BitNet · Issues· 332 open

Latent int16 overflow in ggml_vec_dot_i2_i8_s_1x1: 128 maddubs results per lane against a 32767 ceiling

build/bin/llama-cli throws SIGSEGV, Segmentation fault when using a TL2 GGUF model

Verify evals on Papers with Code

i2_s AVX2 GEMM folds the int16 accumulator 10x more often than its own bound requires (7.5% measured)

[Bug]: BitNet FFN uses SILU instead of ReLU² — perplexity 99.8 vs 17.1 on every CPU backend

BitNet-b1.58-2B-4T produces garbage output on ARM64/NEON (no AVX2) — scalar fallback uses wrong I2_S unpacking scheme

ggml-cpu.c: `src1_cont` undeclared → build fails on ARM CPUs with i8mm (Apple M2/M3/M4, Graviton3, ...)

bitnet-b1.58-2B-4T runs FFN with SiLU instead of relu2 - wrong logits on every backend

Regression: Falcon-E models cannot be loaded — "unknown pre-tokenizer type: 'falcon_e'" (support added in #268 was lost in the llama.cpp submodule update)

convert-hf-to-gguf-bitnet.py --outtype i2_s silently writes F16 (not I2_S) for LlamaForCausalLM BitNet checkpoints (Falcon3 / Falcon-E 1.58bit)

README "Convert from .safetensors" flow is broken: convert-ms-to-gguf-bitnet.py KeyError MODEL_ARCH.BITNET_25, and llama-quantize has no I2_S ftype (setup_env.py i2_s conversion path also affected)

setup_env.py on MacBook M2 failed

[Bug]: I2_S GEMM fast path produces garbage for multi-token prompts on AVX-only CPUs

[Bug]: i2_s quantization produces garbage output on x86 CPUs without AVX2 (Missing fallback)

Warning: `huggingface-cli` is deprecated and no longer works. Use `hf` instead.