INDEX

Explanations

dialogue cues in the form of markup symbols and character conversations

oai_token-act-pair · gpt-3.5-turbo

fragmented or unusual characters and informal speech patterns

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

mercial

-0.50

è£ħ

-0.47

uckland

-0.47

oun

-0.46

nesday

-0.44

cephal

-0.43

avorite

-0.42

 tremend

-0.42

 agric

-0.42

illac

-0.42

POSITIVE LOGITS

Ķ

0.62

ĺ

0.59

Ļ

0.57

¦

0.57

cause

0.56

Ĵ

0.55

ł

0.55

ĵ

0.55

¥

0.54

Ŀ

0.53

Activations Density 0.370%

dialogue cues in the form of markup symbols and character conversations

fragmented or unusual characters and informal speech patterns

No Comments

No Known Activations