INDEX

Explanations

instances of actions or events being done or happening

oai_token-act-pair · gpt-3.5-turbo

concepts related to reward and recognition

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Qiao

-0.68

asketball

-0.62

anka

-0.57

ß

-0.54

ather

-0.54

ansson

-0.54

ESV

-0.53

iannopoulos

-0.52

essen

-0.52

 Skydragon

-0.52

POSITIVE LOGITS

 thereof

1.31

 accordingly

1.09

 thereto

1.06

 them

1.04

 therein

1.00

 theirs

0.95

it

0.91

 THEM

0.79

 thereafter

0.78

 afterwards

0.75

Activations Density 1.818%

instances of actions or events being done or happening

concepts related to reward and recognition

No Comments

No Known Activations