From 4e848124764d13e7febd923c20b7d9115ed5d2a1 Mon Sep 17 00:00:00 2001 From: Andrew Lapp Date: Thu, 16 May 2024 14:16:50 -0500 Subject: [PATCH] fix tests --- tests/generate/test_integration_llamacpp.py | 20 +++++++++++++------- 1 file changed, 13 insertions(+), 7 deletions(-) diff --git a/tests/generate/test_integration_llamacpp.py b/tests/generate/test_integration_llamacpp.py index e133ee027..43d2299bd 100644 --- a/tests/generate/test_integration_llamacpp.py +++ b/tests/generate/test_integration_llamacpp.py @@ -250,16 +250,22 @@ def test_llamacpp_cfg(model): @pytest.mark.parametrize( - "repo,model_path", + "repo,model_path,hf_tokenizer_uri", [ - ("Qwen/Qwen1.5-0.5B-Chat-GGUF", "*q2*.gguf"), - ("QuantFactory/Meta-Llama-3-8B-Instruct-GGUF", "*Q2*.gguf"), - ("TheBloke/phi-2-GGUF", "*Q2*.gguf"), - ("QuantFactory/Phi-3-mini-4k-instruct-GGUF-v2", "*Q2*.gguf"), + ("Qwen/Qwen1.5-0.5B-Chat-GGUF", "*q2*.gguf", "Qwen/Qwen1.5-0.5B-Chat"), + ("TheBloke/phi-2-GGUF", "*Q2*.gguf", "microsoft/phi-2"), ], ) -def test_byte_tokenizer_regression(repo, model_path): +def test_byte_tokenizer_regression(repo, model_path, hf_tokenizer_uri): """Reproduce https://github.com/outlines-dev/outlines/issues/820""" - model = models.llamacpp(repo, model_path) + import llama_cpp + + model = models.llamacpp( + repo, + model_path, + tokenizer=llama_cpp.llama_tokenizer.LlamaHFTokenizer.from_pretrained( + hf_tokenizer_uri + ), + ) generator = generate.choice(model, ["skirt", "dress", "pen", "jacket"]) generator("Pick the odd word out: skirt, dress, pen, jacket")