INDEX

Explanations

statements of what someone thinks or believes

oai_token-act-pair · gpt-3.5-turbo

speculations or hypothetical statements about people's beliefs or actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Orange

-0.64

pires

-0.62

Kag

-0.62

è¦ļéĨĴ

-0.61

SIG

-0.61

{*

-0.60

 Xuan

-0.60

 underscores

-0.60

 Case

-0.60

 Scotia

-0.57

POSITIVE LOGITS

 prefer

1.16

rather

1.15

 gladly

1.14

nt

1.12

 NEVER

1.00

 rather

0.97

've

0.95

 want

0.93

 never

0.93

never

0.92

Activations Density 0.136%

statements of what someone thinks or believes

speculations or hypothetical statements about people's beliefs or actions

No Comments

No Known Activations