INDEX

Explanations

mentions of a specific morning news show

oai_token-act-pair · gpt-3.5-turbo

repeated references to the phrase "Morning Joe."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

inates

-0.93

urally

-0.79

relative

-0.74

cles

-0.73

holes

-0.70

imately

-0.70

entially

-0.69

icy

-0.69

RED

-0.68

requ

-0.66

POSITIVE LOGITS

 Morning

0.92

star

0.91

 Consult

0.88

 Caller

0.81

 Madness

0.73

 Jacket

0.73

 Evening

0.73

 Herald

0.72

 Brief

0.72

 Roundup

0.71

Activations Density 0.041%

mentions of a specific morning news show

repeated references to the phrase "Morning Joe."

No Comments

No Known Activations