Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -7,85 +7,79 @@ tags:
|
|
| 7 |
- prism-coder
|
| 8 |
- qwen3.5
|
| 9 |
- synalux
|
|
|
|
| 10 |
- gguf
|
| 11 |
base_model: Qwen/Qwen3.5-4B
|
| 12 |
pipeline_tag: text-generation
|
| 13 |
---
|
| 14 |
|
| 15 |
-
# prism-coder:
|
| 16 |
|
| 17 |
-
Qwen3.5-4B
|
| 18 |
|
| 19 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 20 |
|
| 21 |
## Quick Start
|
| 22 |
|
| 23 |
```bash
|
|
|
|
| 24 |
ollama pull dcostenco/prism-coder:2b
|
| 25 |
-
ollama run dcostenco/prism-coder:2b
|
| 26 |
-
```
|
| 27 |
-
|
| 28 |
-
## Download
|
| 29 |
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
|
| 33 |
|
| 34 |
-
## BFCL Benchmark
|
| 35 |
|
| 36 |
-
|
| 37 |
|
| 38 |
-
|
| 39 |
-
|----------|------:|-------------|:--------:|
|
| 40 |
-
| save | 17 | session_save_ledger logging | 100% |
|
| 41 |
-
| smem | 17 | Session memory search/recall | 100% |
|
| 42 |
-
| aac | 12 | AAC phrase/communication routing | 100% |
|
| 43 |
-
| hand | 12 | Agent handoff (session_save_handoff) | 100% |
|
| 44 |
-
| irrel | 10 | Irrelevant/general questions β plain text | 90% |
|
| 45 |
-
| load | 9 | Context loading (session_load_context) | 100% |
|
| 46 |
-
| pred | 8 | Prediction/task routing | 100% |
|
| 47 |
-
| know | 7 | Knowledge search/recall | 100% |
|
| 48 |
-
| cmpct | 6 | Compact/prune ledger | 100% |
|
| 49 |
-
| edge | 6 | Edge cases (self-description, ambiguous) | 100% |
|
| 50 |
-
| tran | 6 | Translation routing | 100% |
|
| 51 |
-
| info | 5 | Info/capability queries β plain text | 100% |
|
| 52 |
|
| 53 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 54 |
|
| 55 |
-
|
| 56 |
|
| 57 |
-
|
| 58 |
-
|-------|-----------|------|-------------|----------|
|
| 59 |
-
| **prism-coder:2b (Q3_K_M)** | **2.3 GB** | **99.1%** | **1.79s** | **1** |
|
| 60 |
-
| prism-coder:4b (Q4_K_M) | 3.4 GB | 100% | 1.37s | 0 |
|
| 61 |
-
| old prism-coder:1b7 (Qwen3 SFT) | 2.2 GB | 90.4% | 0.30s | 11 |
|
| 62 |
|
| 63 |
-
|
| 64 |
|
| 65 |
## Architecture
|
| 66 |
|
| 67 |
Qwen3.5-4B uses a hybrid attention architecture:
|
| 68 |
-
- **24 linear attention layers** (Gated DeltaNet) β O(n) inference
|
| 69 |
-
- **8 full attention layers** (standard softmax) β precise
|
| 70 |
-
|
| 71 |
-
The `2b` tag reflects the memory footprint (2.3 GB), not the parameter count (4B). No fine-tuning β the system prompt is baked into the Ollama Modelfile.
|
| 72 |
-
|
| 73 |
-
### Why not SFT?
|
| 74 |
|
| 75 |
-
|
| 76 |
|
| 77 |
## Fleet Position
|
| 78 |
|
| 79 |
| Model | Ollama tag | Size | BFCL | Role |
|
| 80 |
|---|---|---|---|---|
|
| 81 |
-
| **Qwen3.5-4B Q3_K_M** | **`prism-coder:2b`** | **2.3 GB** | **99.1%** | **iPhone / mobile** |
|
| 82 |
-
| Qwen3.5-4B Q4_K_M | `prism-coder:4b` | 3.4 GB | 100% | Verifier / 8 GB+ |
|
| 83 |
-
|
|
| 84 |
-
| prism-coder:32b | `prism-coder:32b` |
|
| 85 |
|
| 86 |
## Links
|
| 87 |
|
| 88 |
-
- [Ollama model page](https://ollama.com/dcostenco/prism-coder
|
| 89 |
-
- [
|
| 90 |
-
- [
|
| 91 |
-
- [Qwen3.5-4B base](https://huggingface.co/Qwen/Qwen3.5-4B)
|
|
|
|
| 7 |
- prism-coder
|
| 8 |
- qwen3.5
|
| 9 |
- synalux
|
| 10 |
+
- prompt-engineering
|
| 11 |
- gguf
|
| 12 |
base_model: Qwen/Qwen3.5-4B
|
| 13 |
pipeline_tag: text-generation
|
| 14 |
---
|
| 15 |
|
| 16 |
+
# prism-coder:4b β Prism Memory Tool Router
|
| 17 |
|
| 18 |
+
Prompt-engineered [Qwen3.5-4B](https://huggingface.co/Qwen/Qwen3.5-4B) for MCP tool routing in the [Prism Coder](https://ollama.com/dcostenco/prism-coder) system. No fine-tuning β the system prompt IS the specialization.
