INDEX

Explanations

keywords related to refining or processing

oai_token-act-pair · gpt-3.5-turbo

references to specific topics or concepts in a text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Invalid

-0.69

 Cout

-0.66

Magikarp

-0.64

CHO

-0.64

hunt

-0.63

 Garfield

-0.62

 Poles

-0.62

channelAvailability

-0.61

ATA

-0.60

CBI

-0.60

POSITIVE LOGITS

actor

1.13

ocused

1.12

raction

1.11

ractive

1.09

ract

1.06

inished

1.06

racted

1.01

utations

0.99

utation

0.98

inement

0.96

Activations Density 0.012%

keywords related to refining or processing

references to specific topics or concepts in a text

No Comments

No Known Activations