INDEX

Explanations

terms related to formal investigations or probes

oai_token-act-pair · gpt-3.5-turbo

mentions of investigations or probes

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

birth

-0.71

 Distribut

-0.67

icio

-0.65

veland

-0.65

Unix

-0.64

ãĤª

-0.64

emonic

-0.60

umeric

-0.60

ulic

-0.59

eton

-0.59

POSITIVE LOGITS

 into

1.41

 INTO

1.24

into

1.18

 Into

1.13

 probing

1.03

 probe

1.02

 undertaken

0.94

 underway

0.93

 probes

0.92

 involving

0.92

Activations Density 0.089%

terms related to formal investigations or probes

mentions of investigations or probes

No Comments

No Known Activations