INDEX

Explanations

words related to titles or headings, specifically recognizing and focusing on specific symbols

oai_token-act-pair · gpt-3.5-turbo

a specific character or symbol, particularly focused on the letter "L"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 mathemat

-0.82

 obser

-0.81

 contrace

-0.79

 confir

-0.78

 incorpor

-0.76

 satell

-0.74

sonian

-0.74

 undermin

-0.73

 federation

-0.71

 strugg

-0.71

POSITIVE LOGITS

ï¸

1.01

ï¸ı

0.95

Â¯

0.93

ttp

0.92

#$

0.91

0.82

âĻ

0.80

âĻ¥

0.80

âĶĢâĶĢâĶĢâĶĢ

0.78

 Originally

0.77

Activations Density 0.214%

words related to titles or headings, specifically recognizing and focusing on specific symbols

a specific character or symbol, particularly focused on the letter "L"

No Comments

No Known Activations