INDEX

Explanations

numerical values classified as counts

oai_token-act-pair · gpt-3.5-turbo

references to specific geographical or administrative divisions

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

 wrench

-0.71

 obser

-0.67

 embr

-0.65

 inflamm

-0.65

BIT

-0.63

 pity

-0.60

 Centauri

-0.60

 misunder

-0.60

 Harbor

-0.60

ovie

-0.58

POSITIVE LOGITS

enance

1.78

ries

1.38

ess

1.24

downs

1.12

esses

1.11

down

1.06

rified

0.98

reys

0.98

eenth

0.96

een

0.87

Activations Density 0.033%

numerical values classified as counts

references to specific geographical or administrative divisions

No Comments

No Known Activations