INDEX

Explanations

acronyms and code-like patterns

oai_token-act-pair · gpt-3.5-turbo

references to television or media entities and their associated programming

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

 Feast

-0.78

sow

-0.67

 latch

-0.65

 impulse

-0.64

zag

-0.63

isite

-0.62

hunt

-0.61

nings

-0.61

zees

-0.61

 caution

-0.59

POSITIVE LOGITS

ulhu

0.94

ãĥ¼ãĥĨãĤ£

0.87

MX

0.77

emporary

0.77

å°Ĩ

0.74

0.71

OA

0.69

section

0.67

nect

0.67

ottage

0.66

Activations Density 0.104%

acronyms and code-like patterns

references to television or media entities and their associated programming

No Comments

No Known Activations