INDEX

Explanations

unique or specific titles or labels

oai_token-act-pair · gpt-3.5-turbo

commands or directives in the text

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

forth

-0.60

pheus

-0.60

 fellowship

-0.57

è£

-0.55

ledged

-0.55

 symp

-0.55

xxxx

-0.54

 facilitated

-0.54

ilaterally

-0.53

fully

-0.53

POSITIVE LOGITS

zbollah

0.79

isoft

0.73

oÄŁan

0.70

 Expand

0.70

initions

0.67

 Ingredients

0.67

 SHARES

0.65

Warning

0.64

Wiki

0.63

 Definition

0.63

Activations Density 0.250%

unique or specific titles or labels

commands or directives in the text

No Comments

No Known Activations