Simplify weight loading in Transformers backend (#21382)

Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
This commit is contained in:
Harry Mellor
2025-07-22 20:29:43 -07:00
committed by GitHub
parent 3ec7170ff1
commit f154bb9ff0
7 changed files with 53 additions and 76 deletions
+1 -1
View File
@@ -500,7 +500,7 @@ _SPECULATIVE_DECODING_EXAMPLE_MODELS = {
}
_TRANSFORMERS_MODELS = {
"TransformersForCausalLM": _HfExamplesInfo("ArthurZ/Ilama-3.2-1B", trust_remote_code=True), # noqa: E501
"TransformersForCausalLM": _HfExamplesInfo("hmellor/Ilama-3.2-1B", trust_remote_code=True), # noqa: E501
"TransformersForMultimodalLM": _HfExamplesInfo("OpenGVLab/InternVL3-1B-hf"),
}
+1 -1
View File
@@ -56,7 +56,7 @@ def check_implementation(
"model,model_impl",
[
("meta-llama/Llama-3.2-1B-Instruct", "transformers"),
("ArthurZ/Ilama-3.2-1B", "auto"), # CUSTOM CODE
("hmellor/Ilama-3.2-1B", "auto"), # CUSTOM CODE
]) # trust_remote_code=True by default
def test_models(
hf_runner: type[HfRunner],