INDEX

Explanations

phrases related to placements or rankings

oai_token-act-pair · gpt-3.5-turbo

references to positions or places within various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 aborted

-0.59

 Breach

-0.57

Wan

-0.57

 confir

-0.56

 whirlwind

-0.56

 decomp

-0.54

 Greenberg

-0.53

Hab

-0.52

 gunshot

-0.52

 Disclosure

-0.52

POSITIVE LOGITS

 atop

1.43

 amongst

1.35

 among

1.27

 amidst

1.02

among

1.00

 alongside

1.00

 beside

0.91

 therein

0.89

inence

0.88

manship

0.87

Activations Density 0.177%

phrases related to placements or rankings

references to positions or places within various contexts

No Comments

No Known Activations