|
| 19 |
|
| 20 |
+
## Downloads
|
| 21 |
+
|
| 22 |
+
| File | Quantization | Size | BFCL Accuracy | Use when |
|
| 23 |
+
|------|-------------|------|---------------|----------|
|
| 24 |
+
| `Qwen3.5-4B-Q3_K_M.gguf` | Q3_K_M | **2.3 GB** | **99.1%** Γ 3 seeds | **iPhone / mobile first gate** |
|
| 25 |
+
| *(stock via Ollama)* | Q4_K_M | 3.4 GB | **100%** Γ 3 seeds | Mac / 8 GB+ devices |
|
| 26 |
|
| 27 |
## Quick Start
|
| 28 |
|
| 29 |
```bash
|
| 30 |
+
# iPhone-optimized (2.3 GB, 99.1%)
|
| 31 |
ollama pull dcostenco/prism-coder:2b
|
|
|
|
|
|
|
|
|
|
|
|
|
| 32 |
|
| 33 |
+
# Full quality (3.4 GB, 100%)
|
| 34 |
+
ollama pull dcostenco/prism-coder:4b
|
| 35 |
+
```
|
| 36 |
|
| 37 |
+
## BFCL Benchmark
|
| 38 |
|
| 39 |
+
### Q3_K_M (prism-coder:2b) β 99.1% Γ 3 seeds
|
| 40 |
|
| 41 |
+
**114/115 Γ 3 shuffled runs = 99.1%, 1 flaky case**
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 42 |
|
| 43 |
+
| Category | Count | Accuracy |
|
| 44 |
+
|----------|------:|:--------:|
|
| 45 |
+
| save | 17 | 100% |
|
| 46 |
+
| smem | 17 | 100% |
|
| 47 |
+
| aac | 12 | 100% |
|
| 48 |
+
| hand | 12 | 100% |
|
| 49 |
+
| irrel | 10 | 90% |
|
| 50 |
+
| load | 9 | 100% |
|
| 51 |
+
| pred | 8 | 100% |
|
| 52 |
+
| know | 7 | 100% |
|
| 53 |
+
| cmpct | 6 | 100% |
|
| 54 |
+
| edge | 6 | 100% |
|
| 55 |
+
| tran | 6 | 100% |
|
| 56 |
+
| info | 5 | 100% |
|
| 57 |
|
| 58 |
+
Single failure: "Write a regex to match email addresses" β knowledge_search instead of plain.
|
| 59 |
|
| 60 |
+
### Q4_K_M (prism-coder:4b) β 100% Γ 3 seeds
|
|
|
|
|
|
|
|
|
|
|
|
|
| 61 |
|
| 62 |
+
**115/115 Γ 3 shuffled runs = 100.0%, 0 flaky**
|
| 63 |
|
| 64 |
## Architecture
|
| 65 |
|
| 66 |
Qwen3.5-4B uses a hybrid attention architecture:
|
| 67 |
+
- **24 linear attention layers** (Gated DeltaNet) β O(n) inference
|
| 68 |
+
- **8 full attention layers** (standard softmax) β precise retrieval
|
|
|
|
|
|
|
|
|
|
|
|
|
| 69 |
|
| 70 |
+
This hybrid design is why prompt-only routing works at 4B scale but not smaller. The 8 full-attention layers are sufficient to hold the routing rules when combined with the DeltaNet layers' pattern matching.
|
| 71 |
|
| 72 |
## Fleet Position
|
| 73 |
|
| 74 |
| Model | Ollama tag | Size | BFCL | Role |
|
| 75 |
|---|---|---|---|---|
|
| 76 |
+
| **Qwen3.5-4B Q3_K_M** | **`dcostenco/prism-coder:2b`** | **2.3 GB** | **99.1%** | **iPhone / mobile** |
|
| 77 |
+
| Qwen3.5-4B Q4_K_M | `dcostenco/prism-coder:4b` | 3.4 GB | 100% | Verifier / 8 GB+ |
|
| 78 |
+
| Qwen3.5-9B Q4_K_M | `dcostenco/prism-coder:9b` | 5.8 GB | 100% | Default router |
|
| 79 |
+
| prism-coder:32b | `dcostenco/prism-coder:32b` | 19 GB | 100% | Complex tasks |
|
| 80 |
|
| 81 |
## Links
|
| 82 |
|
| 83 |
+
- [Ollama model page](https://ollama.com/dcostenco/prism-coder) β pull and run
|
| 84 |
+
- [Prism MCP Server](https://github.com/dcostenco/prism-coder) β the MCP server
|
| 85 |
+
- [Qwen3.5-4B base](https://huggingface.co/Qwen/Qwen3.5-4B) β upstream model
|
|
|