INDEX

Explanations

verbs where something is achievable or doable

oai_token-act-pair · gpt-3.5-turbo

phrases that suggest possibility or hypothetical actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Culture

-0.68

 costing

-0.65

Lah

-0.63

é¾įå¥ĳå£«

-0.61

 cass

-0.61

quickShipAvailable

-0.61

Iss

-0.60

 Closing

-0.60

written

-0.60

 Amar

-0.60

POSITIVE LOGITS

 visualize

1.01

 conceive

1.00

 quantify

0.98

nikov

0.98

 achieve

0.96

 attain

0.93

 penetrate

0.92

 recreate

0.91

 accomplish

0.91

 foresee

0.90

Activations Density 0.063%

verbs where something is achievable or doable

phrases that suggest possibility or hypothetical actions

No Comments

No Known Activations