INDEX

Explanations

names ending with "ns"

oai_token-act-pair · gpt-3.5-turbo

references to locations or places

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

PDATE

-0.75

zzi

-0.68

 subdiv

-0.62

 sorcery

-0.62

BIL

-0.61

 crank

-0.60

CONCLUS

-0.59

 slee

-0.59

 speculative

-0.58

holding

-0.58

POSITIVE LOGITS

fw

1.05

ertodd

1.05

hers

1.03

uit

0.96

ailing

0.95

uits

0.95

ayer

0.91

ettings

0.91

etsk

0.89

ail

0.88

Activations Density 0.012%

names ending with "ns"

references to locations or places

No Comments

No Known Activations