INDEX

Explanations

the word "downtown" in various contexts

oai_token-act-pair · gpt-3.5-turbo

references to the concepts of "up" and "down" as well as their variations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

BuyableInstoreAndOnline

-0.85

 Mori

-0.79

UGC

-0.78

 Canaver

-0.77

Http

-0.75

 Franch

-0.75

REDACTED

-0.74

Closure

-0.73

Frag

-0.73

 Spur

-0.73

POSITIVE LOGITS

ownt

1.05

 downt

1.01

upt

0.98

eenth

0.94

imes

0.92

urned

0.91

upt

0.88

riter

0.87

itled

0.86

ivery

0.85

Activations Density 0.009%

the word "downtown" in various contexts

references to the concepts of "up" and "down" as well as their variations

No Comments

No Known Activations