vrr/kvcache-ai-ktransformers
2
0
Fork 0
mirror of https://github.com/kvcache-ai/ktransformers.git synced 2026-04-28 20:00:06 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 4

add balance-serve, support concurrence

Browse source
This commit is contained in:
Atream 2025-03-31 22:55:32 +08:00
parent 8d0292aa44
commit 25cee5810e
196 changed files with 22077 additions and 565 deletions
Download patch file Download diff file Expand all files Collapse all files

57
csrc/balance_serve/kvc2/config/quant_configs.json Normal file
View file

@ -0,0 +1,57 @@
{
"BF16": {
"block_element_count": 1,
"block_element_size": 2,
"bytes_per_element": 2.0,
"can_be_used_as_vector": true,
"has_min": false,
"has_scale": false,
"name": "BF16",
"reference": "",
"type_of_dot_vector": "BF16"
},
"FP16": {
"block_element_count": 1,
"block_element_size": 2,
"bytes_per_element": 2.0,
"can_be_used_as_vector": true,
"has_min": false,
"has_scale": false,
"name": "FP16",
"reference": "",
"type_of_dot_vector": "FP16"
},
"FP32": {
"block_element_count": 1,
"block_element_size": 4,
"bytes_per_element": 4.0,
"can_be_used_as_vector": true,
"has_min": false,
"has_scale": false,
"name": "FP32",
"reference": "",
"type_of_dot_vector": "FP32"
},
"Q4_0": {
"block_element_count": 32,
"block_element_size": 18,
"bytes_per_element": 0.5625,
"can_be_used_as_vector": false,
"has_min": false,
"has_scale": true,
"name": "Q4_0",
"reference": "https://huggingface.co/docs/hub/gguf",
"type_of_dot_vector": "Q8_0"
},
"Q8_0": {
"block_element_count": 32,
"block_element_size": 34,
"bytes_per_element": 1.0625,
"can_be_used_as_vector": true,
"has_min": false,
"has_scale": true,
"name": "Q8_0",
"reference": "https://huggingface.co/docs/hub/gguf",
"type_of_dot_vector": "Q8_0"
}
}