INDEX

Explanations

names of specific individuals

oai_token-act-pair · gpt-3.5-turbo

proper names, particularly of characters and deities

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

dinand

-0.77

arily

-0.76

iatus

-0.75

andise

-0.75

sson

-0.75

ainment

-0.73

alist

-0.73

antly

-0.72

gart

-0.70

ascript

-0.69

POSITIVE LOGITS

EStreamFrame

0.98

mpeg

0.75

yy

0.74

worthiness

0.74

Wa

0.73

pei

0.73

wei

0.73

ward

0.73

pher

0.73

peria

0.71

Activations Density 0.018%

names of specific individuals

proper names, particularly of characters and deities

No Comments

No Known Activations