INDEX

Explanations

number patterns as part of words

oai_token-act-pair · gpt-3.5-turbo

numerical values or identifiers used in a hierarchical or structured format

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

netflix

-0.68

 Gunn

-0.68

Osw

-0.66

 Whedon

-0.64

worthiness

-0.64

verified

-0.63

icable

-0.63

 endors

-0.62

consumer

-0.62

 Consumer

-0.61

POSITIVE LOGITS

 formation

1.19

 formations

0.94

 hybrid

0.90

 combinations

0.88

ramid

0.87

 Formation

0.84

 triangle

0.83

 rotation

0.79

agonal

0.79

 grid

0.79

Activations Density 0.073%

number patterns as part of words

numerical values or identifiers used in a hierarchical or structured format

No Comments

No Known Activations