INDEX

Explanations

phrases related to political or public support

oai_token-act-pair · gpt-3.5-turbo

references to governmental or political support

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

cha

-0.75

 Hebdo

-0.70

 Grimoire

-0.62

 Wonderland

-0.61

 Flore

-0.60

 Nose

-0.58

tein

-0.58

rave

-0.57

 dilig

-0.56

ueller

-0.56

POSITIVE LOGITS

itism

0.96

arity

0.85

ability

0.79

 among

0.75

byn

0.73

for

0.72

 amongst

0.71

establishment

0.70

votes

0.70

 backing

0.70

Activations Density 0.079%

phrases related to political or public support

references to governmental or political support

No Comments

No Known Activations