INDEX

Explanations

phrases indicating a judgment or verdict of guilt or wrongdoing

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "found" and variations of it in legal or forensic contexts

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

 junction

-0.67

 imagin

-0.66

 horm

-0.65

userc

-0.63

kin

-0.63

GBT

-0.63

Trade

-0.62

isoft

-0.61

hop

-0.60

 modulation

-0.60

POSITIVE LOGITS

 Cosponsors

0.79

-+-+

0.78

itutional

0.78

alties

0.77

nda

0.73

uments

0.73

baum

0.72

mand

0.69

isi

0.68

roit

0.67

Activations Density 0.069%

phrases indicating a judgment or verdict of guilt or wrongdoing

occurrences of the word "found" and variations of it in legal or forensic contexts

No Comments

No Known Activations