INDEX

Explanations

phrases related to quotations or reported speech

oai_token-act-pair · gpt-3.5-turbo

mentions of a specific character or element, represented by the unique activation pattern

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

 minim

-0.78

OTOS

-0.70

 scatter

-0.68

 coffin

-0.67

 decomp

-0.67

 cyan

-0.67

 coast

-0.66

 fairy

-0.66

 scene

-0.66

 protective

-0.66

POSITIVE LOGITS

cause

0.91

then

0.89

said

0.88

according

0.88

since

0.85

especially

0.84

Â¯

0.83

_>

0.79

sure

0.77

yet

0.77

Activations Density 0.181%

phrases related to quotations or reported speech

mentions of a specific character or element, represented by the unique activation pattern

No Comments

No Known Activations