INDEX

Explanations

patterns and trends within data or physical phenomena

oai_token-act-pair · gpt-3.5-turbo

repeated references to various types of patterns

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

GS

-0.82

LV

-0.80

×Ķ

-0.78

×ª

-0.76

nee

-0.75

×Ļ×

-0.75

grim

-0.74

×¨

-0.74

CLA

-0.73

avez

-0.72

POSITIVE LOGITS

 patterns

1.42

 Patterns

1.29

ĸļ

1.11

 pattern

1.06

Pattern

1.03

pattern

1.02

 Pattern

1.02

 arrang

0.96

ãĤ¼ãĤ¦ãĤ¹

0.90

etting

0.85

Activations Density 0.009%

patterns and trends within data or physical phenomena

repeated references to various types of patterns

No Comments

No Known Activations