INDEX

Explanations

references to physical fire or combustion

oai_token-act-pair · gpt-3.5-turbo

references to flames or fire-related imagery

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 6-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.6.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.6.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

alian

-0.87

nai

-0.76

alth

-0.73

Lean

-0.68

istan

-0.66

guyen

-0.66

Govern

-0.65

llah

-0.65

laus

-0.65

 Geographic

-0.65

POSITIVE LOGITS

 flame

0.98

 extinguished

0.98

 flames

0.96

flies

0.95

 hotter

0.95

 exting

0.88

orescence

0.88

 furnace

0.87

 blaze

0.87

 extingu

0.86

Activations Density 0.043%

references to physical fire or combustion

references to flames or fire-related imagery

No Comments

No Known Activations