INDEX

Explanations

phrases mentioning something being at the 'top'

oai_token-act-pair · gpt-3.5-turbo

references to ranking or status within categories

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

ufact

-0.67

 Advent

-0.65

 Manual

-0.63

 Gaul

-0.63

 Territories

-0.61

otive

-0.61

riages

-0.61

 Arri

-0.60

 Instrument

-0.58

 Neon

-0.58

POSITIVE LOGITS

most

1.07

 tier

1.00

ographical

0.98

top

0.89

 notch

0.88

 scorer

0.85

deck

0.85

top

0.85

ographically

0.82

ography

0.80

Activations Density 0.039%

phrases mentioning something being at the 'top'

references to ranking or status within categories

No Comments

No Known Activations