INDEX

Explanations

references to specific text or sources

oai_token-act-pair · gpt-3.5-turbo

references denoted by the term "ref" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Kinnikuman

-0.76

 deed

-0.68

 outfield

-0.68

daq

-0.67

 Outer

-0.66

DAQ

-0.66

 heights

-0.66

 Pharaoh

-0.66

ECD

-0.65

 mighty

-0.65

POSITIVE LOGITS

erences

1.24

inement

1.21

inished

1.05

eree

1.04

ractive

0.98

actor

0.98

utable

0.96

eren

0.94

erential

0.93

resh

0.92

Activations Density 0.006%

references to specific text or sources

references denoted by the term "ref" in various contexts

No Comments

No Known Activations