INDEX

Explanations

words related to authors, editors, and organizers

oai_token-act-pair · gpt-3.5-turbo

references to authors and editors in academic or analytical contexts

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

SPONSORED

-0.80

tains

-0.68

 visitation

-0.65

 Corps

-0.59

CLASSIFIED

-0.58

rises

-0.58

Availability

-0.58

 certs

-0.57

Lear

-0.57

 heals

-0.56

POSITIVE LOGITS

hip

1.27

pring

1.10

mith

1.07

hift

1.04

uggest

1.01

paces

1.00

 intend

0.95

pace

0.92

hips

0.91

']

0.91

Activations Density 0.192%

words related to authors, editors, and organizers

references to authors and editors in academic or analytical contexts

No Comments

No Known Activations