INDEX

Explanations

phrases that indicate a significant or essential change or concept

oai_token-act-pair · gpt-3.5-turbo

the word "fundamentally" and variations of it used to emphasize core changes or differences

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

ries

-0.77

ibal

-0.65

 Guys

-0.64

 supervisors

-0.64

 Commissioners

-0.64

 Leone

-0.64

 Slaughter

-0.64

icip

-0.63

tein

-0.63

 runners

-0.60

POSITIVE LOGITS

 opposed

0.81

 flawed

0.80

 differentiated

0.80

 wedd

0.80

ascript

0.78

 altering

0.74

 incapable

0.74

 embodied

0.72

 reliant

0.71

gebra

0.71

Activations Density 0.021%

phrases that indicate a significant or essential change or concept

the word "fundamentally" and variations of it used to emphasize core changes or differences

No Comments

No Known Activations