INDEX

Explanations

words related to achievements or titles

oai_token-act-pair · gpt-3.5-turbo

various terms related to physical attributes and types of roles or identities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 iPads

-0.85

 positives

-0.81

 sockets

-0.78

 revolutions

-0.77

 embassies

-0.76

 notebooks

-0.75

 clusters

-0.73

 boxes

-0.73

 crates

-0.73

 vans

-0.72

POSITIVE LOGITS

ographer

1.20

ibrarian

1.18

ologist

1.14

earcher

1.12

iologist

1.11

angler

1.08

encer

1.04

alker

1.03

worker

1.03

owner

1.02

Activations Density 0.302%

words related to achievements or titles

various terms related to physical attributes and types of roles or identities

No Comments

No Known Activations