INDEX

Explanations

proper nouns, especially from organizations or institutions

oai_token-act-pair · gpt-3.5-turbo

acronyms and abbreviations related to governmental or organizational entities

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

 Acer

-0.68

 Lizard

-0.64

 Raspberry

-0.63

 Clockwork

-0.63

 Mate

-0.63

 Fury

-0.62

cir

-0.62

 nipples

-0.61

 Known

-0.61

 Years

-0.61

POSITIVE LOGITS

IS

0.99

OPLE

0.95

ITION

0.91

ICA

0.85

ITE

0.84

HS

0.81

ISE

0.79

TS

0.78

TERN

0.77

OPS

0.76

Activations Density 0.190%

proper nouns, especially from organizations or institutions

acronyms and abbreviations related to governmental or organizational entities

No Comments

No Known Activations