INDEX

Explanations

phrases related to official investigations or appointments

oai_token-act-pair · gpt-3.5-turbo

occurrences of inspections and evaluations by authorities

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

 Zombies

-0.65

isSpecialOrderable

-0.64

Interstitial

-0.59

 inheritance

-0.58

 Endless

-0.58

ankind

-0.56

Tomorrow

-0.55

 arcade

-0.55

 stagn

-0.55

 Rebellion

-0.54

POSITIVE LOGITS

 overseeing

0.76

 oversee

0.75

 superv

0.75

 discretion

0.70

 inspecting

0.69

Īè

0.68

 oversaw

0.67

 consulted

0.67

 instructed

0.67

ĳå£«

0.66

Activations Density 1.394%

phrases related to official investigations or appointments

occurrences of inspections and evaluations by authorities

No Comments

No Known Activations