Update model card for validated quantizations

README.md

@@ -9,6 +9,9 @@ base_model_relation: quantized
 tags:
 - gguf
 - bf16
+- q8_0
+- q6_k
+- q5_k_m
 - quantized
 - llama.cpp
 - hrm
@@ -21,12 +24,12 @@ tags:
 
 # HRM-Text-1B GGUF
 
-This repository contains a BF16 GGUF conversion of [`sapientinc/HRM-Text-1B`](https://huggingface.co/sapientinc/HRM-Text-1B).
+This repository contains a BF16 GGUF conversion of [`sapientinc/HRM-Text-1B`](https://huggingface.co/sapientinc/HRM-Text-1B) and validated `Q8_0`, `Q6_K`, and `Q5_K_M` quantizations derived from that BF16 GGUF.
 
-The GGUF uses:
+The GGUF files use:
 
 - `general.architecture = hrm_text`
-- BF16 tensor storage
+- BF16 source tensor storage or standard `llama.cpp` quantized tensor storage
 - the original tokenizer from `tokenizer.json`
 - no injected chat template
 
@@ -55,11 +58,19 @@ Only the normal causal generation path is implemented in the patched runtime. Pr
 | File | Description |
 | --- | --- |
 | `HRM-Text-1B-BF16.gguf` | BF16 GGUF conversion of `sapientinc/HRM-Text-1B` |
+| `HRM-Text-1B-Q8_0.gguf` | Validated `Q8_0` quantization from BF16 |
+| `HRM-Text-1B-Q6_K.gguf` | Validated `Q6_K` quantization from BF16 |
+| `HRM-Text-1B-Q5_K_M.gguf` | Validated `Q5_K_M` quantization from BF16 |
 | `runtime/llama.cpp-hrm_text.patch` | Patch adding `hrm_text` conversion and runtime support to the clean `llama.cpp` base commit |
 | `reports/validation/final_report.md` | Human-readable conversion and validation report |
+| `reports/validation/quantization_report.md` | Quantization report, hashes, and pass/fail summary |
 | `reports/validation/baseline_transformers.json` | Transformers baseline prompts, logits, and continuations |
 | `reports/validation/bf16_tensor_validation.json` | Tensor-level GGUF validation |
 | `reports/validation/bf16_vs_hf.json` | Runtime logit and text validation |
+| `reports/validation/q8_0_vs_bf16.json` | `Q8_0` vs BF16 runtime validation |
+| `reports/validation/q6_k_vs_bf16.json` | `Q6_K` vs BF16 runtime validation |
+| `reports/validation/q5_k_m_vs_bf16.json` | `Q5_K_M` vs BF16 runtime validation |
+| `reports/validation/q4_k_m_vs_bf16.json` | Failed `Q4_K_M` validation report; the `Q4_K_M` GGUF is not uploaded |
 
 ## Provenance
 
@@ -68,10 +79,21 @@ Only the normal causal generation path is implemented in the patched runtime. Pr
 | Source model | `sapientinc/HRM-Text-1B` |
 | Source snapshot SHA | `2285b999f6fb8a5b16e0cc313a9e8e4fe447140d` |
 | Source `model.safetensors` SHA256 | `F8FE2B2BF6948414E8E8D6538659198726D98F967C55B533B7AABE8A1FA9A584` |
-| GGUF SHA256 | `2DD5E2EF55E40C46DB0D0CB4CF1427A4E72DA34FEE36F0D2B73D081D0E1C2010` |
-| GGUF size | `2,367,995,648` bytes |
+| BF16 GGUF SHA256 | `2DD5E2EF55E40C46DB0D0CB4CF1427A4E72DA34FEE36F0D2B73D081D0E1C2010` |
+| BF16 GGUF size | `2,367,995,648` bytes |
 | llama.cpp base commit | `6a257d44633d4a752183ed778b88d2924d0a6b9d` |
 
+## Available GGUF Files
+
+| Variant | File | Size (bytes) | SHA256 |
+| --- | --- | ---: | --- |
+| BF16 | `HRM-Text-1B-BF16.gguf` | `2367995648` | `2DD5E2EF55E40C46DB0D0CB4CF1427A4E72DA34FEE36F0D2B73D081D0E1C2010` |
+| Q8_0 | `HRM-Text-1B-Q8_0.gguf` | `1259126560` | `C0729C267C3421E1F6DE0488AC5448E98EA30E56514DAF210596B70AC3F9786D` |
+| Q6_K | `HRM-Text-1B-Q6_K.gguf` | `972668704` | `24D93CA4EF4A02CFE415E3EA56A78AD65198A165A4157B928004B58DBDA2D93C` |
+| Q5_K_M | `HRM-Text-1B-Q5_K_M.gguf` | `851509024` | `F6CE71A076EC897174C555D810ED6E379767D52F9396D485B42E42BF8DB1D0B7` |
+
+`Q4_K_M` was generated and tested locally but is not uploaded. It introduced a new single-token repetition loop for one validation prompt, so it failed the release gate.
+
 ## Validation Summary
 
 Validation was performed from a clean source snapshot and a clean `llama.cpp` base checkout.
@@ -85,6 +107,15 @@ Validation was performed from a clean source snapshot and a clean `llama.cpp` ba
 | Top-10 overlap | `10/10` for all prompts |
 | Text validation | BF16 GGUF continuations are aligned with Transformers baseline |
 
+Quantized variants were validated against the BF16 GGUF:
+
+| Variant | Token IDs | Top-1 matches | Min top-10 overlap | New loop check | Result |
+| --- | --- | ---: | ---: | --- | --- |
+| Q8_0 | Pass | `4/4` | `9/10` | Pass | Pass |
+| Q6_K | Pass | `4/4` | `9/10` | Pass | Pass |
+| Q5_K_M | Pass | `4/4` | `9/10` | Pass | Pass |
+| Q4_K_M | Pass | `3/4` | `8/10` | Fail | Not uploaded |
+
 Full-vocab mean absolute logit error:
 
 | Prompt | MAE |
@@ -129,7 +160,7 @@ Depending on the generator binary and `llama.cpp` build type, the executable may
 - `hrm_text` is a custom GGUF architecture in this conversion.
 - Generic GGUF runners will not work until they implement the HRM runtime graph.
 - Prefix-LM bidirectional attention with `token_type_ids` is not implemented in the patched `llama.cpp` path.
-- Q8_0 and other quantized variants are intentionally not included in this repository.
+- `Q4_K_M` is intentionally not included because strict validation found a new single-token repetition loop.
 
 ## License