INDEX

Explanations

mentions of television channels or broadcast schedules

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "on"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

Ct

-0.63

 Investig

-0.62

 Jericho

-0.61

 Virus

-0.61

 intervening

-0.60

 psychotic

-0.59

 Canaan

-0.58

CVE

-0.58

 Consumer

-0.57

 rushing

-0.57

POSITIVE LOGITS

ulum

0.82

orrow

0.82

ango

0.79

loo

0.78

ĸļ

0.78

hent

0.77

ulo

0.77

cup

0.76

livion

0.75

*=-

0.74

Activations Density 0.063%

mentions of television channels or broadcast schedules

occurrences of the word "on"

No Comments

No Known Activations