INDEX

Explanations

words closely associated with names or entities, potentially focusing on specific individuals

oai_token-act-pair · gpt-3.5-turbo

vowels and phonetic patterns in words

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

 coord

-0.72

 alloc

-0.68

 consolidation

-0.67

ATTLE

-0.67

 withd

-0.65

eele

-0.65

romeda

-0.61

è¦ļéĨĴ

-0.61

 manufact

-0.61

 Chimera

-0.59

POSITIVE LOGITS

ghan

1.42

hei

1.12

lder

1.03

ÃŁ

1.02

hee

1.00

wei

0.97

ften

0.95

ze

0.95

Mei

0.94

vern

0.93

Activations Density 0.168%

words closely associated with names or entities, potentially focusing on specific individuals

vowels and phonetic patterns in words

No Comments

No Known Activations