Code
Hub
Workspaces
Connect
Indexed graphs
Engine
MCP
copy
hub
/
github.com/ModelTC/LightLLM
/ cuda
Method
cuda
lightllm/models/vit/model.py:200–201 ·
view source on GitHub ↗
(self)
Source
from the content-addressed store, hash-verified
198
return
all_img_embeds, uuids, valid_ids
199
200
def
cuda(self):
201
return
self
202
203
def
load_model(self, weight_dir):
204
pass
Callers
15
exposed_init_model
Method · 0.45
sample
Function · 0.45
_get_post_sample_tensors
Function · 0.45
beam_prefill
Method · 0.45
prefill_normal
Method · 0.45
prefill_overlap
Method · 0.45
prefill_mtp
Method · 0.45
decode_mtp
Method · 0.45
prefill_overlap_mtp
Method · 0.45
decode_overlap_mtp
Method · 0.45
prefill_normal
Method · 0.45
prefill_mtp
Method · 0.45
Calls
no outgoing calls
Tested by
15
test_scatter_token_to_cpu
Function · 0.36
test_gather_token
Function · 0.36
test_tma_align
Function · 0.36
test_per_token_group_quant_fp8
Function · 0.36
tppart_model_infer
Function · 0.36
test_decode_attentions
Function · 0.36
test_decode_attentions
Function · 0.36
test_kernel
Function · 0.36
test_fp8_block_gemm
Function · 0.36
test_gen_prefill_params_basic
Function · 0.36
test_multimodal_emb
Function · 0.36
test_gen_mtp_new_input_ids_0
Function · 0.36