INDEX

Explanations

phrases related to explosives

oai_token-act-pair · gpt-3.5-turbo

references to explosive devices or situations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

cedented

-0.87

SEA

-0.83

wright

-0.81

pai

-0.81

ploma

-0.81

atche

-0.81

haps

-0.80

aird

-0.79

alian

-0.79

igan

-0.75

POSITIVE LOGITS

eru

1.01

 explosive

0.94

 incendiary

0.87

 decomp

0.84

iating

0.80

 deton

0.79

onite

0.76

 bursts

0.75

endiary

0.75

aneously

0.73

Activations Density 0.020%

phrases related to explosives

references to explosive devices or situations

No Comments

No Known Activations