matt/ollama37
1
0
Fork 0
mirror of https://github.com/dogkeeper886/ollama37.git synced 2025-12-19 04:07:01 +00:00
Code Issues Packages Projects Releases Wiki Activity

unbound max num gpu layers (#591)

Browse Source 
---------

Co-authored-by: Michael Yang <mxyng@pm.me>
This commit is contained in:
Bruce MacDonald
2023-09-25 23:36:46 +01:00
committed by GitHub
parent b934bf23e6
commit 86279f4ae3
4 changed files with 36 additions and 29 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
llm/gguf.go
View File

@@ -195,6 +195,16 @@ func (llm *ggufModel) Decode(r io.Reader) error {
return nil
}
func (llm *ggufModel) NumLayers() int64 {
value, exists := llm.kv[fmt.Sprintf("%s.block_count", llm.ModelFamily())]
if !exists {
return 0
}
v := value.(uint32)
return int64(v)
}
func (ggufModel) readU8(r io.Reader) uint8 {
var u8 uint8
binary.Read(r, binary.LittleEndian, &u8)