Update README.md

README.md

@@ -74,12 +74,12 @@ SPLADE models are a fine balance between retrieval effectiveness (quality) and r
 
 **TL;DR of Our attempt & results**
 1. FLOPS tuning: Seperate **Seq lens and Severely restrictive FLOPs schedule and token budget** doc(128) & query(24) NOT 256 unlike Official SPLADE++. Inspired from **SparseEmbed** 
-3. Init Weights: Vanilla **bert-base-uncased**. No corpus awarness unlike Official splade++ / ColBERT
-4. Yet achieves competitive effectiveness of MRR@10 **37.22** in ID data (& OOD) and a retrieval latency of - **47.27ms**. (multi-threaded) all On **Consumer grade-GPUs** with **only 5 negatives per query**.
+3. Init Weights: **Middle Trained bert-base-uncased with MLM Loss**. Some corpus awarness like Official splade++ / ColBERT
+4. Yet achieves competitive effectiveness of MRR@10 **37.8** in ID data (& OOD) and a retrieval latency of - **48.27ms**. (multi-threaded) all On **Consumer grade-GPUs** with **only 5 negatives per query**.
 4. For Industry setting: Effectiveness on custom domains needs more than just **Trading FLOPS for tiny gains** and The Premise "SPLADE++ are not well suited to mono-cpu retrieval" does not hold.
 5. Owing to query-time inference latency we still need 2 models one for query & doc, This is a Doc model and Query model will be **released soon.**
 
-<img src="./ID.png" width=750 height=650/>
+<img src="./ID.png" width=850 height=750/>
 
 *Note: The paper refers to the best performing models as SPLADE++, hence for consistency we are reusing the same.*
 
@@ -130,7 +130,7 @@ This is why Google's SparseEmbed is interesting as they also achieve SPLADE qual
 document terms with a linear complexity as ColBERT’s late interaction i.e. all query-document term pairs takes a quadratic complexity. The Challenge with SparseEmbed is it uses a hyperparameter called **Top-k to restrict number of tokens used to learn contextual dense representations.**  Say 64 and 256 tokens for query and passage encoding.
 But it is unclear how well these hyperparameters are transferable to other domains or languages (where the notion of tokens changes a lot like our mother tongue Tamil which is Agglutinative in nature). 
 
-<img src="./Metrics.png" width=800/>
+<img src="./Metrics.png" width=1000/>
 
 <details>
 
@@ -251,7 +251,7 @@ print("SPLADE BOW rep:\n", bow_rep)
 ```
 
 ## BEIR Zeroshot OOD performance:
-<img src="./splade_v1.png" width=100% height=850/>
+<img src="./splade_v2.png" width=100%/>
 
 
 ## Training details: