INDEX

Explanations

references to time in terms of years

oai_token-act-pair · gpt-3.5-turbo

phrases related to yearly occurrences or trends

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

 contag

-0.66

 Kardashian

-0.60

 guiActiveUn

-0.58

ecast

-0.58

atile

-0.58

 Haram

-0.57

IRC

-0.57

atility

-0.56

 RELE

-0.56

 Borders

-0.55

POSITIVE LOGITS

nings

1.26

ning

1.19

book

1.18

long

1.10

books

1.09

ned

1.07

 olds

1.04

lings

1.04

ns

0.97

days

0.93

Activations Density 0.085%

references to time in terms of years

phrases related to yearly occurrences or trends

No Comments

No Known Activations