INDEX

Explanations

phrases indicating a spokesperson or representatives

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

ãĤ¦ãĤ¹

-0.80

reci

-0.70

Ranked

-0.67

strip

-0.67

Stud

-0.66

veyard

-0.65

entry

-0.65

rooms

-0.64

CLASS

-0.64

Thumbnail

-0.64

POSITIVE LOGITS

Gov

0.93

 Attorney

0.87

 Senator

0.79

 spokesperson

0.78

 Amnesty

0.77

 Concern

0.77

Ms

0.76

 Mayor

0.76

 George

0.75

 Agriculture

0.75

Activations Density 0.031%

phrases indicating a spokesperson or representatives

No Comments

No Known Activations