INDEX

Explanations

specific subjects or titles within a larger context

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "Subject" and related terms

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

gas

-0.71

 brace

-0.66

ice

-0.66

 cycling

-0.66

 animation

-0.65

 poke

-0.64

 bulldo

-0.64

otte

-0.64

 advocate

-0.64

 palm

-0.63

POSITIVE LOGITS

Subject

4.24

 Subject

2.82

subject

2.08

 Subjects

1.55

Reviewer

1.46

soever

1.33

Posts

1.24

Reply

1.19

Anonymous

1.14

Merit

1.07

Activations Density 0.037%

specific subjects or titles within a larger context

occurrences of the word "Subject" and related terms

No Comments

No Known Activations