INDEX

Explanations

words related to discussions or descriptions of conditions or states

oai_token-act-pair · gpt-3.5-turbo

occurrences of the prefix "di-" indicating a relation to diverse topics

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

witz

-0.63

 totality

-0.63

ICA

-0.62

INGTON

-0.61

ï¸ı

-0.60

 spare

-0.60

ULAR

-0.60

INE

-0.60

ICAL

-0.58

 DOWN

-0.57

POSITIVE LOGITS

abol

1.40

urnal

1.35

ocese

1.26

abolic

1.26

agon

1.18

abet

1.12

adem

1.12

plom

1.10

ablo

1.06

alling

1.03

Activations Density 0.019%

words related to discussions or descriptions of conditions or states

occurrences of the prefix "di-" indicating a relation to diverse topics

No Comments

No Known Activations