INDEX

Explanations

terms related to numerical quantities or measurements

oai_token-act-pair · gpt-3.5-turbo

phrases indicating one-off events or occurrences

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

SIG

-0.73

 Borders

-0.70

Kuh

-0.66

ãĤ¼ãĤ¦ãĤ¹

-0.65

Dag

-0.64

Cla

-0.64

ãĤ©

-0.64

Hoo

-0.63

vel

-0.63

 Chero

-0.62

POSITIVE LOGITS

sided

1.32

dimensional

1.20

year

1.16

minute

1.14

day

1.09

handed

1.09

legged

1.06

sized

1.05

piece

1.05

time

1.05

Activations Density 0.041%

terms related to numerical quantities or measurements

phrases indicating one-off events or occurrences

No Comments

No Known Activations