INDEX

Explanations

mentions of imaginary creatures or entities commonly associated with fiction or mythology

oai_token-act-pair · gpt-3.5-turbo

references to monsters

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Fairfax

-0.84

 Caldwell

-0.82

elect

-0.75

uates

-0.74

yang

-0.73

ansk

-0.72

 Mint

-0.72

Elect

-0.71

iland

-0.70

rin

-0.70

POSITIVE LOGITS

 monster

3.49

 monsters

2.90

monster

2.54

 Monster

2.52

Monster

2.35

 Monsters

2.23

 beast

1.73

 monstrous

1.66

 creature

1.53

 Godzilla

1.43

Activations Density 0.024%

mentions of imaginary creatures or entities commonly associated with fiction or mythology

references to monsters

No Comments

No Known Activations