INDEX

Explanations

emphatic expressions like "What a" followed by an adjective or noun

oai_token-act-pair · gpt-3.5-turbo

expressions of admiration or emphasis

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

:[

-0.64

=-=-=-=-

-0.63

Reports

-0.62

[/

-0.62

ATURES

-0.62

aneously

-0.61

=[

-0.60

pts

-0.60

 Lans

-0.59

Ont

-0.59

POSITIVE LOGITS

lot

0.98

 heck

0.90

ils

0.88

sexual

0.85

 bunch

0.83

 person

0.83

historic

0.82

sembly

0.82

 mistake

0.80

 couple

0.78

Activations Density 0.144%

emphatic expressions like "What a" followed by an adjective or noun

expressions of admiration or emphasis

No Comments

No Known Activations