INDEX

Explanations

specific mentions of a television network

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "NBC"

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

 Gard

-0.74

 guard

-0.72

 Origin

-0.71

 gardens

-0.66

adv

-0.65

 inner

-0.64

Imp

-0.64

 Heaven

-0.64

 guards

-0.64

End

-0.63

POSITIVE LOGITS

NBC

4.11

CBS

1.83

ABC

1.71

FOX

1.47

NBC

1.39

CNN

1.33

CTV

1.24

Netflix

1.24

NPR

1.17

Bloomberg

1.17

Activations Density 0.005%

specific mentions of a television network

occurrences of the word "NBC"

No Comments

No Known Activations