INDEX

Explanations

phrases related to direct actions or instructions

oai_token-act-pair · gpt-3.5-turbo

phrases indicating direct actions or connections

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

gerald

-0.80

glers

-0.77

awar

-0.70

ifully

-0.70

 Daily

-0.69

 Garry

-0.67

Wa

-0.65

 Miracle

-0.65

 Baxter

-0.64

 Boys

-0.63

POSITIVE LOGITS

 contradicted

0.78

 contradicts

0.76

 opposite

0.75

forward

0.74

 identifiable

0.73

 direct

0.73

ebted

0.69

 contradict

0.68

 benefited

0.68

 BELOW

0.68

Activations Density 0.015%

phrases related to direct actions or instructions

phrases indicating direct actions or connections

No Comments

No Known Activations