INDEX

Explanations

numerical listings and rankings of various items or entities, such as "top 10" or "best"

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate rankings or lists of items

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

ittee

-0.82

riber

-0.79

phasis

-0.78

clair

-0.78

adra

-0.78

reon

-0.77

ighter

-0.75

ipient

-0.74

ADRA

-0.74

matter

-0.74

POSITIVE LOGITS

 quotes

1.13

 moments

1.07

 places

1.07

 unsolved

1.02

 destinations

1.00

 cities

1.00

 stories

0.98

 ways

0.98

 locations

0.98

 albums

0.97

Activations Density 0.149%

numerical listings and rankings of various items or entities, such as "top 10" or "best"

phrases that indicate rankings or lists of items

No Comments

No Known Activations