INDEX

Explanations

time-related references, particularly mentioning the span of years

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "last" followed by a number, indicating time periods

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

lees

-0.73

olor

-0.71

pherd

-0.71

aband

-0.69

planes

-0.65

á

-0.64

ollow

-0.64

 Granger

-0.64

apt

-0.64

selves

-0.64

POSITIVE LOGITS

decade

1.30

 millennium

1.02

few

1.01

 month

1.01

 week

0.98

 gasp

0.98

 century

0.97

 ditch

0.96

 installment

0.96

 couple

0.95

Activations Density 0.046%

time-related references, particularly mentioning the span of years

occurrences of the word "last" followed by a number, indicating time periods

No Comments

No Known Activations