INDEX

Explanations

numerical values embedded within a context

oai_token-act-pair · gpt-3.5-turbo

references to cards and ratings related to items or discussions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

cffff

-0.77

 horizont

-0.71

 mathemat

-0.67

Sph

-0.67

 eleph

-0.67

 dwarf

-0.67

↵Âł

-0.66

 indo

-0.64

artifacts

-0.64

 chimpan

-0.63

POSITIVE LOGITS

 href

0.74

abel

0.68

"]=>

0.68

ivation

0.65

Topic

0.64

 Location

0.64

URL

0.64

ource

0.64

 Date

0.63

iculty

0.63

Activations Density 0.225%

numerical values embedded within a context

references to cards and ratings related to items or discussions

No Comments

No Known Activations