INDEX

Explanations

words related to actions or processes that can be done correctly or appropriately

oai_token-act-pair · gpt-3.5-turbo

actions related to implementation, usage, and application of various concepts

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

paio

-0.67

 Vaughan

-0.65

ibur

-0.62

ajo

-0.61

ahoo

-0.60

 Baal

-0.59

govtrack

-0.59

 Flavoring

-0.58

Vie

-0.58

whatever

-0.57

POSITIVE LOGITS

 properly

1.07

 correctly

0.97

 individually

0.79

 incorrectly

0.75

 appropriately

0.70

 exceeds

0.69

 improperly

0.69

 unchecked

0.68

ividually

0.67

 happens

0.66

Activations Density 0.114%

words related to actions or processes that can be done correctly or appropriately

actions related to implementation, usage, and application of various concepts

No Comments

No Known Activations