INDEX

Explanations

phrases related to important or impactful actions or events

oai_token-act-pair · gpt-3.5-turbo

words or phrases that include a specific character sequence or formatting

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Downs

-0.77

 filler

-0.69

 strat

-0.68

 recycling

-0.68

 surrender

-0.67

 nutrient

-0.66

 rubbish

-0.63

 flowering

-0.63

 bluff

-0.63

 prostitute

-0.63

POSITIVE LOGITS

ï¸ı

1.21

Â£

1.00

should

1.00

Ö¼

0.95

âĶĢâĶĢâĶĢâĶĢ

0.93

owned

0.92

âĸł

0.91

âĻ

0.90

Â¯Â¯

0.90

Pg

0.89

Activations Density 0.220%

phrases related to important or impactful actions or events

words or phrases that include a specific character sequence or formatting

No Comments

No Known Activations