INDEX

Explanations

phrases indicating the completion of an action or event

oai_token-act-pair · gpt-3.5-turbo

phrases expressing degrees or concepts of significance or importance

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Cosponsors

-0.71

grounds

-0.68

 uncond

-0.65

 underestimated

-0.65

imens

-0.64

 ItemLevel

-0.63

 Crate

-0.63

chwitz

-0.62

abilities

-0.61

aucuses

-0.61

POSITIVE LOGITS

 sorts

0.97

 hers

0.78

 fortune

0.73

 humour

0.67

 luck

0.67

 doom

0.66

 Merlin

0.65

 sadness

0.63

 course

0.63

 bitters

0.63

Activations Density 0.236%

phrases indicating the completion of an action or event

phrases expressing degrees or concepts of significance or importance

No Comments

No Known Activations