INDEX

Explanations

dates written with ordinal numbers

oai_token-act-pair · gpt-3.5-turbo

dates and numeric values

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 confir

-0.62

 streng

-0.60

 counselling

-0.59

 grips

-0.58

channelAvailability

-0.56

 Genie

-0.56

 sugg

-0.56

 nodd

-0.55

 explan

-0.55

 glim

-0.54

POSITIVE LOGITS

th

1.82

rd

1.28

tha

1.08

TH

1.08

ths

1.08

teenth

1.03

nd

1.01

st

1.01

thal

0.91

thus

0.90

Activations Density 0.079%

dates written with ordinal numbers

dates and numeric values

No Comments

No Known Activations