INDEX

Explanations

certain formatting patterns, likely related to citations or lists, rather than specific content words

oai_token-act-pair · gpt-3.5-turbo

punctuation and formatting elements

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

iberal

-0.76

 POLITICO

-0.74

 politic

-0.70

 divest

-0.70

interrupted

-0.69

ecast

-0.69

 nonpartisan

-0.67

 democratically

-0.66

 Soros

-0.66

govtrack

-0.66

POSITIVE LOGITS

 Manufacturer

0.87

 invention

0.81

 Invention

0.80

 Experiment

0.79

 Designs

0.79

 Insect

0.78

Recipe

0.77

 Specifications

0.74

 Instructor

0.71

 seller

0.71

Activations Density 1.348%

certain formatting patterns, likely related to citations or lists, rather than specific content words

punctuation and formatting elements

No Comments

No Known Activations