fade-kv 0.3.0

FADE: Frequency-Adaptive Decay Encoding — attention-aware tiered KV cache compression for LLM inference.

FADE

Frequency-Adaptive Decay Encoding — drop-in KV cache compression for HuggingFace transformers. Shrinks the KV cache 3–5× with near-baseline quality.

from fade import FadeConfig, create_tiered_cache

cache = create_tiered_cache(model, config=FadeConfig.safe())
out = model.generate(input_ids, past_key_values=cache, max_new_tokens=256)

Works with model.generate() — greedy, sampling, beam search. No manual decode loop needed.

How it works

Tokens live in tiers based on age and attention importance:

Tier	What's stored	When
FP16	Full precision	First N_SINK tokens + last RECENT_WINDOW tokens
INT4	Bit-packed 4-bit	Middle-aged tokens (the bulk of the cache)
INT2	Grouped 2-bit	Optional deeper compression (lossy)
PQ	Product-quantized codes	~2 bits/element via trained codebook (Phase 3)
Evicted	Nothing	Dropped when INT4_BUDGET is finite

When tokens are evicted, surviving K tensors are un-RoPE'd at old positions and re-RoPE'd with contiguous StreamingLLM positions.

Install

From PyPI:

pip install fade-kv

From source (development):

python -m venv .venv
.\.venv\Scripts\Activate.ps1
pip install torch  # match your CUDA version: https://pytorch.org/get-started/locally/
pip install -e ".[dev]"

Optional extras: pip install fade-kv[cuda] (accelerate), fade-kv[eval] (datasets), fade-kv[codebook] (scikit-learn for PQ).

Quick start

Presets

from fade import FadeConfig, create_tiered_cache

# Safe: ~3-4x compression, 100% greedy match. No eviction.
cache = create_tiered_cache(model, config=FadeConfig.safe())

# Balanced: ~5x compression with H2O eviction.
cache = create_tiered_cache(model, config=FadeConfig.balanced())

# Aggressive: ~7-8x compression. Validate on your workload first.
cache = create_tiered_cache(model, config=FadeConfig.aggressive())

Custom config

cache = create_tiered_cache(model, config=FadeConfig(
    phase="2",
    n_sink=4,
    recent_window=64,
    int4_budget=400,
    eviction_policy="h2o",       # "h2o", "ema", "position", or "learned"
    middle_k_bits=4,             # K stays INT4 (outlier-sensitive)
    middle_v_bits=2,             # V at INT2 (~30% more compression)
))

Rotated 2-bit backend (~6× compression)

from fade.backends import get_backend

cache = create_tiered_cache(model, config=FadeConfig.safe(),
    quant_backend=get_backend("rotated", head_dim=64, bits=2))

Random orthogonal rotation spreads per-channel outliers before quantization, making 2-bit viable. Uses native PyTorch — no external dependencies.

Manual decode with tier reassignment

from fade.patch import forward_with_tracking, load_model
from fade.policy import reassign_tiers
from fade.tracker import AttentionTracker

model, tokenizer = load_model("Qwen/Qwen2.5-3B-Instruct", attn_impl="auto", need_attentions=True)
cache = create_tiered_cache(model, config=FadeConfig.balanced())
tracker = AttentionTracker(num_layers=model.config.num_hidden_layers)

out = forward_with_tracking(model, input_ids, cache, tracker=tracker)
for step in range(max_tokens):
    out = forward_with_tracking(model, next_token, cache, tracker=tracker)
    if (step + 1) % 64 == 0:
        reassign_tiers(cache, tracker, model.config.num_hidden_layers)

Eviction policies

Policy	Quality	Speed	Needs attention?
h2o	Best	Normal	Yes (prefill only)
ema	Good	Normal	Yes (decode only)
position	Fair	Fast	No
learned	Good*	Fast	No

*Learned policy requires a trained checkpoint: python scripts/train_eviction_mlp.py

Supported models

FADE auto-detects the RoPE scheme from the model config:

• Qwen2 / Qwen3 — vanilla RoPE, GQA
• Llama / Llama-3.1 — vanilla + frequency-dependent scaling
• Mistral — vanilla RoPE, sliding-window
• Phi-3 — vanilla RoPE
• Gemma-2 — vanilla RoPE
• Gemma 4 — proportional RoPE with partial_rotary_factor + per-layer-type dispatch
• Falcon — ALiBi (non-RoPE; re-RoPE is a no-op)
• Qwen 3.5 / 3.6 — hybrid DeltaNet + softmax attention. FADE auto-detects layer_types and skips DeltaNet layers (only full-attention layers are tiered).

