INDEX

Explanations

words related to pharmaceuticals or drug-like substances

oai_token-act-pair · gpt-3.5-turbo

references to pharmaceutical products, specifically pills

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

 pillar

-0.91

 pillars

-0.84

 Guth

-0.75

 Rush

-0.62

 Heisman

-0.61

 Charge

-0.60

 petrol

-0.59

kW

-0.58

 metab

-0.57

 taxed

-0.57

POSITIVE LOGITS

pillar

2.73

pill

2.24

hook

1.10

pace

0.81

gypt

0.80

cific

0.77

hops

0.74

aign

0.74

vised

0.73

furt

0.69

Activations Density 0.003%

words related to pharmaceuticals or drug-like substances

references to pharmaceutical products, specifically pills

No Comments

No Known Activations