INDEX

Explanations

architecture-related terms or references

oai_token-act-pair · gpt-3.5-turbo

references to architects and architecture-related terms

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

mma

-0.78

othal

-0.72

FEC

-0.72

nir

-0.70

TPS

-0.70

ibaba

-0.67

outh

-0.65

awar

-0.64

FFFF

-0.63

 Mississ

-0.63

POSITIVE LOGITS

urally

1.43

itect

1.29

 architect

1.03

ural

1.02

 architects

0.79

 architectural

0.78

scape

0.77

 Architect

0.76

raper

0.75

ured

0.75

Activations Density 0.023%

architecture-related terms or references

references to architects and architecture-related terms

No Comments

No Known Activations