RoPE scaling types: linear, llama3, ntk, dynamic, yarn, proportional. Non-RoPE models (ALiBi, Bloom, MPT) work via the NoRope sentinel.

Batching

Two modes:

• Shared-tier (default): all rows share positions and tier decisions. For lockstep decoding.
• Per-sequence (apply_tier_assignment_per_sequence): each row gets independent [B, S] tiers. For continuous-batching where sequences diverge.

Performance

• Fused INT4 FlashAttention kernel — single Triton kernel reads packed INT4 K/V, computes attention with online softmax, writes fp16 output. Never materializes fp16 K/V. 4.9× faster than the old dequant+SDPA path, within 1.4× of pure fp16 FlashAttention on RTX 3060.
• Pre-allocated FP16 buffer — doubling buffer eliminates torch.cat on every decode step.
• torch.compile — cache.enable_compile() wraps _materialize between graph-break boundaries.
• Dequant-cache age eviction — cache.max_dequant_age = N periodically refreshes cached dequant buffers.
• Benchmarks — python benchmarks/tps.py (decode throughput), python benchmarks/divergence.py (quality).

# Use the fused kernel directly:
from fade.kernels.fused_int4_attn import fused_int4_sdpa
out = fused_int4_sdpa(q, k_packed, k_scale, v_packed, v_scale)

Checkpointing

sd = cache.cache_state_dict()
torch.save(sd, "cache.pt")
cache.load_cache_state_dict(torch.load("cache.pt"))

Observability

from fade.telemetry import JsonlExporter, attach_telemetry
attach_telemetry(cache, JsonlExporter("events.jsonl"))

Debug dump: cache.dump_debug("snapshot.json")

PQ codebook

from fade.codebook import PQCodebook
cb = PQCodebook.train(calibration_vectors, sub_dim=32, num_centroids=256)
cache.set_codebooks(cb)  # enables TIER_PQ in tier assignment

Train codebooks from a real model: python scripts/train_codebook.py

Results

Config	Model	KV cache	Compression
INT4 symmetric	Qwen2.5-0.5B, 2K tok	6.78 MiB	3.5×
Rotated 4-bit	Qwen2.5-0.5B, 2K tok	6.78 MiB	3.5× (better quality)
Rotated 2-bit	Qwen2.5-0.5B, 2K tok	3.88 MiB	6.2×
Phase 2 H2O	Qwen2.5-3B, 595 tok	6.3 MiB	5× (with eviction)

Project layout

fade/
  cache.py           # TieredKVCache with 5 tiers (FP16/INT4/INT2/PQ/evict)
  config.py          # FadeConfig with presets
  quant.py           # INT4/INT2 quantization + bit-packing
  rope.py            # 7 RoPE schemes incl. Gemma 4 proportional
  policy.py          # Tier assignment: h2o, ema, position
  learned_policy.py  # Learned eviction MLP
  tracker.py         # AttentionTracker (per-layer EMA)
  patch.py           # load_model, create_tiered_cache, forward_with_tracking
  codebook.py        # PQ codebook train/encode/decode
  telemetry.py       # Structured telemetry + exporters
  kernels/           # Fused INT4 FlashAttention kernel + unpack kernel + fallback
  serving/           # vLLM / SGLang adapter stubs
  eval/              # Perplexity, needle, quality suite
examples/            # quickstart.py
experiments/         # run_baseline.py, run_tiered.py
benchmarks/          # tps.py, divergence.py
scripts/             # train_eviction_mlp.py, train_codebook.py
tests/               # 136 tests, all CPU, no downloads

Gotchas

1. Attention impl: eager only needed for H2O prefill. Use load_model(attn_impl="auto").
2. Transformers version: verified on 4.45 and 5.3. Weekly canary CI runs against transformers@main.
3. Memory: use cache.compressed_storage_bytes(), not nvidia-smi.
4. RoPE precision: all math in float32, cast through model dtype to match rounding.
5. Hybrid models: Qwen 3.5/3.6 DeltaNet layers are auto-skipped — only full-attention layers are tiered.
6. Triton kernels: fused attention via fused_int4_sdpa(), unpack-only via int4_sdpa(force_triton=True). Run check_fused_parity() to validate on your hardware.

License

Apache-2.0. See LICENSE.