INDEX

Explanations

words indicating a successful or temporary action

oai_token-act-pair · gpt-3.5-turbo

words indicating successful or forced actions

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

soc

-0.81

icip

-0.76

ERAL

-0.73

iquette

-0.70

Liter

-0.70

 intentions

-0.66

 enthus

-0.65

ilion

-0.65

 tendency

-0.65

quickShipAvailable

-0.65

POSITIVE LOGITS

 transported

0.96

 awoken

0.95

 transitioned

0.94

 reinvent

0.92

 housed

0.90

 reproduce

0.90

 detonated

0.89

 lobbied

0.88

 evacuated

0.88

 reproduced

0.86

Activations Density 0.147%

words indicating a successful or temporary action

words indicating successful or forced actions

No Comments

No Known Activations