INDEX

Explanations

mentions of authoritative statements or responses to reports and allegations

oai_token-act-pair · gpt-3.5-turbo

punctuation marks, particularly commas, indicating a focus on lists or separate clauses

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

grain

-0.67

igmatic

-0.67

hedral

-0.65

eco

-0.65

ety

-0.62

alogue

-0.61

merce

-0.61

iple

-0.60

ordinate

-0.60

reprene

-0.59

POSITIVE LOGITS

 citing

1.13

 namely

1.10

 albeit

1.05

 noting

1.02

 although

0.97

 however

0.93

 wherein

0.89

 whereby

0.87

 according

0.86

 implying

0.86

Activations Density 0.854%

mentions of authoritative statements or responses to reports and allegations

punctuation marks, particularly commas, indicating a focus on lists or separate clauses

No Comments

No Known Activations