diff --git a/examples/values/falcon-40b.yaml b/examples/values/falcon-40b.yaml index 8fed4f11a4bcfc9b220df63c967e32a0b008e3e8..1c309d350a57bbdf0ff4c3ec5a50adddcd5087d1 100644 --- a/examples/values/falcon-40b.yaml +++ b/examples/values/falcon-40b.yaml @@ -3,7 +3,7 @@ deployment: image: quay.io/chenhunghan/ialacol:latest env: DEFAULT_MODEL_HG_REPO_ID: TheBloke/WizardLM-Uncensored-Falcon-40B-GGML - DEFAULT_MODEL_FILE: wizard-falcon40b.ggmlv3.q4_K_S.bin + DEFAULT_MODEL_FILE: wizardlm-uncensored-falcon-40b.ggccv1.q4_0.bin resources: {} cache: diff --git a/examples/values/falcon-7b.yaml b/examples/values/falcon-7b.yaml index 1b59128d62222beb673ef88b53fc4a2d99c63997..d7b8d4803a720cf4998ee05f6460fdfed5fd8a3b 100644 --- a/examples/values/falcon-7b.yaml +++ b/examples/values/falcon-7b.yaml @@ -3,7 +3,9 @@ deployment: image: quay.io/chenhunghan/ialacol:latest env: DEFAULT_MODEL_HG_REPO_ID: TheBloke/WizardLM-Uncensored-Falcon-7B-GGML - DEFAULT_MODEL_FILE: wizard-falcon-7b.ggmlv3.q4_1.bin + DEFAULT_MODEL_FILE: wizardlm-7b-uncensored.ggccv1.q4_0.bin + # the file name of the model file is missing `falcon` so we need to set the model type + MODE_TYPE: falcon resources: {} cache: diff --git a/get_llm.py b/get_llm.py index eef6bd544bd7ee71955463ce244d96387e8cc366..f112c98aeb4fbfa8e4b4c23f19b98e2ac5f442eb 100644 --- a/get_llm.py +++ b/get_llm.py @@ -38,7 +38,10 @@ async def get_llm( ctransformer_model_type = "dolly-v2" if "stablelm" in body.model: ctransformer_model_type = "gpt_neox" - + + MODE_TYPE = get_env("MODE_TYPE", "") + if len(MODE_TYPE) > 0: + ctransformer_model_type = MODE_TYPE MODELS_FOLDER = get_env("MODELS_FOLDER", "models") return AutoModelForCausalLM.from_pretrained(