INDEX

Explanations

proper nouns or names, possibly related to authors, characters, or locations

oai_token-act-pair · gpt-3.5-turbo

numerical identifiers and rankings related to items or categories

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

itably

-0.66

 pasture

-0.66

ovember

-0.66

oglu

-0.65

oÄŁ

-0.63

 countryside

-0.62

 gardens

-0.62

isine

-0.61

 bloodstream

-0.60

 gain

-0.58

POSITIVE LOGITS

CB

1.04

JV

0.74

Jackson

0.72

JB

0.71

HM

0.71

JJ

0.71

0.70

orah

0.70

Favorite

0.69

Activations Density 0.050%

proper nouns or names, possibly related to authors, characters, or locations

numerical identifiers and rankings related to items or categories

No Comments

No Known Activations