INDEX

Explanations

specific technical terms, possibly related to technology or procedures

oai_token-act-pair · gpt-3.5-turbo

symbols, punctuation, or special characters indicating excitement or emphasis

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 inflamm

-0.65

 hindsight

-0.59

 conclud

-0.59

 shenan

-0.58

 gestation

-0.57

 notor

-0.57

 ammunition

-0.56

 necessities

-0.56

 comr

-0.55

 possessions

-0.55

POSITIVE LOGITS

hua

0.76

pora

0.71

jong

0.69

hu

0.69

arth

0.68

 Sapphire

0.67

andra

0.66

tics

0.65

mia

0.65

osp

0.65

Activations Density 0.566%

specific technical terms, possibly related to technology or procedures

symbols, punctuation, or special characters indicating excitement or emphasis

No Comments

No Known Activations