INDEX

Explanations

phrases related to different outcomes or results

oai_token-act-pair · gpt-3.5-turbo

constructions that describe outcomes or results, particularly those involving the phrases "end up" or "wound up."

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

arers

-0.70

 initiative

-0.66

CW

-0.64

progress

-0.64

 sprang

-0.62

 shine

-0.62

ector

-0.61

 shone

-0.61

hid

-0.61

 pertaining

-0.60

POSITIVE LOGITS

glomer

0.74

aucus

0.74

 Schwar

0.73

untled

0.72

antically

0.69

ipel

0.67

 TAMADRA

0.66

 Penguin

0.66

skirts

0.65

otom

0.64

Activations Density 0.115%

phrases related to different outcomes or results

constructions that describe outcomes or results, particularly those involving the phrases "end up" or "wound up."

No Comments

No Known Activations