INDEX

Explanations

specific mentions of the word "icon"

oai_token-act-pair · gpt-3.5-turbo

references to various types of icons and iconography

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Hegel

-0.68

UGE

-0.60

ndra

-0.60

 Analy

-0.59

 incon

-0.59

 Thoughts

-0.58

 Intermediate

-0.58

hari

-0.57

nda

-0.57

 Winds

-0.56

POSITIVE LOGITS

ocl

1.61

ically

1.16

ography

1.12

nect

1.11

icity

1.10

icons

1.09

icon

0.99

ographic

0.99

ologies

0.98

oc

0.95

Activations Density 0.043%

specific mentions of the word "icon"

references to various types of icons and iconography

No Comments

No Known Activations