INDEX

Explanations

terms related to preconceived notions or beliefs

oai_token-act-pair · gpt-3.5-turbo

concepts related to perceptions and misconceptions

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

 royalty

-0.68

 scrub

-0.67

 braces

-0.67

 ubiqu

-0.66

 kosher

-0.62

scr

-0.61

 gelatin

-0.61

 counters

-0.61

sol

-0.60

 removal

-0.60

POSITIVE LOGITS

ceived

4.09

ceive

2.75

ceptions

2.14

ception

1.87

cept

1.46

structed

1.41

ceptive

1.39

ceivable

1.38

ceiver

1.25

CEPT

1.23

Activations Density 0.023%

terms related to preconceived notions or beliefs

concepts related to perceptions and misconceptions

No Comments

No Known Activations