INDEX

Explanations

terms related to American football defensive players

oai_token-act-pair · gpt-3.5-turbo

references to undrafted free agents in a sports context

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

enegger

-0.70

 Feminist

-0.67

elvet

-0.64

rage

-0.61

arge

-0.60

 Conflict

-0.59

netflix

-0.58

gently

-0.57

arten

-0.56

 labelled

-0.55

POSITIVE LOGITS

 rookie

0.83

 cornerback

0.83

 depth

0.78

 linebacker

0.78

 linebackers

0.77

 reperto

0.77

 Bengals

0.76

 quarterback

0.75

 scouts

0.74

 rusher

0.74

Activations Density 1.473%

terms related to American football defensive players

references to undrafted free agents in a sports context

No Comments

No Known Activations