Sage-S2.1-Flash-oQ3e-mtp

oQ3e mixed-precision MLX conversion of Sage S2.1 Flash for Apple Silicon.

Effective weight storage: 3.69 BPW ยท 148.35 GB ยท group size 64.

Sage S2.1 Flash is a scientific multimodal and agentic model from Shanghai Artificial Intelligence Laboratory / InternLM, built on the GLM-5.3-Flash architecture. This conversion retains the full expert set, vision encoder, multi-token-prediction head and original tokenizer, processor and chat template. No additional training, abliteration or expert pruning was applied.

Conversion

  • Source: creator BF16/FP32 checkpoint, internlm/Sage-S2.1-Flash@76f4513db7464485605a002a06a6190de0460574.
  • Format: MLX safetensors; oQ3e mixed precision, affine packing, group size 64.
  • Tooling: oMLX 0.8.0rc1 native oQe streaming converter.
  • Calibration: 128 samples, 512 tokens each, using the bundled code/multilingual corpus; collected independently for this quant.
  • Weights: 29 root shards, 148.35 GB / 138.16 GiB, including text, vision and MTP. Runtime and KV cache need additional memory.

Conversion details are in CONVERSION.json.

Mixed precision

3.69 effective bits per weight (BPW), including affine scales/biases, unquantized tensors and safetensors headers.

Component Weight precision Original parameters Stored GB
Routed MoE experts 3-bit / 4-bit 311,653,564,416 137.254404
Attention projections 8-bit 6,088,556,544 6.469091
Shared experts 8-bit 1,082,130,432 1.149764
Dense MLP projections 8-bit 452,984,832 0.481296
Token embeddings 8-bit 634,388,480 0.674038
Output head 8-bit 634,388,480 0.674038
Vision encoder BF16 563,627,008 1.127254
Linear-attention convolutions BF16 3,342,336 0.006685
Indexer pooling/compression 8-bit / BF16 89,662,464 0.101173
Text/MTP normalization BF16 422,144 0.000844
MTP fusion projection BF16 33,554,432 0.067109
MoE routers/correction biases FP32 50,737,248 0.202949
Hyperconnections FP32 35,391,870 0.141567
Recurrent decay controls FP32 280,704 0.001123
Safetensors headers โ€” โ€” 0.000402
Total 3.69 effective BPW 321,323,031,390 148.351738

Stored GB uses decimal units. Component totals include MTP and exclude the separate scientific projectors. The nine original scientific projectors add 0.168 GB and remain in FP32.

Per-block precision map
Block Routed expert matrices Other quantized matrices Unquantized tensors Stored GB
Text 0 โ€” (dense MLP) 8-bit BF16 / FP32 0.310052
Text 1 โ€” (dense MLP) 8-bit BF16 / FP32 0.310052
Text 2 โ€” (dense MLP) 8-bit BF16 / FP32 0.310052
Text 3 3-bit 8-bit BF16 / FP32 3.338730
Text 4 3-bit 8-bit BF16 / FP32 3.351972
Text 5 3-bit 8-bit BF16 / FP32 3.351972
Text 6 3-bit 8-bit BF16 / FP32 3.351972
Text 7 3-bit 8-bit BF16 / FP32 3.338730
Text 8 3-bit 8-bit BF16 / FP32 3.351972
Text 9 3-bit 8-bit BF16 / FP32 3.351972
Text 10 3-bit 8-bit BF16 / FP32 3.351972
Text 11 3-bit 8-bit BF16 / FP32 3.338730
Text 12 3-bit 8-bit BF16 / FP32 3.351972
Text 13 3-bit 8-bit BF16 / FP32 3.351972
Text 14 3-bit 8-bit BF16 / FP32 3.351972
Text 15 3-bit 8-bit BF16 / FP32 3.338730
Text 16 3-bit 8-bit BF16 / FP32 3.351972
Text 17 3-bit 8-bit BF16 / FP32 3.351972
Text 18 3-bit 8-bit BF16 / FP32 3.351972
Text 19 3-bit 8-bit BF16 / FP32 3.338730
Text 20 3-bit 8-bit BF16 / FP32 3.351972
Text 21 3-bit 8-bit BF16 / FP32 3.351972
Text 22 3-bit 8-bit BF16 / FP32 3.351972
Text 23 3-bit 8-bit BF16 / FP32 3.338730
Text 24 3-bit 8-bit BF16 / FP32 3.351972
Text 25 3-bit 8-bit BF16 / FP32 3.351972
Text 26 3-bit 8-bit BF16 / FP32 3.351972
Text 27 3-bit 8-bit BF16 / FP32 3.338730
Text 28 3-bit 8-bit BF16 / FP32 3.351972
Text 29 3-bit 8-bit BF16 / FP32 3.351972
Text 30 3-bit 8-bit BF16 / FP32 3.351972
Text 31 3-bit 8-bit BF16 / FP32 3.338730
Text 32 3-bit 8-bit BF16 / FP32 3.351972
Text 33 3-bit 8-bit BF16 / FP32 3.351972
Text 34 3-bit 8-bit BF16 / FP32 3.351972
Text 35 3-bit 8-bit BF16 / FP32 3.338730
Text 36 3-bit 8-bit BF16 / FP32 3.351972
Text 37 3-bit 8-bit BF16 / FP32 3.351972
Text 38 3-bit 8-bit BF16 / FP32 3.351972
Text 39 3-bit 8-bit BF16 / FP32 3.338730
Text 40 3-bit 8-bit BF16 / FP32 3.351972
Text 41 3-bit 8-bit BF16 / FP32 3.351972
Text 42 3-bit 8-bit BF16 / FP32 3.351972
Text 43 3-bit 8-bit BF16 / FP32 3.338730
Text 44 3-bit 8-bit BF16 / FP32 3.351972
MTP 0 4-bit 8-bit BF16 / FP32 4.308687

Complete per-module/tensor precision data (CSV) and exact accounting (JSON) are included.

Running

Use oMLX 0.8.0rc1 for text, images, tool calling and native MTP decoding. Keep the included Sage tokenizer, processor and chat template together.

Scientific inputs require the matching SciPerception encoders and integration in addition to the included projector files.

Other precision variants

Each variant is converted directly from the same pinned creator BF16/FP32 checkpoint, not requantized from another variant.

Credits and license

Model: Shanghai Artificial Intelligence Laboratory / InternLM. GLM architecture: Z.ai. Conversion tooling: oMLX / oQe. Community conversion: Hygaard.

Licensed under MIT. See NOTICE for attribution and modification notes.

Downloads last month
67
Safetensors
Model size
321B params
Tensor type
U32
ยท
F32
ยท
BF16
ยท
MLX
Hardware compatibility
Log In to add your hardware

3-bit

Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for Hygaard/Sage-S2.1-Flash-oQ3e-mtp

Quantized
(12)
this model