Update chatcode/config.json
Browse files- chatcode/config.json +8 -6
chatcode/config.json
CHANGED
|
@@ -1,18 +1,20 @@
|
|
| 1 |
{
|
| 2 |
-
"name": "PocketVision Chat &
|
| 3 |
"base_model": "Qwen/Qwen3-4B-Instruct-2507",
|
| 4 |
"base_revision": "cdbee75f17c01a7cc42f958dc650907174af0554",
|
| 5 |
"dataset": "HuggingFaceTB/smoltalk",
|
| 6 |
"dataset_revision": "5feaf2fd3ffca7c237fc38d1861bc30365d48ffa",
|
| 7 |
"sources": [
|
| 8 |
-
{"config": "self-oss-instruct", "purpose": "programming", "weight": 0.
|
| 9 |
-
{"config": "smol-magpie-ultra", "purpose": "conversation", "weight": 0.
|
|
|
|
|
|
|
| 10 |
],
|
| 11 |
"seed": 42,
|
| 12 |
"train_examples": 20000,
|
| 13 |
-
"validation_examples":
|
| 14 |
-
"test_examples":
|
| 15 |
"max_sequence_length": 2048,
|
| 16 |
"max_new_tokens": 768,
|
| 17 |
-
"system_prompt": "You are a helpful assistant
|
| 18 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"name": "PocketVision Chat, Code & Writing",
|
| 3 |
"base_model": "Qwen/Qwen3-4B-Instruct-2507",
|
| 4 |
"base_revision": "cdbee75f17c01a7cc42f958dc650907174af0554",
|
| 5 |
"dataset": "HuggingFaceTB/smoltalk",
|
| 6 |
"dataset_revision": "5feaf2fd3ffca7c237fc38d1861bc30365d48ffa",
|
| 7 |
"sources": [
|
| 8 |
+
{"config": "self-oss-instruct", "purpose": "programming", "weight": 0.5, "license": "odc-by", "original": "bigcode/self-oss-instruct-sc2-exec-filter-50k"},
|
| 9 |
+
{"config": "smol-magpie-ultra", "purpose": "conversation", "weight": 0.3, "license": "apache-2.0", "original": "HuggingFaceTB/smoltalk"},
|
| 10 |
+
{"config": "smol-rewrite", "purpose": "writing and editing", "weight": 0.1, "license": "apache-2.0", "original": "HuggingFaceTB/smoltalk"},
|
| 11 |
+
{"config": "smol-summarize", "purpose": "source summarization", "weight": 0.1, "license": "apache-2.0", "original": "HuggingFaceTB/smoltalk"}
|
| 12 |
],
|
| 13 |
"seed": 42,
|
| 14 |
"train_examples": 20000,
|
| 15 |
+
"validation_examples": 512,
|
| 16 |
+
"test_examples": 512,
|
| 17 |
"max_sequence_length": 2048,
|
| 18 |
"max_new_tokens": 768,
|
| 19 |
+
"system_prompt": "You are a helpful assistant for conversation, programming, and source-based writing. Explain ideas clearly. For programming, provide readable code and explain assumptions and edge cases. For reference articles, use a neutral tone, clear lead and headings, distinguish facts from opinions, preserve uncertainty, and use only supplied source excerpts for factual claims. Never invent citations, URLs, quotations, authors, dates, or source verification. Request source material when it is missing. Do not claim a topic qualifies for Wikipedia or that generated prose can be published there. Ask a short clarifying question when essential information is missing. Never claim to have run code or tests unless you actually have."
|
| 20 |
}
|