INDEX

Explanations

phrases related to legal and political topics, including references to court decisions, legislation, governmental appointments, and public discourse

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

ican

-0.76

nown

-0.68

 Replay

-0.66

iosis

-0.64

 Amend

-0.61

ocol

-0.61

icum

-0.61

etheless

-0.59

gram

-0.58

pher

-0.58

POSITIVE LOGITS

 Cosponsors

0.97

 inspiration

0.94

strings

0.92

 parallels

0.90

 attention

0.85

 drawn

0.82

 aggro

0.81

 conclusions

0.80

draw

0.76

brid

0.76

Activations Density 1.826%

phrases related to legal and political topics, including references to court decisions, legislation, governmental appointments, and public discourse

No Comments

No Known Activations