INDEX

Explanations

commands or instructions listed in a structured format

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate guidelines or protocols

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

minist

-0.80

unts

-0.77

Newsletter

-0.73

aukee

-0.73

cest

-0.70

obyl

-0.69

Ott

-0.69

izont

-0.68

opter

-0.67

wr

-0.67

POSITIVE LOGITS

:-

0.96

:"

0.93

:{

0.87

>:

0.86

:]

0.85

 follows

0.82

bourg

0.80

0.75

:#

0.70

:(

0.69

Activations Density 0.014%

commands or instructions listed in a structured format

phrases that indicate guidelines or protocols

No Comments

No Known Activations