INDEX

Explanations

information or details that are worth noting

oai_token-act-pair · gpt-3.5-turbo

expressions emphasizing important or noteworthy information

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

 handlers

-0.64

Cre

-0.61

onz

-0.61

rendered

-0.60

 helicop

-0.59

 handler

-0.57

idal

-0.57

 transmitted

-0.56

CGI

-0.56

 driveway

-0.55

POSITIVE LOGITS

 lest

0.74

inational

0.73

ress

0.67

quished

0.67

antage

0.65

uate

0.64

 parallels

0.63

skirts

0.62

ikes

0.62

tymology

0.61

Activations Density 0.163%

information or details that are worth noting

expressions emphasizing important or noteworthy information

No Comments

No Known Activations