INDEX

Explanations

proper names and related terms

oai_token-act-pair · gpt-3.5-turbo

elements of names or phrases associated with specific individuals or entities

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

 GOODMAN

-0.81

 whales

-0.68

 Nile

-0.65

GGGG

-0.62

 trainers

-0.62

 underwater

-0.60

 disapp

-0.58

 faint

-0.57

 Predator

-0.56

 caucuses

-0.56

POSITIVE LOGITS

uble

0.90

schild

0.90

enegger

0.85

kov

0.82

ramer

0.80

vere

0.79

inger

0.79

berger

0.77

lich

0.77

feld

0.76

Activations Density 0.061%

proper names and related terms

elements of names or phrases associated with specific individuals or entities

No Comments

No Known Activations