INDEX

Explanations

mentions of underground locations or structures

oai_token-act-pair · gpt-3.5-turbo

references to underground environments or concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

cific

-0.88

awed

-0.87

lda

-0.86

ouf

-0.84

eln

-0.81

ogun

-0.81

emaker

-0.81

erva

-0.81

nir

-0.79

nesota

-0.79

POSITIVE LOGITS

 tunnels

0.96

 cavern

0.96

 Bunker

0.94

 bunk

0.93

 bunker

0.93

 workings

0.91

 lair

0.88

 labyrinth

0.84

 plumbing

0.83

 reservoirs

0.83

Activations Density 0.068%

mentions of underground locations or structures

references to underground environments or concepts

No Comments

No Known Activations