Granite-4.2-3B-ROCmFPX-Q8_0-AGENT-GGUF

Granite 4.2-3B — ROCmFPX 8-bit AGENT routing for AMD Strix Halo (gfx1151)

I built this agent-tier 8-bit ROCmFPX quantization of ibm-granite/granite-4.2-3b on my Strix Halo box. Granite 4.2 has native function calling with thinking/no-thinking modes, so the agent tier is the one I would serve for tool work.

The file

Type histogram, read from the finished file:

Q8_0 x185, Q8_0_ROCMFPX x97, F32 x81

What this build type is — and what it protects

Q8_0_ROCMFPX_AGENT (ftype 115) is my agent/tool-call coherent 8-bit routing. It keeps the LM head, token embeddings and the attention-output projections at full plain Q8_0 (185 tensors here) and routes the rest through the ROCmFPX 8-bit UE4M3-scale layout (97 tensors). Tool-call structure is fragile to output-side precision loss, so the routing difference vs my plain tier (which keeps zero tensors at plain Q8_0) is exactly what the extra ~70 MiB buys.

tie_word_embeddings is false on Granite 4.2, so output.weight is a real standalone tensor. I verified the head types by reading the finished file back by exact tensor name (token_embd.weight and output.weight — exact match, not substring).

How I built it

1. Manifest gate: pulled ibm-granite/granite-4.2-3b file list from the HF API with ?blobs=true and recorded the real shard bytes (2 safetensors shards, 7,319,517,120 bytes total — never the index total_size).
2. Downloaded and byte-verified all 15 files against that manifest.
3. Converted with convert_hf_to_gguf.py from my rocmfpx-dspark-halo tree (4eca07e), --outtype bf16 → 363 tensors, 7,323,461,696 bytes.
4. Quantized with the same tree's llama-quantize at 16 threads. Dry-run estimate 3,668.60 MiB; the real file landed within a few MiB of it.

Measured on my box

amd-halo: AMD Ryzen AI Max+ 395 (Strix Halo, gfx1151), ROCm 7.13.0, 125 GiB unified memory. Functional check, not an idle-box benchmark: 8 other llama-server seats were live on this machine while I tested (MemAvailable ~11.6 GiB before load, ~6.6 GiB after). Server flags: -dev ROCm0 -fa on -ngl 999 --no-mmap -fit off -np 1 -b 2048 -c 8192 -t 16 --jinja, port 8497, greedy:

Sample output (greedy, prompt "Explain in one clear sentence what granite rock is primarily made of."):

<think></think>Granite rock is primarily made of quartz and feldspar minerals.

⚠️ Stock llama.cpp will not load this file

Q8_0_ROCMFPX_AGENT is a custom tensor format that exists only in the ROCmFPX fork of llama.cpp.

llama-server -m granite-4.2-3b-Q8_0_ROCMFPX_AGENT.gguf -dev ROCm0 -fa on -ngl 999 -c 8192

Not measured

No benchmark sweeps, no context sweeps, no perplexity — per my build discipline this is the 3-tier publish set and one functional check per tier.

Provenance & license

Converted and quantized from ibm-granite/granite-4.2-3b (Apache 2.0). This quantized build is released under the same Apache 2.0 license. The ROCmFPX runtime is a third-party fork; its own terms apply to the runtime, not to these weights.

<!-- VARIANTS:START -->

All my quants of Granite-4.2-3B

All measured by me on a Ryzen AI MAX+ 395 (Strix Halo, gfx1151, ROCm 7.2.4) with the whole model on GPU (-ngl 999), 128-token greedy generation. A dash means I haven't measured that one yet — I won't put a number in a card I didn't measure.

Base model: ibm-granite/granite-4.2-3b

<!-- VARIANTS:END -->