mamba-architecture
Tools and guidance for using Mamba state-space models in high-performance machine learning tasks.
Install
mkdir -p .claude/skills/mamba-architecture && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/2145" && unzip -o skill.zip -d .claude/skills/mamba-architecture && rm skill.zipInstalls to .claude/skills/mamba-architecture
Activation
This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.
State-space model with O(n) complexity vs Transformers' O(n²). 5× faster inference, million-token sequences, no KV cache. Selective SSM with hardware-aware design. Mamba-1 (d_state=16) and Mamba-2 (d_state=128, multi-head). Models 130M-2.8B on HuggingFace.Key capabilities
- →Configure Mamba-1/Mamba-2 model architectures
- →Implement SSM blocks within PyTorch
- →Replace transformer layers with state-space layers
- →Optimize sequence generation for linear complexity
How it works
It abstracts the integration of selective state-space models by calling specialized `mamba-ssm` routines that replace standard self-attention mechanisms with hardware-aware linear operators.
Inputs & outputs
When to use mamba-architecture
- →Implementing SSM models
- →Optimizing sequence generation inference
- →Replacing Transformer layers with Mamba
About this skill
Mamba - Selective State Space Models
Quick start
Mamba is a state-space model architecture achieving O(n) linear complexity for sequence modeling.
Installation:
# Install causal-conv1d (optional, for efficiency)
pip install causal-conv1d>=1.4.0
# Install Mamba
pip install mamba-ssm
# Or both together
pip install mamba-ssm[causal-conv1d]
Prerequisites: Linux, NVIDIA GPU, PyTorch 1.12+, CUDA 11.6+
Basic usage (Mamba block):
import torch
from mamba_ssm import Mamba
batch, length, dim = 2, 64, 16
x = torch.randn(batch, length, dim).to("cuda")
model = Mamba(
d_model=dim, # Model dimension
d_state=16, # SSM state dimension
d_conv=4, # Conv1d kernel size
expand=2 # Expansion factor
).to("cuda")
y = model(x) # O(n) complexity!
assert y.shape == x.shape
Common workflows
Workflow 1: Language model with Mamba-2
Complete LM with generation:
from mamba_ssm.models.mixer_seq_simple import MambaLMHeadModel
from mamba_ssm.models.config_mamba import MambaConfig
import torch
# Configure Mamba-2 LM
config = MambaConfig(
d_model=1024, # Hidden dimension
n_layer=24, # Number of layers
vocab_size=50277, # Vocabulary size
ssm_cfg=dict(
layer="Mamba2", # Use Mamba-2
d_state=128, # Larger state for Mamba-2
headdim=64, # Head dimension
ngroups=1 # Number of groups
)
)
model = MambaLMHeadModel(config, device="cuda", dtype=torch.float16)
# Generate text
input_ids = torch.randint(0, 1000, (1, 20), device="cuda", dtype=torch.long)
output = model.generate(
input_ids=input_ids,
max_length=100,
temperature=0.7,
top_p=0.9
)
Workflow 2: Use pretrained Mamba models
Load from HuggingFace:
from transformers import AutoTokenizer
from mamba_ssm.models.mixer_seq_simple import MambaLMHeadModel
# Load pretrained model
model_name = "state-spaces/mamba-2.8b"
tokenizer = AutoTokenizer.from_pretrained("EleutherAI/gpt-neox-20b") # Use compatible tokenizer
model = MambaLMHeadModel.from_pretrained(model_name, device="cuda", dtype=torch.float16)
# Generate
prompt = "The future of AI is"
input_ids = tokenizer(prompt, return_tensors="pt").input_ids.to("cuda")
output_ids = model.generate(
input_ids=input_ids,
max_length=200,
temperature=0.7,
top_p=0.9,
repetition_penalty=1.2
)
generated_text = tokenizer.decode(output_ids[0])
print(generated_text)
Available models:
state-spaces/mamba-130mstate-spaces/mamba-370mstate-spaces/mamba-790mstate-spaces/mamba-1.4bstate-spaces/mamba-2.8b
Workflow 3: Mamba-1 vs Mamba-2
Mamba-1 (smaller state):
from mamba_ssm import Mamba
model = Mamba(
d_model=256,
d_state=16, # Smaller state dimension
d_conv=4,
expand=2
).to("cuda")
Mamba-2 (multi-head, larger state):
from mamba_ssm import Mamba2
model = Mamba2(
d_model=256,
d_state=128, # Larger state dimension
d_conv=4,
expand=2,
headdim=64, # Head dimension for multi-head
ngroups=1 # Parallel groups
).to("cuda")
Key differences:
- State size: Mamba-1 (d_state=16) vs Mamba-2 (d_state=128)
- Architecture: Mamba-2 has multi-head structure
- Normalization: Mamba-2 uses RMSNorm
- Distributed: Mamba-2 supports tensor parallelism
Workflow 4: Benchmark vs Transformers
Generation speed comparison:
# Benchmark Mamba
python benchmarks/benchmark_generation_mamba_simple.py \
--model-name "state-spaces/mamba-2.8b" \
--prompt "The future of machine learning is" \
--topp 0.9 --temperature 0.7 --repetition-penalty 1.2
# Benchmark Transformer
python benchmarks/benchmark_generation_mamba_simple.py \
--model-name "EleutherAI/pythia-2.8b" \
--prompt "The future of machine learning is" \
--topp 0.9 --temperature 0.7 --repetition-penalty 1.2
Expected results:
- Mamba: 5× faster inference
- Memory: No KV cache needed
- Scaling: Linear with sequence length
When to use vs alternatives
Use Mamba when:
- Need long sequences (100K+ tokens)
- Want faster inference than Transformers
