INDEX

Explanations

references to strokes, specifically related to handwriting or physical movements

oai_token-act-pair · gpt-3.5-turbo

references to "strokes," particularly in medical or procedural contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 crate

-0.72

ctuary

-0.65

 shelter

-0.65

 bins

-0.64

 Oversight

-0.63

 Official

-0.63

 Free

-0.62

Ara

-0.62

 packages

-0.62

 Fest

-0.61

POSITIVE LOGITS

stroke

4.59

stro

3.03

 stroke

2.81

 strokes

2.52

 Stro

1.74

cyl

1.06

 stro

0.97

blow

0.95

bledon

0.93

strike

0.92

Activations Density 0.027%

references to strokes, specifically related to handwriting or physical movements

references to "strokes," particularly in medical or procedural contexts

No Comments

No Known Activations