INDEX

Explanations

words starting with 'r' followed by a number representing some level of intensity or indicating a ranking

oai_token-act-pair · gpt-3.5-turbo

occurrences of the letter "r."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 GOODMAN

-0.74

 EntityItem

-0.72

renheit

-0.70

 Virtue

-0.70

 Spartan

-0.68

lihood

-0.67

terson

-0.66

 Contra

-0.65

 Isles

-0.65

 Transactions

-0.65

POSITIVE LOGITS

ascal

1.12

ipples

1.08

ump

1.08

umbled

1.07

attle

1.06

angers

1.05

ipper

1.04

ipp

1.04

haps

1.03

ussia

1.02

Activations Density 0.016%

words starting with 'r' followed by a number representing some level of intensity or indicating a ranking

occurrences of the letter "r."

No Comments

No Known Activations