INDEX

Explanations

phrases related to different or multiple worlds

oai_token-act-pair · gpt-3.5-turbo

references to different fictional worlds or universes

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

ãĥį

-0.71

thumbnails

-0.69

ged

-0.67

onomy

-0.67

rav

-0.65

ging

-0.64

Rush

-0.64

draw

-0.64

ãĤ·ãĥ£

-0.64

-0.63

POSITIVE LOGITS

chool

0.97

 worlds

0.89

 collide

0.88

peak

0.82

pace

0.79

 wide

0.78

hops

0.77

 Worlds

0.76

imilar

0.76

hell

0.74

Activations Density 0.039%

phrases related to different or multiple worlds

references to different fictional worlds or universes

No Comments

No Known Activations