INDEX

Explanations

specific paired characters that signify a specific programming construct or syntax

oai_token-act-pair · gpt-3.5-turbo

constructs related to complexity or complexity-related concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

hement

-0.86

 Mechdragon

-0.68

uto

-0.64

erville

-0.63

hao

-0.62

Lug

-0.61

 Cato

-0.60

 Ripple

-0.60

 Shen

-0.60

 Cent

-0.60

POSITIVE LOGITS

ÂłÂłÂłÂłÂłÂłÂłÂł

0.87

ÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂł

0.87

shall

0.74

ÂłÂłÂłÂł

0.74

need

0.73

prototype

0.72

ever

0.71

edition

0.70

voy

0.70

 recess

0.69

Activations Density 0.117%

specific paired characters that signify a specific programming construct or syntax

constructs related to complexity or complexity-related concepts

No Comments

No Known Activations