INDEX

Explanations

terms related to underlying issues or structures

oai_token-act-pair · gpt-3.5-turbo

references to foundational or core issues

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

ooters

-0.82

chin

-0.79

zzi

-0.77

alde

-0.76

oping

-0.75

arcity

-0.75

cture

-0.74

asia

-0.74

efully

-0.73

itto

-0.72

POSITIVE LOGITS

 underlying

0.87

 layers

0.77

SourceFile

0.75

layer

0.75

 underpin

0.74

 assumptions

0.74

 motivations

0.72

Versions

0.71

 fundamentals

0.71

 infrastructure

0.68

Activations Density 0.011%

terms related to underlying issues or structures

references to foundational or core issues

No Comments

No Known Activations