INDEX

Explanations

proper nouns or specific names followed by the word "opened"

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "opened."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

grade

-0.73

bil

-0.64

ivable

-0.64

hov

-0.63

iable

-0.62

dos

-0.62

orum

-0.61

ford

-0.60

ffic

-0.59

ency

-0.59

POSITIVE LOGITS

 Doors

0.91

 doors

0.77

 exting

0.74

nesday

0.74

OWS

0.74

 wounds

0.73

urated

0.73

opening

0.73

SSL

0.73

 hostilities

0.73

Activations Density 0.035%

proper nouns or specific names followed by the word "opened"

occurrences of the word "opened."

No Comments

No Known Activations