INDEX

Explanations

questions or statements posing a contrary position

oai_token-act-pair · gpt-3.5-turbo

phrases that express negation or doubt

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

furt

-0.77

ãĤ£

-0.71

Ö¼

-0.68

Fra

-0.68

urst

-0.66

het

-0.63

ELD

-0.62

eteenth

-0.61

vironment

-0.60

facts

-0.60

POSITIVE LOGITS

hin

0.85

?".

0.85

 sooner

0.83

?!"

0.79

?",

0.79

!?"

0.76

?]

0.76

?"

0.75

?).

0.75

?),

0.74

Activations Density 0.098%

questions or statements posing a contrary position

phrases that express negation or doubt

No Comments

No Known Activations