EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

Update version to v0.1.13 (#280)

Browse Source
This commit is contained in:
Lianmin Zheng
2024-03-11 05:49:27 -07:00
committed by GitHub
parent 13662fd533
commit 4aa5dd2c5f
11 changed files with 35 additions and 21 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
test/srt/model/test_llama_extend.py
View File

@@ -28,8 +28,8 @@ def test_generate_worker(model_path, tp_rank, tp_size):
reqs = []
for i in range(len(prompts)):
req = Req(i, None, None)
req.input_ids = tokenizer.encode(prompts[i])[:cut_num]
input_ids = tokenizer.encode(prompts[i])[:cut_num]
req = Req(i, prompts[i], input_ids)
req.sampling_params = sampling_params
reqs.append(req)
@@ -60,7 +60,7 @@ def test_generate_worker(model_path, tp_rank, tp_size):
# Decode
for i in range(6):
batch.prepare_for_decode(next_token_ids.cpu().numpy())
logits = model.forward(batch, ForwardMode.DECODE)
logits, _ = model.forward(batch, ForwardMode.DECODE)
next_token_ids, next_token_probs = batch.sample(logits)
print(