INDEX

Explanations

actions or processes performed by objects or systems

oai_token-act-pair · gpt-3.5-turbo

instances of the word "act" in various contexts

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

rolet

-0.76

lege

-0.72

rib

-0.68

usra

-0.67

oller

-0.65

rax

-0.62

ashtra

-0.61

eeee

-0.60

ron

-0.60

 Regions

-0.60

POSITIVE LOGITS

 opposed

1.01

pired

0.99

ynchron

0.98

criptions

0.90

 well

0.89

bestos

0.87

pires

0.86

 follows

0.84

pers

0.84

 advertised

0.80

Activations Density 0.113%

actions or processes performed by objects or systems

instances of the word "act" in various contexts

No Comments

No Known Activations