INDEX

Explanations

words related to politics, control, and authority

oai_token-act-pair · gpt-3.5-turbo

concepts related to political and social dynamics

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

ymm

-0.56

arnaev

-0.55

xon

-0.53

atom

-0.52

ULTS

-0.52

onz

-0.52

qua

-0.51

ãĥ£

-0.51

cci

-0.51

ortium

-0.51

POSITIVE LOGITS

 itself

0.67

 herself

0.60

 himself

0.55

 yourself

0.54

 peripher

0.54

POV

0.54

 altogether

0.53

 motif

0.52

pedia

0.52

 entails

0.51

Activations Density 1.094%

words related to politics, control, and authority

concepts related to political and social dynamics

No Comments

No Known Activations