File size: 1,605 Bytes
7f753b6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
models:
  # Base instructed model
  - model: unsloth/Qwen2.5-3B-Instruct
    parameters:
      weight: 1
      density: 1

  # Merged LoRA models
  - model: genstruct/merged_model
    parameters:
      weight: 1.0
      density: 1.0

  # - model: summary/merged_model
  #   parameters:
  #     weight: 1.0
  #     density: 1.0

  - model: kg/merged_model
    parameters:
      weight: 1.0
      density: 1.0

  #### THIS BREAKS KG!!!
  # - model: pII/merged_model
  #   parameters:
  #     weight: 1.0
  #     density: 1.0

  # #### Breaks KG!
  # - model: preference/merged_model
  #   parameters:
  #     weight: 1.0
  #     density: 1.0

  - model: triples/merged_model
    parameters:
      weight: 1.0
      density: 1.0

  # - model: suitable/merged_model
  #   parameters:
  #     weight: 1.0
  #     density: 1.0

  # - model: feedback/merged_model
  #   parameters:
  #     weight: 1.0
  #     density: 1.0

# Merge configuration
merge_method: ties
base_model: unsloth/Qwen2.5-3B
parameters:
  normalize: true
  int8_mask: true
dtype: bfloat16

# # Tokenizer configuration
# tokenizer_source: Qwen/Qwen1.5-14B-Chat
# tokenizer_parameters:
#   trust_remote_code: true
  
# # Output configuration
# output:
#   precision: bfloat16
#   model_format: safetensors
#   max_shard_size: "4GB"

# # Training configuration (for potential fine-tuning)
# training:
#   learning_rate: 2e-5
#   warmup_steps: 100
#   gradient_checkpointing: true
#   gradient_accumulation_steps: 4
  
# # Hardware optimization
# hardware:
#   mixed_precision: true
#   cuda_memory_fraction: 0.95
#   optimize_model_memory: true