Feature Extraction
MLX
Safetensors
ministral3
apple-silicon
quantized
mixed-precision
axquant
axq
development
experimental
mistral3
6bit
6-bit
embedding
sentence-similarity
4-bit precision
Instructions to use AutomatosX/AX-Nemotron-3-Embed-8B-MLX-AXQ-6bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use AutomatosX/AX-Nemotron-3-Embed-8B-MLX-AXQ-6bit with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] huggingface-cli download --local-dir AX-Nemotron-3-Embed-8B-MLX-AXQ-6bit AutomatosX/AX-Nemotron-3-Embed-8B-MLX-AXQ-6bit
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Atomic Chat
Download axquant_manifest.json from AutomatosX/AX-Nemotron-3-Embed-8B-MLX-AXQ-6bit: direct link, hf CLI and curl.
- Browser
- Download file 5.4 kB
-
https://huggingface.co/AutomatosX/AX-Nemotron-3-Embed-8B-MLX-AXQ-6bit/resolve/main/axquant_manifest.json
- Command line
-
hf download hf://AutomatosX/AX-Nemotron-3-Embed-8B-MLX-AXQ-6bit/axquant_manifest.json
-
curl -L -o axquant_manifest.json https://huggingface.co/AutomatosX/AX-Nemotron-3-Embed-8B-MLX-AXQ-6bit/resolve/main/axquant_manifest.json
5.4 kB
| { | |
| "axquant_version": "1.6.1", | |
| "calibration": null, | |
| "created_at": "2026-08-08T18:35:07.890564Z", | |
| "effective_bpw": 5.999827974534051, | |
| "files": [ | |
| { | |
| "path": "README.md", | |
| "sha256": "4c7e67b971f351f520872282de0d91924dd6c610ff59c12e61a41262798a9e0e", | |
| "size_bytes": 7763 | |
| }, | |
| { | |
| "path": "axquant_plan.json", | |
| "sha256": "30d4f09e558fe4723eb5aa38dfb9e526f96650247df1838a8dfbe8bc6f7dd5e5", | |
| "size_bytes": 291159 | |
| }, | |
| { | |
| "path": "axquant_quantizer_execution.json", | |
| "sha256": "73bd459d778fc15a56c6de69dce9a62cf31ef4c96a5ae36365289002f4a1a714", | |
| "size_bytes": 56871 | |
| }, | |
| { | |
| "path": "axquant_runtime.json", | |
| "sha256": "6cc9a88b8701bb5b01b85e69ab418740ab74fcf973806bf27ab055a87be175a1", | |
| "size_bytes": 1628 | |
| }, | |
| { | |
| "path": "config.json", | |
| "sha256": "f84fb5bb0d016c64a1f902b22e736d8a12f7244e20554bccfe3494e16fd4d175", | |
| "size_bytes": 66967 | |
| }, | |
| { | |
| "path": "model-00001-of-00002.safetensors", | |
| "sha256": "ab6cc4a47269a89c197e0d46ca3be7f49105c31f1c05b3c8401b9b7888374cbf", | |
| "size_bytes": 5337300123 | |
| }, | |
| { | |
| "path": "model-00002-of-00002.safetensors", | |
| "sha256": "fdd9e25a1ac44ef21262d7e648ac3c33877352e3118bec946795b02d5b967a05", | |
| "size_bytes": 627130954 | |
| }, | |
| { | |
| "path": "model-manifest.json", | |
| "sha256": "55b7b7c8b072799833d92d5ac989eb57eaaf1964df8d6cff0aebe7e98a2aefe8", | |
| "size_bytes": 127644 | |
| }, | |
| { | |
| "path": "model.safetensors.index.json", | |
| "sha256": "ae909649f3e16180ce8c6fb84303532d83ff2ea78fedbf1db739041d87e360ed", | |
| "size_bytes": 67330 | |
| }, | |
| { | |
| "path": "tokenizer.json", | |
| "sha256": "797410dfb649a5b9ba92bc4fef7dbf4022d00e73de6867c4ac199a8846439421", | |
| "size_bytes": 17077678 | |
| }, | |
| { | |
| "path": "tokenizer_config.json", | |
| "sha256": "7bbb77c55282bc679b86d21a1b3953ed8d5d63aaf75f10800c20596eab30b980", | |
| "size_bytes": 21139 | |
| } | |
| ], | |
| "format": "mlx", | |
| "logical_parameters": 7952683008, | |
| "main_logical_parameters": 7952683008, | |
| "main_weight_file_size_bytes": 5964431077, | |
| "measured_main_bpw": 5.999918337999975, | |
| "measured_total_bpw": 5.999918337999975, | |
