INDEX

Explanations

words related to violations, especially in sports contexts

oai_token-act-pair · gpt-3.5-turbo

mentions of fouls and related misconduct

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

SPONSORED

-0.72

igslist

-0.71

yrim

-0.68

[/

-0.66

iolet

-0.66

agall

-0.66

ãĥŃ

-0.65

ocobo

-0.64

FINE

-0.64

_>

-0.64

POSITIVE LOGITS

ball

0.82

 smelling

0.80

ing

0.78

ling

0.78

lings

0.74

luent

0.74

 foul

0.73

balls

0.73

weather

0.71

ries

0.70

Activations Density 0.051%

words related to violations, especially in sports contexts

mentions of fouls and related misconduct

No Comments

No Known Activations