INDEX

Explanations

word pairs indicating conflict or competition

oai_token-act-pair · gpt-3.5-turbo

instances of confrontational or competitive situations

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

NV

-0.68

 Koen

-0.65

Pag

-0.63

awn

-0.63

Wag

-0.61

 sake

-0.61

Vu

-0.61

 Andersen

-0.57

Nem

-0.57

 Hollow

-0.57

POSITIVE LOGITS

senal

0.84

Submit

0.76

 withd

0.71

ailed

0.69

Apply

0.68

UES

0.66

POST

0.64

submit

0.63

imity

0.63

agra

0.62

Activations Density 0.117%

word pairs indicating conflict or competition

instances of confrontational or competitive situations

No Comments

No Known Activations