

1.01

Activations Density 0.119%

references to specific groups of people or demographic categories

No Comments

No Known Activations