Merge pull request #159 from VE-FORBRYDERNE/fairseq
Don't blacklist </s> token in "s" newline mode
This commit is contained in:
commit
f127918114
|
@ -1995,7 +1995,7 @@ def load_model(use_gpu=True, gpu_layers=None, disk_layers=None, initial_load=Fal
|
||||||
shutil.rmtree("cache/")
|
shutil.rmtree("cache/")
|
||||||
|
|
||||||
if(vars.badwordsids is vars.badwordsids_default and vars.model_type not in ("gpt2", "gpt_neo", "gptj")):
|
if(vars.badwordsids is vars.badwordsids_default and vars.model_type not in ("gpt2", "gpt_neo", "gptj")):
|
||||||
vars.badwordsids = [[v] for k, v in tokenizer.get_vocab().items() if any(c in str(k) for c in "<>[]")]
|
vars.badwordsids = [[v] for k, v in tokenizer.get_vocab().items() if any(c in str(k) for c in "<>[]") if vars.newlinemode != "s" or str(k) != "</s>"]
|
||||||
|
|
||||||
patch_causallm(model)
|
patch_causallm(model)
|
||||||
|
|
||||||
|
@ -2162,7 +2162,7 @@ def load_model(use_gpu=True, gpu_layers=None, disk_layers=None, initial_load=Fal
|
||||||
vars.modeldim = int(tpu_mtj_backend.params.get("d_embed", tpu_mtj_backend.params["d_model"]))
|
vars.modeldim = int(tpu_mtj_backend.params.get("d_embed", tpu_mtj_backend.params["d_model"]))
|
||||||
tokenizer = tpu_mtj_backend.tokenizer
|
tokenizer = tpu_mtj_backend.tokenizer
|
||||||
if(vars.badwordsids is vars.badwordsids_default and vars.model_type not in ("gpt2", "gpt_neo", "gptj")):
|
if(vars.badwordsids is vars.badwordsids_default and vars.model_type not in ("gpt2", "gpt_neo", "gptj")):
|
||||||
vars.badwordsids = [[v] for k, v in tokenizer.get_vocab().items() if any(c in str(k) for c in "<>[]")]
|
vars.badwordsids = [[v] for k, v in tokenizer.get_vocab().items() if any(c in str(k) for c in "<>[]") if vars.newlinemode != "s" or str(k) != "</s>"]
|
||||||
else:
|
else:
|
||||||
loadsettings()
|
loadsettings()
|
||||||
|
|
||||||
|
|
Loading…
Reference in New Issue