INDEX

Explanations

specific characters or symbols (e.g., "Ċ") occurring repeatedly in the text

oai_token-act-pair · gpt-3.5-turbo

occurrences of numerical data or references

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

hement

-0.84

 appe

-0.84

Ń·

-0.81

 undermin

-0.77

uca

-0.74

illance

-0.74

 endeav

-0.72

 purs

-0.70

byss

-0.68

 deceived

-0.68

POSITIVE LOGITS

ÂłÂłÂłÂł

0.98

ÂłÂłÂłÂłÂłÂłÂłÂł

0.93

Advertisement

0.92

ÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂł

0.89

SPONSORED

0.87

Meanwhile

0.87

Trivia

0.86

ÂłÂłÂł

0.85

Loading

0.85

Posted

0.83

Activations Density 1.825%

specific characters or symbols (e.g., "Ċ") occurring repeatedly in the text

occurrences of numerical data or references

No Comments

No Known Activations