INDEX

Explanations

phrases related to breaking rules or records

oai_token-act-pair · gpt-3.5-turbo

instances of phrases and terms related to legal violations and legislative actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ADD

-0.70

eton

-0.67

 Reserved

-0.66

EMP

-0.63

anni

-0.63

 Voters

-0.62

rily

-0.62

ogg

-0.60

enhagen

-0.59

 Matters

-0.59

POSITIVE LOGITS

 seal

0.80

 stride

0.79

 barrier

0.79

advertisement

0.76

 membranes

0.74

 protocol

0.74

 bottleneck

0.73

 chain

0.73

 boundaries

0.71

 bounds

0.71

Activations Density 0.136%

phrases related to breaking rules or records

instances of phrases and terms related to legal violations and legislative actions

No Comments

No Known Activations