INDEX

Explanations

instances where two specific words are mentioned in the same context, as well as words related to numerical percentages

oai_token-act-pair · gpt-3.5-turbo

phrases related to value and quantity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 exting

-0.65

jah

-0.65

ardless

-0.64

ãĤ¦ãĤ¹

-0.61

farious

-0.60

////////////////

-0.60

orously

-0.60

razil

-0.58

istically

-0.58

anity

-0.58

POSITIVE LOGITS

NBA

0.61

 Knicks

0.61

 Orioles

0.60

WWE

0.60

Â¶

0.58

.)

0.57

 Tsarnaev

0.57

MLB

0.57

 BART

0.56

MMA

0.56

Activations Density 2.214%

instances where two specific words are mentioned in the same context, as well as words related to numerical percentages

phrases related to value and quantity

No Comments

No Known Activations