INDEX

Explanations

terms related to being influenced or guided by a specific driving force or motivation

oai_token-act-pair · gpt-3.5-turbo

terms related to frameworks or systems that influence behavior or outcomes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 square

-0.85

 balls

-0.73

 conc

-0.68

 present

-0.66

 insign

-0.66

 squares

-0.65

 correct

-0.65

 nails

-0.65

 person

-0.64

 photograph

-0.64

POSITIVE LOGITS

dominated

2.19

heavy

2.10

driven

2.10

enabled

1.81

focused

1.72

turned

1.68

intensive

1.65

laden

1.63

centered

1.62

centric

1.55

Activations Density 0.035%

terms related to being influenced or guided by a specific driving force or motivation

terms related to frameworks or systems that influence behavior or outcomes

No Comments

No Known Activations