INDEX

Explanations

Roman numerals

oai_token-act-pair · gpt-3.5-turbo

instances of a specific character or symbol representation

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

 disadvant

-0.76

 misunder

-0.67

 mathemat

-0.64

 Palestin

-0.63

 hemor

-0.63

 contrace

-0.60

 Seym

-0.59

 obser

-0.59

 whiff

-0.58

 fortun

-0.58

POSITIVE LOGITS

ï¸ı

1.34

ï¸

0.93

âĹ

0.82

Balt

0.80

Â¯Â¯

0.78

âĸł

0.75

0.73

âĸ

0.73

âĻ

0.71

âĪ

0.69

Activations Density 0.440%

Roman numerals

instances of a specific character or symbol representation

No Comments

No Known Activations