INDEX

Explanations

instances where someone is able to accomplish a task successfully

oai_token-act-pair · gpt-3.5-turbo

instances of the word "able" indicating capability or proficiency

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Alone

-0.72

 Parents

-0.72

 Caval

-0.70

 Yose

-0.69

 Deer

-0.68

 Famous

-0.66

origin

-0.66

 sterling

-0.65

 Warcraft

-0.65

 Tradition

-0.64

POSITIVE LOGITS

ioned

0.93

bodied

0.90

reys

0.89

't

0.80

itud

0.78

llor

0.77

 manage

0.77

 access

0.76

awaru

0.76

¶ħ

0.75

Activations Density 0.030%

instances where someone is able to accomplish a task successfully

instances of the word "able" indicating capability or proficiency

No Comments

No Known Activations