INDEX

Explanations

terms related to news articles and headlines

oai_token-act-pair · gpt-3.5-turbo

mentions of Google Maps and related geographical references

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

 nodd

-0.58

Ire

-0.55

":[

-0.50

 shenan

-0.49

ividual

-0.45

nep

-0.45

eworthy

-0.44

 confir

-0.44

 Gree

-0.44

bilt

-0.44

POSITIVE LOGITS

)</

0.59

 Abandon

0.49

*)

0.46

othe

0.45

chwitz

0.44

ategor

0.44

 Strategies

0.44

 Converted

0.44

DragonMagazine

0.43

 FIGHT

0.42

Activations Density 0.888%

terms related to news articles and headlines

mentions of Google Maps and related geographical references

No Comments

No Known Activations