Mixedbread mxbai-rerank-large-v2
dense 1.776B
Fields
id
models/mxbai-rerank-large-v2Family
mxbai-rerank
Vendor
Mixedbread AI
Release year
2024
Architecture
dense
Params (B)
1.776
Active params (B)
1.776
Hidden size
1,536
Layers
28
KV heads
2
Head dim
128
GQA ratio
6
Max context
32,768
Context scaling method
native
Activation
silu
Attention variant
cross-encoder (Qwen2 causal decoder), full attention over query+passage
KV cache (bytes/token)
28,672
FLOPs per active token
3,550,000,000
Open weights
yes
License
Apache-2.0
notes
RERANKER: cross-encoder using a Qwen2 causal decoder backbone (hidden 1536, 28 layers, 12 heads, 2 KV heads, vocab 151936, 32768 max_position_embeddings). KV cache applies (28 * 2 * 128 * 2 * 2 = 28672 bytes/token at bf16) but is irrelevant for reranking (single forward pass per pair). 1.776B params, 3.55e9 flops/token — GPU-class work. Apache-2.0. GGUF and ONNX ports exist.
task
reranking
Notes
RERANKER: cross-encoder using a Qwen2 causal decoder backbone (hidden 1536, 28 layers, 12 heads, 2 KV heads, vocab 151936, 32768 max_position_embeddings). KV cache applies (28 * 2 * 128 * 2 * 2 = 28672 bytes/token at bf16) but is irrelevant for reranking (single forward pass per pair). 1.776B params, 3.55e9 flops/token — GPU-class work. Apache-2.0. GGUF and ONNX ports exist.
Cross-references
This record references no other records — it is a leaf. It is cited by 0.