INDEX

Explanations

phrases related to military actions or operations

oai_token-act-pair · gpt-3.5-turbo

references to police actions or operations, specifically raids

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

DonaldTrump

-0.67

chell

-0.62

 Hurricanes

-0.61

Ian

-0.59

Seb

-0.59

Fed

-0.58

 Gork

-0.58

 Warp

-0.58

internet

-0.58

 Quincy

-0.57

POSITIVE LOGITS

ers

0.95

 raid

0.85

 raids

0.83

Raid

0.79

rays

0.79

bing

0.78

arded

0.75

ishly

0.74

iversary

0.73

 raided

0.72

Activations Density 0.037%

phrases related to military actions or operations

references to police actions or operations, specifically raids

No Comments

No Known Activations