INDEX

Explanations

names and terms related to specific individuals in different contexts

oai_token-act-pair · gpt-3.5-turbo

the word "as" in various contexts

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

OUGH

-0.62

 Squid

-0.60

Submit

-0.60

 thumbnail

-0.59

ship

-0.57

ships

-0.56

OWS

-0.56

 contiguous

-0.55

 Sting

-0.55

 better

-0.55

POSITIVE LOGITS

aurus

1.30

daq

1.22

cius

1.18

semble

1.17

pora

1.16

utra

1.14

adena

1.11

henko

1.11

lav

1.10

quez

1.10

Activations Density 0.056%

names and terms related to specific individuals in different contexts

the word "as" in various contexts

No Comments

No Known Activations