INDEX

Explanations

words related to forceful actions or processes, especially ones involving breaking

oai_token-act-pair · gpt-3.5-turbo

phrases related to the action of breaking

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

affer

-0.69

imental

-0.69

erva

-0.67

gel

-0.63

iva

-0.62

ikk

-0.61

oka

-0.61

velength

-0.61

accompanied

-0.60

erala

-0.60

POSITIVE LOGITS

neck

1.07

fast

0.90

breakers

0.87

down

0.86

points

0.85

break

0.80

downs

0.79

breaking

0.77

breaks

0.77

 down

0.76

Activations Density 0.028%

words related to forceful actions or processes, especially ones involving breaking

phrases related to the action of breaking

No Comments

No Known Activations