INDEX

Explanations

words related to specific themed items, places, events, and concepts

oai_token-act-pair · gpt-3.5-turbo

terms that indicate specific themes or motifs in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

IVER

-0.78

OIL

-0.71

 ILCS

-0.69

eded

-0.65

ãģį

-0.64

oil

-0.63

cular

-0.63

å¥³

-0.62

gans

-0.61

Ability

-0.61

POSITIVE LOGITS

 themed

1.07

nesday

0.95

themed

0.86

ragon

0.86

theme

0.80

ulously

0.78

 costumes

0.72

 festivities

0.72

 artwork

0.72

 decoration

0.72

Activations Density 0.023%

words related to specific themed items, places, events, and concepts

terms that indicate specific themes or motifs in various contexts

No Comments

No Known Activations