INDEX

Explanations

verbs or phrases related to actions done to prevent or stop something negative from happening

oai_token-act-pair · gpt-3.5-turbo

phrases related to stopping or halting negative processes or events

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ledged

-0.92

çīĪ

-0.81

served

-0.77

FACE

-0.76

eland

-0.70

erved

-0.68

warts

-0.67

leground

-0.67

ell

-0.65

=-=-

-0.64

POSITIVE LOGITS

 flow

1.31

 bleeding

1.26

 progress

1.24

 progression

1.16

 advance

1.12

 tide

1.11

 momentum

1.11

 growth

1.07

 madness

1.07

 onslaught

1.05

Activations Density 0.246%

verbs or phrases related to actions done to prevent or stop something negative from happening

phrases related to stopping or halting negative processes or events

No Comments

No Known Activations