INDEX

Explanations

specific references to television channels

oai_token-act-pair · gpt-3.5-turbo

references to specific television channels

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

stad

-0.70

hiro

-0.65

 Raiders

-0.64

dden

-0.64

enhagen

-0.63

berman

-0.63

pperc

-0.63

 Hots

-0.63

 Gloria

-0.63

 Linda

-0.62

POSITIVE LOGITS

 channel

0.88

 channels

0.85

lain

0.81

eval

0.81

ages

0.77

channel

0.76

ing

0.76

agall

0.76

Channel

0.76

washer

0.75

Activations Density 0.028%

specific references to television channels

references to specific television channels

No Comments

No Known Activations