INDEX

Explanations

positive news or information

oai_token-act-pair · gpt-3.5-turbo

phrases that express positive sentiments or news

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

ĸļ

-0.79

ascript

-0.75

edIn

-0.75

-0.69

ismo

-0.68

uilding

-0.67

forth

-0.67

TIT

-0.65

ategory

-0.65

eters

-0.65

POSITIVE LOGITS

 news

1.01

 stuff

0.94

 outweigh

0.92

 thing

0.90

ol

0.89

 guys

0.87

 part

0.86

 folks

0.86

 bits

0.84

 deed

0.81

Activations Density 0.046%

positive news or information

phrases that express positive sentiments or news

No Comments

No Known Activations