INDEX

Explanations

mentions of the word "soft" or "softness."

oai_token-act-pair · gpt-3.5-turbo

references to software-related terms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 anten

-0.64

 Gavin

-0.63

 adul

-0.62

 Pluto

-0.62

aceous

-0.62

 Icelandic

-0.62

Param

-0.62

Gle

-0.61

ãĤ¶

-0.60

 Tasman

-0.60

POSITIVE LOGITS

soft

1.08

sov

0.96

rene

0.86

xus

0.85

tek

0.84

lasses

0.79

weapons

0.78

hes

0.77

lus

0.77

ense

0.77

Activations Density 0.027%

mentions of the word "soft" or "softness."

references to software-related terms

No Comments

No Known Activations