INDEX

Explanations

phrases related to a downward motion or negative event

oai_token-act-pair · gpt-3.5-turbo

phrases that include the word "down."

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

kefeller

-0.65

aceous

-0.63

archives

-0.59

itia

-0.57

ument

-0.56

¶ħ

-0.56

andum

-0.55

okemon

-0.55

ĸļ

-0.55

ugh

-0.54

POSITIVE LOGITS

grading

1.21

graded

1.21

stairs

1.10

 stairs

1.08

LOAD

1.07

grades

1.05

hill

0.96

river

0.91

wind

0.81

sized

0.81

Activations Density 0.065%

phrases related to a downward motion or negative event

phrases that include the word "down."

No Comments

No Known Activations