INDEX

Explanations

phrases related to political or societal context

oai_token-act-pair · gpt-3.5-turbo

phrases indicating familiarity or awareness of a topic

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 disbanded

-0.69

andowski

-0.68

 redistributed

-0.64

lam

-0.62

 subcontract

-0.61

roup

-0.61

 teamed

-0.60

 vetoed

-0.60

 promot

-0.60

 merge

-0.58

POSITIVE LOGITS

 comprehension

1.23

 viewing

1.19

 reading

1.16

Reading

1.11

 glance

1.11

 inspection

1.04

 unfamiliar

1.02

 watching

1.01

 glimpse

1.00

 understanding

1.00

Activations Density 0.495%

phrases related to political or societal context

phrases indicating familiarity or awareness of a topic

No Comments

No Known Activations