INDEX

Explanations

mentions of geographic locations, towns, and people's names

oai_token-act-pair · gpt-3.5-turbo

specific names and notable places

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

 dracon

-0.74

 envy

-0.63

etheless

-0.62

jriwal

-0.62

 trope

-0.58

 fallacy

-0.57

 doom

-0.56

 transitioning

-0.56

 priceless

-0.56

 puzz

-0.55

POSITIVE LOGITS

oz

0.85

ito

0.82

ich

0.79

ak

0.79

ani

0.78

ona

0.78

jan

0.77

oli

0.77

ema

0.77

aj

0.76

Activations Density 0.511%

mentions of geographic locations, towns, and people's names

specific names and notable places

No Comments

No Known Activations