Hugging Face's logo

TiGa-RCE
/
Qwen3-Embedding-0.6B-MLX-oQ8e

Feature Extraction
MLX
Safetensors
qwen3
embeddings
sentence-similarity
quantization
omlx
oq8e
8-bit precision
Model card Files
xet
 Community
Qwen3-Embedding-0.6B-MLX-oQ8e
File size: 1,891 Bytes
0af7da1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
{
  "schema_version": 1,
  "published_repository": "TiGa-RCE/Qwen3-Embedding-0.6B-MLX-oQ8e",
  "family": "Qwen3-Embedding-0.6B",
  "variant": "oQ8e",
  "upstream_repository": "Qwen/Qwen3-Embedding-0.6B",
  "upstream_revision": "97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3",
  "upstream_revision_evidence": "exact source snapshot retained in local Hugging Face download metadata",
  "direct_parent": "TiGa-RCE/Qwen3-Embedding-0.6B-MLX-BF16",
  "direct_parent_weight_hashes": [
    {
      "file": "model.safetensors",
      "sha256": "87d77ce03148999dec3d5b73168df202d4b15862ea335658ec015e9bba07b14f",
      "bytes": 1191588265
    }
  ],
  "conversion": {
    "method": "oQe calibrated mixed-precision affine quantization",
    "nominal_bits": 8,
    "group_size": 64,
    "importance_matrix": true,
    "importance_matrix_samples": 128,
    "importance_matrix_sequence_length": 512,
    "stack": {
      "omlx": "0.5.3",
      "mlx_lm": "0.31.3",
      "mlx": "0.32.0"
    },
    "lossy_parent": false
  },
  "weight_files": [
    {
      "file": "model.safetensors",
      "sha256": "7c72d2475abca87ccf56f7360e673b98dd6d40442a793e868f80a40cd266009d",
      "bytes": 633152522
    }
  ],
  "evaluation": {
    "pair_count": 24,
    "top1": 1.0,
    "recall_at_5": 1.0,
    "mrr": 1.0,
    "mean_aligned_embedding_cosine_vs_bf16": 0.9994478225708008,
    "minimum_aligned_embedding_cosine_vs_bf16": 0.9986600279808044,
    "score_rmse_vs_bf16": 0.0030556409619748592,
    "queries_with_rank_change": 0,
    "gate_passed": true,
    "gate_criteria": {
      "top1_delta_min": 0.0,
      "recall_at_5_delta_min": 0.0,
      "mrr_delta_min": -0.01,
      "minimum_aligned_embedding_cosine_min": 0.99,
      "queries_with_rank_change_max": 2
    }
  },
  "collection": "https://huggingface.co/collections/TiGa-RCE/mlx-embedding-quantization-matrix-q-oq-oqe-at-4-6-8-bit-6a68d11afb238d4fe967d70b"
}