INDEX

Explanations

numerical counts or scores

oai_token-act-pair · gpt-3.5-turbo

terms related to counting or summing numerical data

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

 directions

-0.63

rugged

-0.60

ubes

-0.60

ured

-0.57

Prin

-0.57

ented

-0.57

 resistant

-0.57

 conditioned

-0.56

designed

-0.56

 Enhancement

-0.55

POSITIVE LOGITS

 tally

3.78

 tallied

1.94

 totals

1.61

 count

1.55

 reckoning

1.33

 counting

1.32

 ledger

1.31

 recount

1.19

 counted

1.07

 toll

1.06

Activations Density 0.013%

numerical counts or scores

terms related to counting or summing numerical data

No Comments

No Known Activations