INDEX

Explanations

words related to names and specific terms, possibly related to political figures or events

oai_token-act-pair · gpt-3.5-turbo

tokens related to names or identifiers, particularly focused on a specific character or entity throughout the text

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

ahime

-0.72

kamp

-0.61

nw

-0.60

 holiest

-0.56

 Franch

-0.56

 tresp

-0.54

jong

-0.54

 developing

-0.51

ung

-0.51

eton

-0.51

POSITIVE LOGITS

dden

0.68

avis

0.67

ody

0.66

Marginal

0.64

Pick

0.62

atche

0.62

icz

0.62

ĵĺ

0.61

iversal

0.60

anyl

0.60

Activations Density 0.532%

words related to names and specific terms, possibly related to political figures or events

tokens related to names or identifiers, particularly focused on a specific character or entity throughout the text

No Comments

No Known Activations