INDEX

Explanations

terms related to intensity or urgency, particularly with negative connotations

oai_token-act-pair · gpt-3.5-turbo

phrases indicating tension or suspense

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

heit

-0.75

STATE

-0.69

iliary

-0.64

esson

-0.63

States

-0.63

Ô

-0.63

 hereafter

-0.63

esses

-0.62

FML

-0.62

vernment

-0.61

POSITIVE LOGITS

monds

0.70

picked

0.67

 clay

0.67

 etched

0.67

 stones

0.67

 rods

0.66

sharp

0.65

 paste

0.65

 bushes

0.63

 stone

0.63

Activations Density 0.183%

terms related to intensity or urgency, particularly with negative connotations

phrases indicating tension or suspense

No Comments

No Known Activations