INDEX

Explanations

names or terms ending in 'rn1'

oai_token-act-pair · gpt-3.5-turbo

calculations or references related to square roots

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

 superv

-0.80

 Keys

-0.65

 armed

-0.65

 cosmetic

-0.64

 superheroes

-0.63

 compensated

-0.61

 marketing

-0.61

 counselors

-0.61

 checkout

-0.60

 counseling

-0.60

POSITIVE LOGITS

rt

4.25

RT

1.95

rg

1.66

rn

1.56

rf

1.51

rm

1.50

rl

1.41

rs

1.38

rc

1.35

lt

1.34

Activations Density 0.014%

names or terms ending in 'rn1'

calculations or references related to square roots

No Comments

No Known Activations