How to use CalebMaresca/matrix-game-embeddings-ft-v1 with sentence-transformers:
from sentence_transformers import SentenceTransformer model = SentenceTransformer("CalebMaresca/matrix-game-embeddings-ft-v1") sentences = [ " What is the proposed alternative to imposing random events purely by chance on players?", "half a day). \n• They are perceived to be new and innovative (despite being around since 1987). \n• They are easy to transport, requiring only pen and paper – with perhaps a few maps and \ncounters. \n• They work well in multi-domain, multi-agency contexts allowing all Actors to participate \nequally. \nA few Words of Warning \n• The fact that a Matrix Game requires little infrastructure can be a problem – it just \ndoesn't look sexy and the strengths that it can be done quickly with the minimum of \nfuss, can be reduced by efforts to make it look cool/expensive. \n• The non-quantitative nature of the game can frustrate analysts. \n• Matrix Games require an experience facilitator to run them.", "inform the other players of their stated intentions. In many cases these are not really \n\"arguments\" as part of the game, so shouldn't count as their action for the turn, unless they \nwish to specify a measurable effect (such as increasing their approval ratings). \nTrade Agreements \nIn some games, trade forms a very important part of the game narrative. In most cases this \ncan be treated simply as part of the normal ebb and flow of the argument process. \nHowever, in some circumstances, particularly when timescales are long, trade can require \ngreater attention as to the nuances of the economic benefits and impacts. In these cases, it \nmay be necessary to get the two sides to make additional arguments as to what they expect", "possible throughout the game, having “random events” happen completely at random is \nproblematic. An Actor may be disadvantaged purely by chance, more than once during the \ngame, which can reduce their immersion and engagement. The narrative develops during \nthe game based on the decisions of the players and their reactions to the decisions of other \nplayers. Having random events imposed on them by chance breaks this “cause and effect” \ncycle and degrades the game flow. \nThe alternative is to give the random event to the participants. They will then make a \ndecision as to how this can contribute to the narrative being developed by the players. They" ] embeddings = model.encode(sentences) similarities = model.similarity(embeddings, embeddings) print(similarities.shape) # [4, 4]
