INDEX

Explanations

names or terms related to medical or health issues

oai_token-act-pair · gpt-3.5-turbo

specific names and terms associated with individuals or concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 Tatt

-0.69

 Atom

-0.66

 aimed

-0.64

Tou

-0.63

view

-0.62

 seats

-0.61

 Hare

-0.61

raq

-0.60

anche

-0.59

 Drag

-0.59

POSITIVE LOGITS

igl

2.58

peace

1.74

sov

1.12

pez

1.09

vic

1.05

cox

1.03

vier

1.02

 choke

1.01

worldly

1.00

oliath

0.97

Activations Density 0.039%

names or terms related to medical or health issues

specific names and terms associated with individuals or concepts

No Comments

No Known Activations