INDEX

Explanations

the word "sar" or variations involving the stem "sar" typically linked to chemical or toxicological 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_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

stakes

-0.76

riad

-0.76

 Friendly

-0.75

eas

-0.72

ingly

-0.70

enance

-0.69

Trend

-0.67

aeper

-0.66

ORD

-0.66

 Panasonic

-0.64

POSITIVE LOGITS

xual

0.94

ned

0.91

emy

0.89

ovie

0.86

osaurs

0.86

ergic

0.85

ews

0.85

eteen

0.81

uke

0.81

uous

0.80

Activations Density 0.010%

the word "sar" or variations involving the stem "sar" typically linked to chemical or toxicological contexts

No Comments

No Known Activations