INDEX

Explanations

comparisons between different quantities or levels

oai_token-act-pair · gpt-3.5-turbo

a pattern of comparison or increase related to various subjects

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

 Procedure

-0.73

xtap

-0.72

 Reconstruction

-0.67

DragonMagazine

-0.67

 Divinity

-0.66

 Passage

-0.64

fuck

-0.64

=-=-

-0.63

ress

-0.63

POSITIVE LOGITS

 than

1.26

 expensive

1.17

 stringent

1.10

 costly

1.08

 efficient

1.05

 sophisticated

1.04

 importantly

1.04

 favourable

0.98

 complicated

0.97

 manageable

0.95

Activations Density 0.151%

comparisons between different quantities or levels

a pattern of comparison or increase related to various subjects

No Comments

No Known Activations