INDEX

Explanations

scientific and pseudo-scientific terms

oai_token-act-pair · gpt-3.5-turbo

terms related to scientific disciplines and methodologies

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

ILA

-0.77

ecause

-0.73

NCT

-0.71

skirts

-0.71

 DRAG

-0.68

mington

-0.68

ABE

-0.66

 Emblem

-0.65

drops

-0.63

boats

-0.63

POSITIVE LOGITS

ific

1.44

cient

1.05

ral

0.89

itude

0.85

ificate

0.84

ocratic

0.81

ology

0.81

rily

0.81

ificent

0.80

arily

0.78

Activations Density 0.023%

scientific and pseudo-scientific terms

terms related to scientific disciplines and methodologies

No Comments

No Known Activations