INDEX

Explanations

mentions of specific individuals or companies

oai_token-act-pair · gpt-3.5-turbo

phrases related to personal attributes or characteristics of individuals

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

 agre

-0.65

ipel

-0.64

reth

-0.60

:(

-0.57

 Canaan

-0.57

imester

-0.57

 contend

-0.57

 stockpile

-0.56

*:

-0.56

ramid

-0.55

POSITIVE LOGITS

pires

0.73

fol

0.70

azine

0.69

should

0.66

front

0.66

stood

0.62

Mom

0.60

Drive

0.60

Pros

0.59

pite

0.59

Activations Density 0.853%

mentions of specific individuals or companies

phrases related to personal attributes or characteristics of individuals

No Comments

No Known Activations