| "mtp_acceptance_retention": null, | |
| "mtp_distribution": {}, | |
| "mtp_measured_speedup": null, | |
| "mtp_policy": { | |
| "candidate_bits": [ | |
| 8, | |
| 16 | |
| ], | |
| "min_bits": 8, | |
| "mode": "protected", | |
| "optimize_for_acceptance": true, | |
| "preserve_external_sidecar": true, | |
| "protect_norms": true, | |
| "protect_output_head": true | |
| }, | |
| "mtp_present": false, | |
| "mtp_weight_file_size_bytes": 0, | |
| "plan_sha256": "21b591094cae166a5de3e03dd9de8d59b4b307aacecb0b5c8d1401fcca9cb564", | |
| "profile": "general", | |
| "protected_weight_file_size_bytes": 0, | |
| "quantizer": "axquant", | |
| "runtime": { | |
| "ax_engine": { | |
| "decode_kernel": null, | |
| "fused_mtp": null, | |
| "kernel_evidence": "unmeasured", | |
| "model_manifest": "model-manifest.json", | |
| "preferred_group_size": 32 | |
| }, | |
| "compatible_runtimes": [ | |
| { | |
| "compatibility_level": "B", | |
| "manifest": "config.json", | |
| "mtp_support": "none", | |
| "name": "mlx-lm", | |
| "notes": [ | |
| "Standard backbone inference is the compatibility target.", | |
| "AXQuant MTP metadata may be ignored by MLX-LM." | |
| ], | |
| "standard_inference": true, | |
| "standard_mlx_weights": true, | |
| "support_level": "standard-inference" | |
| } | |
| ], | |
| "created_at": "2026-08-08T18:35:04.917442Z", | |
| "kv_cache": null, | |
| "memory_policy": { | |
| "kv_cache_precision": "runtime-default", | |
| "mtp_buffers": "not-required", | |
| "prefix_cache": "runtime-managed", | |
| "unified_memory_safety_margin": "benchmark-required" | |
| }, | |
| "mtp": { | |
| "acceptance_retention": null, | |
| "detected": false, | |
| "draft_tokens": null, | |
| "enabled_by_default": false, | |
| "head_precision": null, | |
| "measured_speedup": null, | |
| "optimized": false, | |
| "recommended_temperature_max": null, | |
| "sidecar_file": null, | |
| "verification_mode": null | |
| }, | |
| "optimization_scope": "text-path", | |
| "primary_runtime": { | |
| "compatibility_level": "A", | |
| "manifest": "model-manifest.json", | |
| "mtp_support": "none", | |
| "name": "ax-engine", | |
| "notes": [ | |
| "Runtime claims require a passing AX Engine doctor and benchmark report." | |
| ], | |
| "standard_inference": true, | |
| "standard_mlx_weights": true, | |
| "support_level": "optimized" | |
| }, | |
| "schema_version": "axquant.runtime.v1" | |
| }, | |
| "schema_version": "axquant.artifact.v2", | |
| "software_versions": { | |
| "ax_engine": "6.11.1", | |
| "axquant": "1.6.1", | |
| "mlx": "0.32.0", | |
| "mlx_lm": "0.31.3", | |
| "pydantic": "2.13.4", | |
| "python": "3.13.15", | |
| "safetensors": "0.8.0" | |
| }, | |
| "source_model": { | |
| "architecture": "Ministral3Model", | |
| "format": "mlx", | |
| "local_path": null, | |
| "model_id": "nvidia/Nemotron-3-Embed-8B-BF16", | |
| "revision": null | |
| }, | |
| "target_class": "6bit", | |
| "weight_distribution": { | |
| "4bit": { | |
| "fraction": 0.4282548222497944, | |
| "parameters": 3405774848 | |
| }, | |
| "6bit": { | |
| "fraction": 0.5042014902349796, | |
| "parameters": 4009754624 | |
| }, | |
| "8bit": { | |
| "fraction": 0.06750814932016463, | |
| "parameters": 536870912 | |
| }, | |
| "bf16": { | |
| "fraction": 3.5538195061426996e-05, | |
| "parameters": 282624 | |
| } | |
| }, | |
| "weight_file_size_bytes": 5964431077 | |
| } | |