INDEX

Explanations

the word "pack" and related terms

oai_token-act-pair · gpt-3.5-turbo

references to groups or clusters, particularly emphasizing the term "pack."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

perature

-0.78

 ende

-0.75

 Azerb

-0.69

sembly

-0.69

perties

-0.67

itars

-0.66

mit

-0.66

vous

-0.65

ptoms

-0.65

Downloadha

-0.64

POSITIVE LOGITS

 pack

1.25

 packs

1.21

 Packs

1.12

pack

1.10

 Pack

1.04

Pack

1.00

packs

0.98

ers

0.79

 PACK

0.78

ings

0.75

Activations Density 0.007%

the word "pack" and related terms

references to groups or clusters, particularly emphasizing the term "pack."

No Comments

No Known Activations