- Memory-constrained (no KV cache)
- Building streaming applications
- Linear scaling important
Advantages:
- O(n) complexity: Linear vs quadratic
- 5× faster inference: No attention overhead
- No KV cache: Lower memory usage
- Million-token sequences: Hardware-efficient
- Streaming: Constant memory per token
Use alternatives instead:
- Transformers: Need best-in-class performance, have compute
- RWKV: Want RNN+Transformer hybrid
- RetNet: Need retention-based architecture
- Hyena: Want convolution-based approach
Common issues
Issue: CUDA out of memory
Reduce batch size or use gradient checkpointing:
model = MambaLMHeadModel(config, device="cuda", dtype=torch.float16)
model.gradient_checkpointing_enable() # Enable checkpointing
Issue: Slow installation
Install binary wheels (not source):
pip install mamba-ssm --no-build-isolation
Issue: Missing causal-conv1d
Install separately:
pip install causal-conv1d>=1.4.0
Issue: Model not loading from HuggingFace
Use MambaLMHeadModel.from_pretrained (not AutoModel):
from mamba_ssm.models.mixer_seq_simple import MambaLMHeadModel
model = MambaLMHeadModel.from_pretrained("state-spaces/mamba-2.8b")
Advanced topics
Selective SSM: See references/selective-ssm.md for mathematical formulation, state-space equations, and how selectivity enables O(n) complexity.
Mamba-2 architecture: See references/mamba2-details.md for multi-head structure, tensor parallelism, and distributed training setup.
Performance optimization: See references/performance.md for hardware-aware design, CUDA kernels, and memory efficiency techniques.
Hardware requirements
- GPU: NVIDIA with CUDA 11.6+
- VRAM:
- 130M model: 2GB
- 370M model: 4GB
- 790M model: 8GB
- 1.4B model: 14GB
- 2.8B model: 28GB (FP16)
- Inference: 5× faster than Transformers
- Memory: No KV cache (lower than Transformers)
Performance (vs Transformers):
- Speed: 5× faster inference
- Memory: 50% less (no KV cache)
- Scaling: Linear vs quadratic
Resources
- Paper (Mamba-1): https://arxiv.org/abs/2312.00752 (Dec 2023)
- Paper (Mamba-2): https://arxiv.org/abs/2405.21060 (May 2024)
- GitHub: https://github.com/state-spaces/mamba ⭐ 13,000+
- Models: https://huggingface.co/state-spaces
- Docs: Repository README and wiki
When not to use it
- →Tasks requiring standard Transformer pre-training data
- →Environments without CUDA/GPU support
Prerequisites
Limitations
- →Limited ecosystem support compared to standard Transformers
- →High dependency on specific GPU hardware configurations
How it compares
It enables training and inference with linear O(n) complexity, avoiding the KV cache memory bottlenecks of Transformers.
Compared to similar skills
mamba-architecture side by side with the closest alternatives in the catalog.
| Skill | Installs | Updated | Safety | Difficulty |
|---|---|---|---|---|
| mamba-architecture (this skill) | 2 | 7mo | Review | Advanced |
| gguf-quantization | 6 | 7mo | Review | Intermediate |
| pytorch-fsdp2 | 6 | 6mo | No flags | Advanced |
| huggingface-accelerate | 3 | 7mo | Review | Intermediate |
Try saying
Example prompts that trigger this skill in your AI assistant.
More by davila7
View all by davila7 →You might also like
gguf-quantization
davila7
GGUF format and llama.cpp quantization for efficient CPU/GPU inference. Use when deploying models on consumer hardware, Apple Silicon, or when needing flexible quantization from 2-8 bit without GPU requirements.
pytorch-fsdp2
Orchestra-Research
Adds PyTorch FSDP2 (fully_shard) to training scripts with correct init, sharding, mixed precision/offload config, and distributed checkpointing. Use when models exceed single-GPU memory or when you need DTensor-based sharding with DeviceMesh.
huggingface-accelerate
davila7
Simplest distributed training API. 4 lines to add distributed support to any PyTorch script. Unified API for DeepSpeed/FSDP/Megatron/DDP. Automatic device placement, mixed precision (FP16/BF16/FP8). Interactive config, single launch command. HuggingFace ecosystem standard.
model-pruning
davila7
Reduce LLM size and accelerate inference using pruning techniques like Wanda and SparseGPT. Use when compressing models without retraining, achieving 50% sparsity with minimal accuracy loss, or enabling faster inference on hardware accelerators. Covers unstructured pruning, structured pruning, N:M sparsity, magnitude pruning, and one-shot methods.
metal-kernel
pytorch
Write Metal/MPS kernels for PyTorch operators. Use when adding MPS device support to operators, implementing Metal shaders, or porting CUDA kernels to Apple Silicon. Covers native_functions.yaml dispatch, host-side operators, and Metal kernel implementation.
optimizing-attention-flash
davila7
Optimizes transformer attention with Flash Attention for 2-4x speedup and 10-20x memory reduction. Use when training/running transformers with long sequences (>512 tokens), encountering GPU memory issues with attention, or need faster inference. Supports PyTorch native SDPA, flash-attn library, H100 FP8, and sliding window attention.