INDEX

Explanations

words related to technical specifications or attributes, such as product features or ingredients

oai_token-act-pair · gpt-3.5-turbo

references to sexually explicit content

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

Reviewer

-1.04

Downloadha

-0.89

TPPStreamerBot

-0.78

GROUND

-0.75

phe

-0.75

inventoryQuantity

-0.74

UGH

-0.73

assetsadobe

-0.73

HAHAHAHA

-0.70

 behavi

-0.70

POSITIVE LOGITS

ensions

1.30

reme

1.28

ract

1.23

racted

1.19

ension

1.17

uple

1.09

ractor

1.09

raction

1.05

ended

1.03

ortion

0.96

Activations Density 0.012%

words related to technical specifications or attributes, such as product features or ingredients

references to sexually explicit content

No Comments

No Known Activations