INDEX

Explanations

patterns related to numerical ranges

oai_token-act-pair · gpt-3.5-turbo

phrases indicating ranges or quantities

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

 meshes

-0.62

reports

-0.60

ergic

-0.59

ãĤ°

-0.59

æ³

-0.58

ubi

-0.57

ippi

-0.57

çĲ

-0.57

ãĤ±

-0.56

hemat

-0.56

POSITIVE LOGITS

ggles

1.00

pload

0.96

pless

0.95

 insure

0.80

asted

0.79

gas

0.77

 compensate

0.76

 ensure

0.75

lling

0.75

 intensify

0.73

Activations Density 0.092%

patterns related to numerical ranges

phrases indicating ranges or quantities

No Comments

No Known Activations