models.ini

 1[ggml-org/embeddinggemma-2-GGUF:Q8_0]
 2hf = ggml-org/embeddinggemma-2-GGUF:Q8_0
 3alias = embedding
 4n-gpu-layers = 99
 5ctx-size = 2048
 6embedding = true
 7
 8[F2LLM-v2-330M-GGUF:Q8_0]
 9model = /home/pavle/.local/share/llama-server/models/f2llm-v2-330m-q8_0.gguf
10alias = f2llm-v2-330m
11n-gpu-layers = 99
12embedding = true
13
14[unsloth/Qwen3.8-27B-GGUF:UD-IQ3_S]
15hf = unsloth/Qwen3.8-27B-GGUF:UD-IQ3_S
16alias = qwen3.8-27b-q3
17n-gpu-layers = 999
18n-gpu-layers-draft = 999
19fit = off
20flash-attn = on
21jinja = true
22ctx-size = 131072
23temp = 0.6
24top-p = 0.95
25top-k = 20
26min-p = 0.0
27parallel = 1
28spec-type = draft-mtp
29spec-draft-n-max = 2
30repeat-penalty = 1.0
31presence-penalty = 1.5
32reasoning = on
33chat-template-kwargs = {"reasoning_effort":"medium"}
34
35[ukisai/Swift-1.5-Qwen3.8-27B-GGUF:Q3_K_S]
36hf = ukisai/Swift-1.5-Qwen3.8-27B-GGUF:Q3_K_S
37alias = swift-1.5-27b-q3,reviewer
38n-gpu-layers = auto
39n-gpu-layers-draft = 99
40fit = off
41flash-attn = on
42jinja = true
43ctx-size = 131072
44temp = 0.6
45top-p = 0.95
46top-k = 20
47min-p = 0.0
48parallel = 1
49spec-type = draft-mtp
50spec-draft-n-max = 2
51repeat-penalty = 1.0
52presence-penalty = 1.5
53reasoning = on
54chat-template-kwargs = {"reasoning_effort":"medium"}
55
56[ornith-ai/Ornith-1.5-9B-GGUF:Q6_K]
57hf = ornith-ai/Ornith-1.5-9B-GGUF:Q6_K
58alias = ornith-1.5-9B-q6,classifier
59n-gpu-layers = 99
60n-gpu-layers-draft = 99
61fit = off
62flash-attn = on
63jinja = true
64ctx-size = 8192
65temp = 0.2
66top-p = 0.95
67top-k = 20
68min-p = 0.0
69parallel = 1
70spec-type = draft-mtp
71spec-draft-n-max = 2
72repeat-penalty = 1.0
73presence-penalty = 1.5
74reasoning = on
75chat-template-kwargs = {"reasoning_effort":"low"}
76
77[unsloth/gemma-4-26B-A4B-it-qat-GGUF:Q4_K_XL]
78hf = unsloth/gemma-4-26B-A4B-it-qat-GGUF:UD-Q4_K_XL
79alias = gemma4-26b-a4b
80n-gpu-layers = 99
81n-gpu-layers-draft = 99
82ngl=999
83flash-attn = on
84ctx-size = 131072
85batch-size = 2048
86ubatch-size = 2048
87temp = 0.85
88top-p = 0.9
89top-k = 32
90spec-type = draft-mtp
91spec-draft-n-max = 4
92fit = off
93reasoning = auto