INDEX

Explanations

phrases that indicate comparison or relationship between multiple entities

oai_token-act-pair · gpt-3.5-turbo

the term "respective" indicating individual relationships or properties associated with multiple subjects

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

trak

-0.70

live

-0.70

aqu

-0.67

Detect

-0.66

cephal

-0.66

chrom

-0.66

walk

-0.65

Rail

-0.65

river

-0.64

ISC

-0.63

POSITIVE LOGITS

 respective

0.92

 proportions

0.78

 names

0.75

 allocation

0.71

«ĺ

0.71

 newcom

0.71

ãĤ¼ãĤ¦ãĤ¹

0.71

 teasp

0.70

ibilities

0.70

 eleph

0.70

Activations Density 0.005%

phrases that indicate comparison or relationship between multiple entities

the term "respective" indicating individual relationships or properties associated with multiple subjects

No Comments

No Known Activations