INDEX

Explanations

Mentions of specific titles or names, likely of individuals

oai_token-act-pair · gpt-3.5-turbo

references to individuals, particularly titles followed by their names

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

 Kodi

-0.68

reach

-0.62

versions

-0.62

CTV

-0.61

izzard

-0.58

 grading

-0.58

CFL

-0.58

 fruitful

-0.57

igham

-0.57

 Olympia

-0.56

POSITIVE LOGITS

udder

0.73

ashtra

0.71

srf

0.70

ĪĴ

0.66

 guiActive

0.64

apest

0.63

Ö¼

0.62

essor

0.61

rupted

0.61

itched

0.60

Activations Density 0.215%

Mentions of specific titles or names, likely of individuals

references to individuals, particularly titles followed by their names

No Comments

No Known Activations