Falcon-H1-1.5B-Instruct — LiteRT-LM

tiiuae/Falcon-H1-1.5B-Instruct converted to the LiteRT-LM (.litertlm) format for on-device inference with Google's LiteRT-LM runtime. Requires litert-lm ≥ 0.15. Sibling of litert-community/Falcon-H1-0.5B-Instruct — same conversion, same patch.

Falcon-H1 is TII's fully-hybrid design: every one of the 24 layers runs a grouped-query attention branch and a Mamba2 selective-scan branch in parallel on the same input and sums them. Each layer therefore carries both a KV cache and constant-size conv + SSM recurrent state.

File	Recipe	Size
Falcon-H1-1.5B-Instruct_int8.litertlm	int8 dynamic on linears + embedding (convs and the scan stay float); fp32 activations declared for GPU	1.61 GB

2026-09-21: chat template updated to accept the 0.18 content-parts form (string form unchanged); weights, tokenizer and executor metadata byte-identical.

Correctness

• Logits parity vs PyTorch: the float export matches the HF model teacher-forced across 8 decode positions — max|logit diff| 1.6e-04, correlation 1.000000, top-1 and top-5 identical at every position.
• 8-question sanity gate: CPU 7/8, GPU 6/8. The shared miss ("17+25") is answered "32" verbatim by the HF bf16 reference too — the model's own level, not conversion damage. The GPU-only miss (8×7) is a borderline int8 greedy flip; nothing degenerates.
• Prompt-length robustness: hermetic prefill-chunk sweep (fresh engine per length, 12–60 tokens) — all clean.
• iPhone 17 Pro (Metal): the 8-item composite quality probe answers 8/8 on GPU and 8/8 on CPU, identical answers on both backends.

Usage

litert-lm run ./Falcon-H1-1.5B-Instruct_int8.litertlm --prompt "What is the capital of France? Answer in one word."

# GPU
litert-lm run ./Falcon-H1-1.5B-Instruct_int8.litertlm --backend gpu --cache no --prompt "..."

Multi-length prefill signatures (1–1024) are exported so the runtime picks tight chunks. The bundle carries the tokenizer and the stock ChatML-style Falcon-H1 chat template.

Performance

litert-lm benchmark (litert-lm 0.16.0), Apple M4 Max, -p 256 -d 256 --runs 3 --cache no, quiet machine:

Backend	Prefill (256)	Decode	TTFT
GPU	1447 tok/s	102.9 tok/s	0.19 s
CPU	238 tok/s	33.7 tok/s	1.11 s

On device (cold start, single runs, 146-token composite prompt, quality harness):

Device	Backend	Prefill	Decode	TTFT	Peak memory
iPhone 17 Pro	GPU (Metal)	186.8 tok/s	25.8 tok/s	0.88 s	2.54 GB
iPhone 17 Pro	CPU	50.3 tok/s	15.5 tok/s	3.16 s	1.02 GB

Honest notes:

• Where GPU execution is verified. macOS (Metal), iPhone 17 Pro (Metal), Pixel 8a (Arm Mali, OpenCL) and Qualcomm Adreno — the Galaxy S26 section below is that measurement.
• On low-end Android the GPU buys prefill and time-to-first-token, not decode (decode is memory-bandwidth-bound there; the CPU path reads int8 weights while the fp32-activation GPU path reads expanded ones). Pick the backend for your workload: long prompts favour the GPU, long answers favour the CPU. On Apple hardware the GPU wins across the board.
• GPU runs with fp32 activations (declared in the bundle) — expect a corresponding memory multiple over CPU.
• Arithmetic at this scale is fragile in the base model itself (see the 8Q note above); int8 adds borderline greedy flips on exactly those items.

Galaxy S26 — GPU backend

The published bundle runs on the Android GPU backend: LiteRT takes the whole graph and the model generates.

file	GPU backend	delegation	peak
Falcon-H1-1.5B-Instruct_int8.litertlm	runs	41683 / 41683 ops across 12 subgraphs on LiteRT GPU	2398 MB

Measured on a Samsung Galaxy S26 (SM-S942Q / SM8850, Android 16) with litert_lm_advanced_main from litert-lm 0.16.0, --backend=gpu --sampler_backend=cpu, prompt What is the capital of France?. Peak is the process high-water mark (VmHWM) sampled during that same run. Gated 2026-08-24.

No speed rows, on purpose. On this handset the GPU backend wins prefill and does not win decode, so a GPU throughput figure only means something beside a CPU row from the same handset, and no S26 CPU row exists for this model yet.

GPU wiring, including the Gallery import toggle: GPU guide.

Conversion notes

Converted with litert-torch plus a hybrid-cache patch (reproduction script + patch: hf-to-litertlm falcon_h1_work/):

• Composite hybrid cache layer: every layer holds KV + conv + recurrent state at ONE layer index — a cache layer class that is full-attention and Mamba2 at the same time (the runtime binds states by tensor name, so co-residency is just packaging).
• Folded selective scan: the Mamba2 scan is re-expressed as batched matmuls with chunk and head axes folded into the batch axis (all tensors rank ≤ 4, no BROADCAST_TO, no int64 index math) — this is what makes the graph fully delegable on GPU.
• Falcon-specific wiring: the µP multiplier vector (mup_vector, a non-persistent model-level buffer) and ssm_in_multiplier are preserved in the traced scan; the exporter's timestamp-index kwargs are re-injected at the attention layer (FalconH1's layer loop drops kwargs).
• Prefill-pad guard: the runtime runs partially-filled prefill chunks; pad positions are made exact identity steps for the SSM and the stored conv window is gathered at the last valid column.
• Quantization: post-hoc dynamic int8 over linears + embedding only; convs and the scan stay float.

Raspberry Pi 5 (CPU)

Measured on a Raspberry Pi 5 Model B Rev 1.1 (8 GB, Raspberry Pi OS 64-bit) with litert-lm benchmark 0.16.1: CPU backend, 4 threads, 256 prefill + 256 decode tokens, --cache memory (the compile cache lives and dies with the process, so every invocation compiles the model from scratch; nothing is reused between runs), one warm-up plus one timed iteration per invocation, 3 invocations per file with cooldown in between. Values are the median across invocations (min–max in parentheses). No thermal throttling occurred during these runs (vcgencmd get_throttled stayed 0x0). Every file listed produced coherent text in a real generation on this backend before its numbers were recorded.

File	Prefill (tok/s)	Decode (tok/s)	TTFT	Peak RSS
Falcon-H1-1.5B-Instruct_int8.litertlm	39.4 (39.2–39.5)	3.4 (3.3–3.4)	6.8 s	2.5 GB

License and changes

Distributed under the Falcon LLM License (inherited from the base model — see the license link). Changes from the original work: weights converted from safetensors bf16 to LiteRT flatbuffers and quantized as described above; tokenizer and chat template repackaged unmodified. This repository is a community conversion and is not affiliated with TII.