Fizzarolli commited on
Commit
26c68b8
·
verified ·
1 Parent(s): f2e26d7

Create README.md

Browse files
Files changed (1) hide show
  1. README.md +55 -0
README.md ADDED
@@ -0,0 +1,55 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ datasets:
3
+ - allenai/tulu-3-sft-personas-instruction-following
4
+ - simplescaling/s1K-1.1
5
+ - simplescaling/s1K-claude-3-7-sonnet
6
+ - FreedomIntelligence/Medical-R1-Distill-Data
7
+ - OpenCoder-LLM/opc-sft-stage1
8
+ - cognitivecomputations/SystemChat-2.0
9
+ - anthracite-org/kalo-opus-instruct-22k-no-refusal
10
+ - allura-org/scienceqa_sharegpt
11
+ - KodCode/KodCode-V1-SFT-R1
12
+ license: apache-2.0
13
+ language:
14
+ - en
15
+ - zh
16
+ base_model:
17
+ - mistralai/Mistral-Small-24B-Base-2501
18
+ library_name: transformers
19
+ tags:
20
+ - instruct
21
+ - conversational
22
+ ---
23
+
24
+ # Sertraline 24b
25
+
26
+ ![sertraline_summary1.jpg](https://cdn-uploads.huggingface.co/production/uploads/634262af8d8089ebaefd410e/RxTTCZtfi2caEy7uIzP5h.jpeg
27
+ )
28
+
29
+ ## About
30
+
31
+ An *actually decent* instruct SFT tune of Mistral Small 3.
32
+
33
+ ## System Prompts
34
+
35
+ I tested with the following Claude-like system prompts, however they were not trained in and any similar prompts can likely be used:
36
+
37
+ ### Non-Reasoning
38
+ ```
39
+ You are Claude, a helpful and harmless AI assistant created by Anthropic.
40
+ ```
41
+
42
+ ### Reasoning
43
+ ```
44
+ You are Claude, a helpful and harmless AI assistant created by Anthropic. Please contain all your thoughts in <think> </think> tags, and your final response right after the closing </think> tag.
45
+ ```
46
+
47
+ For reasoning, it's recommended to force the thinking (by prefilling `<think>\n` on the newest assistant response), as well as not including previous thought blocks in new requests.
48
+
49
+ ## Instruct Template
50
+
51
+ v7-Tekken, same as the original instruct model.
52
+
53
+ ## Dataset
54
+
55
+ This model was trained on [allura-org/inkstructmix-v0.2.1](https://hf.co/datasets/allura-org/inkstructmix-v0.2.1).