INDEX

Explanations

phrases related to reporting news or events

oai_token-act-pair · gpt-3.5-turbo

phrases related to reporting and media coverage

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ç¥ŀ

-0.81

ophers

-0.76

ãĤ«

-0.74

stones

-0.67

wark

-0.63

 admire

-0.60

ahime

-0.58

Args

-0.58

 metaphor

-0.57

lime

-0.57

POSITIVE LOGITS

 sightings

0.75

 wrongdoing

0.68

 nightly

0.68

è¦ļéĨĴ

0.66

lehem

0.64

 quarterly

0.64

 Fully

0.63

DERR

0.63

 irregularities

0.62

 Bulk

0.61

Activations Density 0.214%

phrases related to reporting news or events

phrases related to reporting and media coverage

No Comments

No Known Activations