INDEX

Explanations

references to legal or political conflicts

oai_token-act-pair · gpt-3.5-turbo

conjunctions, particularly focusing on the word "and" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

onica

-0.73

othal

-0.72

BOOK

-0.72

Submit

-0.71

llah

-0.66

DX

-0.66

OTA

-0.66

ubi

-0.66

maxwell

-0.66

meric

-0.64

POSITIVE LOGITS

 boasting

1.04

 hence

0.97

 thus

0.93

 thereby

0.91

 nicknamed

0.91

 costing

0.89

 flanked

0.89

 featuring

0.89

 comprising

0.89

 raining

0.88

Activations Density 0.369%

references to legal or political conflicts

conjunctions, particularly focusing on the word "and" in various contexts

No Comments

No Known Activations