INDEX

Explanations

scientific terms and journal references

oai_token-act-pair · gpt-3.5-turbo

references to scientific studies and journal publications

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

chwitz

-0.79

 swast

-0.76

tsky

-0.73

ardless

-0.73

 ceilings

-0.72

 candles

-0.72

roo

-0.70

indle

-0.69

 joking

-0.68

 furn

-0.68

POSITIVE LOGITS

Phys

1.05

ournals

1.03

 Perspect

1.02

 Pediatrics

1.01

doi

0.99

 Neuroscience

0.97

 Biol

0.94

 Psychiatry

0.92

 Proceedings

0.92

 PLoS

0.91

Activations Density 0.188%

scientific terms and journal references

references to scientific studies and journal publications

No Comments

No Known Activations