INDEX

Explanations

commands or instructions that involve telling someone to do something

oai_token-act-pair · gpt-3.5-turbo

instances of the word "ordered" indicating commands or directives

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

pmwiki

-0.76

cit

-0.76

doi

-0.74

bil

-0.74

td

-0.73

ï¿½ï¿½ï¿½ï¿½

-0.71

ãĤ¨

-0.71

 ÎĶ

-0.70

odor

-0.69

Ott

-0.69

POSITIVE LOGITS

 ordering

0.94

 ordered

0.92

 orders

0.88

etary

0.81

avorite

0.77

 confir

0.77

eering

0.76

lies

0.73

rul

0.73

 psychiat

0.72

Activations Density 0.015%

commands or instructions that involve telling someone to do something

instances of the word "ordered" indicating commands or directives

No Comments

No Known Activations